Designing better futures

The book Volatility Trading has lessons about risk for many

Volatility Trading is a good tool for learning about the ins and out of volatility.  Volatility Trading is accessible and informative in the same style as Ernest Chan's book, Quantitative Trading.

Both Quantitative Trading and Volatility Trading offer tutorials from helpful practitioner teachers.  There is a mix of quantified academic rigor in the form of teaching about the tools of the trade and real world humility.  

I Don't trade options or volatility so why should I care?

One of the flaws of the financial system today is practitioner's belief that volatility is risk.  This is a sad case of using the wrong metric to describe what is experienced by the investor or allocator.  This failure occurs at the personal and billion $ pension fund portfolio levels where trustees from august institutions hold sway on allocation committees.

Volatility Trading begins explaining the basics of calculating Volatility and then launches into 6 separate ways to measure volatility.  A risk manager's ears should perk up now.  Portfolio Value at Risk VaR which is driven by volatility and co-variance can vary significantly depending on how things are measured. 

Like the old joke about the accountant, engineer and lawyer goes.  When asked what 2+2 equals, the accountant answers 4, the engineer answers 3.999 and the lawyer looks around and says, "what would you like it to be?".  Most quantitative risk measurements are elastic and prone to subjective biases, ask Moody's or S&P about CDO ratings. 

If you are on the receiving or delivering end of a VaR report or work with any derivative whose value is determined by path dependence you might want to perk up and take a look at this book as it is a good introduction to the field of volatility and indirectly hedging and risk.

If you have a deep background in mathematics etc, the Volatility Trading's math components may feel introductory and light.  For math newbies, things are clearly articulated and laid out like a useful dummies guide to volatility, hedging and money management. 

Living risk is very different from modeling it.

  Sinclair does a good job of trying to explain what it is like to have a position on and go through drawdowns.  Sinclair quotes Mike Tyson in the opening of one chapter, "Everyone has a plan, until they get hit."

Again for risk managers, learning to be more skeptical of volatility by reading someone who trades it for a living is useful.

The book also touches on Kelly money management techniques which can be a useful way to think about leverage, exposure and ultimate trading goals.

Volatility is a limited measure of risk that is detrimentally overused.

Volatility is simple to calculate, easy to apply across asset types and easy to understand. Unfortunately, that doesn't mean it is as worthwhile as it gets held up to be.  Volatility tells you much the car and driver have swerved in the past. Volatility doesn't tell you if the car or driver are going to get you where you want to go.

I am a sucker for cool Math: if you hate math, see if you like this video

Great math is beauty in form the same way architecture or painting can be poetry in motion.  Yes it is a bit of syesthasia for me. 

Great communicators are those who illuminate and share new worlds of understanding.  Here is Marcus du Sautoy talking about the history of Symmetry.  The language of math can be incredible and Marcus seems like a great teacher who is excited to share what he sees and understands.

Thinking in Prime Numbers

Visualizing relationships, be they mathematical, cultural or physical is a powerful skill.  Here is a great site about an interesting quest for visualizing prime numbers.

 I am a big fan of Phi, the Golden Ratio.  One of the best books on this is Mario Livio's The Golden Ratio.  Livio is a serious mathematician, historian and story teller who brings alive the beauty of seeing how things relate.  The golden ratio shows up in many optimal packing and density solutions involving complex systems, it abounds in nature from the circling of an attacking hawk to the positioning of leaves on stems. 

My guess is we haven't scratched the surface of applied uses for constants like Phi or Feigenbaums constants.  For those interested in how growth and form emerge from nature and complex systems I would suggest studying morphogenesis, starting with the classic On Growth and Form.

All of the above end up relating to and explaining a simple elegant shape such as the a nautilus shell.

Flaw of Averages a useful book and tool

Sam Savage's latest offering Flaw of Averages is a useful book and set of thought tools he calls mindles.    If you have a PHd in statistics or mathematics, your job or role rarely begins and ends with you.  Often your models will need to be explained to people.   More importantly the places where the models could or would go wrong need to get explained.

The Flaw of Averages provides a great read and introduction to basic concepts in probability and real life situations where things behave in un-intended ways, such as the hidden stories in the "average". 

The book's core strength is the author's fun and helpful voice .  He is the stats teacher you wish you, your kids or co-workers wish you had.

Sam Savage has spent a long time in industry explaining and teaching how statistics and models work across sectors.  His chief skills is as a creator of simple stories and mental models that let managers and regular people understand the value of statistics applied.  Dr. Savage sells powerful "light bulb goes on in the head moments" for a living and sells the tools that let people have those light bulb moments in their own respective domains.

If you want a way to understand something from a new angle or some interesting new thought tools, take a look at the Flaw of Averages.  Quite a few useful examples and free simulations are available at his website. 

Topics covered include:

• Flaw of averages in finance and real estate
• Flaw of averages in supply chain management
• Examples of more successful product launches
• Petroleum (peak oil is hinted at)
• Many other real world applications and examples that would allow managers to share powerful concepts.  If you can't afford a $2k/day stats manager to assess some of your business issues, try a few hours of reading this book.

Disclosure: 10 years ago, I bought one of Sam's modeling tools for excel and over the course of 4 years proceeded to lose it somewhere between living in Dubai, Denmark and Brussels.  I called Sam up explained the situation and he let me pick up an upgrade for minimal cost.  He is not only a great teacher/practitioner but a nice fellow as well.

Free resource for learning about finance, statistics, physics etc.

Education is one of the most important ways we have of passing "progress" on.  The Khan academy is a great free video resource for learning.

The available subjects are very cool: including the current economic crisis, the Geihtner plan etc.  Also tutorials on great stuff in math.  If you hated math or some subject in college or high school, give it a shot here, you may find a new way to look at and understand the world.

Calculus  |  Precalculus  |  Trigonometry  |  Algebra  |  Finance  |  Pre-algebra  |  Arithmetic  |  Geometry  |  Physics  |  SAT Preparation  |  Probability  |  Linear Algebra  |  Differential Equations  |  Credit Crisis  |  Banking and Money  |  Paulson Bailout  |  California Standards Test: Algebra II  |  California Standards Test: Algebra I  |  California Standards Test: Geometry  |  Venture Capital and Capital Markets  |  Statistics  |  Geithner Plan  |  Current Economics

Here is a video about the housing bubble.  You may not agree with it, but the explanation is clear and lucid.

Deep play: Math, coral & crochet.

For those of you who get it, you will realize this video is a lot deeper in terms of symbolic meaning, linkage, learning and evolutionary creation of symbolic and applied knowledge.  For the rest, enjoy the pretty coral.

Margaret Wertheim leads a project to re-create the creatures of the coral reefs using a crochet technique invented by a mathematician -- celebrating the amazements of the reef, and deep-diving into the hyperbolic geometry underlying coral creation.   

I would be interested in hearing comments about this as most of the really deep math people I know will understand immediately why "play" is important.  pushing symbols about isn't interesting, but synthesizing new relationships or constructs within a bounded environment is very cool. 

For the fairly limited maths used in "advanced" finance and risk, just curious what kind of "wooly" form would pop out of a good CMBS or CDO model.  My guess is a CDO would look something like a teratoma and be immediately recognized as unstable or unfunctional due to flawed form.

Naive trading strategies & young carbon markets

This week I am off to the wall street green trading summit on behalf of the seeking alphasters. Thanks lads.  I thought I would take a look at how the emissions (carbon) market is evolving from a traders and economic efficiency perspective.   Here is a snapshot of carbon prices or the right to emit 1 tonne of carbon about 11 euros:

Truly efficient markets have few opportunities for naive trading strategies to work long term.  A naive trading strategy is something so blindingly obvious only your brother in-law or crazy uncle thinks it will work.  Much to your and those pesky economists frustration it often does work.

I spent +20 years doing quant stuff and think about numbers and culture in my spare time, so cranking out naive spreadsheets is simple. caveat emptor, slippage, liquidity and real concerns haven't been addressed here, so your mileage may vary significantly.  This article is released to provoke thought not trades per se.

Here is the data and the trading system related to this article in Excel 2007.Download MOZ9 naive strategy

The original Naive strategy for equities is buy and hold

Naive trading or allocation strategies are the baseline by which all alpha or performance should be measured.  OK, I know the Beta puritans are screaming EMH, frontiers etc., but bear with me.

A naive strategy is used as a test or null hypothesis of other approaches or processes you are going to employ. If you can't beat the naive approach then just join it. Yes, value investing is a trading process which can be compared to by and hold.   HT: for the fundamentalists out there.

Markets are not stable they typically evolve or "regime shift" over time, just like an ecology.  These regime shifts typically involve participants eliminating or diminishing naive strategies for periods of time as their behavior (rent seeking) leads to shifts.  Most likely some physics student quant guy will eat your naive lunch with a souped up server looking at Lyapunov exponents and slicing VWAP into ever tinier pieces net hops and milliseconds for execution. 

I took a look at an Emission market traded in Europe to see how a Naive strategy would perform.  Here is the market and instrument description.

Below is the outcome, a 1.96 Sharpe ratio* with a strategy using the cunning logic: if today was up, buy tomorrow, else short tomorrow.  The Sharpe or risk reward ratio of index buy and hold equities is usually around 0.50.  So this naive strategy at first glance appears to yield roughly 4X better risk return in what is probably a low-correlated (as traded) asset class.

There are major caveats here, the logic implies, you can tell moments before the close which trade to make and there are no transaction costs or taxes removed.  Also please note this is a futures contract, so the volume is anemic to non existent for 2-3rds of the contract life. you need to research rolled contracts to really see what is under the hood here.  Look at the data yourself, I included the volume in the spreadsheet.
 
The interesting thing is that a naive strategy appears to provide interesting returns.  For those unfamiliar with a sharpe ratio, if one were to de-leverage the position so that the risk profile had a volatility of 15% per annum which is less than the historical returns of the stock market, the returns would be greater than 29% per year.

This is an interesting example of a young market where the naive strategy still appears to offer significant alpha returns. My thesis is, because there isn't liquidity and more sophisticated participants playing yet, the various Naive niches, like and ecology are unexploited.  Of course lucrative naive approaches will eventually attract participants and liquidity, leading to a regime shifts.  Strategies or market process are chaotically unstable based on simple agent behaviours and opportunity seeking.

I am not advocating a trading strategy or process in this article, merely highlighting an interesting young market and the use of studying naive approaches to assess trading allocation processes and their viabilities.  If you run a fund and want to talk quant carve out, shoot me an e-mail, getting portfolio level Sharpes north of 2.0 isn't really that tough.

*m sharpe didn't have an annual risk free rate removed, to simplify leverage assumptions etc.

Grand Unification Theory of equities: a little something to upset everyone.

OK it's not quite a GUT for equities, but a neat trick involving Munger, Buffet, Graham, Sharpe, Claude Shannon, Thorpe, Kelly, Chan, Mr. Market and Douglas Adams.  Thus even if it isn't the GUT for Finance, this blog post is highly buzzword compliant.

Charlie Munger always says, "invert" which means look at the end or out come first.  I think he is right. So lets look at the end goal first.

The goal of any trader, investor or asset allocator should be in the words of Warren Buffet, "to maximize terminal wealth", the long run in Keynesian terms.  Maximizing terminal wealth is done be most efficiently allocating our risk or bets over time.

The brilliant Claude Shannon created information theory which lead to eventually to the Kelly formula used for optimal bet allocation and card counting. Ed Thorpe, card counter, hedge fund manager and mathematical bon vivant among others has made a lot of money using the Kelly formula to allocate risk in the hedge fund world.  You can read about gamblers, mafioso, card counting and information theory in William Poundstone's truly entertaining read Fortune's Formula.

Recently Ernest Chan published Quantitative trading.  For initiates to quant trading this book rules.  The introduction to basic strategies is well done, clear and concise.

One section in the book ties the Sharpe ratio and the Kelly formula together giving the optimal leverage size for maximizing terminal wealth based on Sharpe ratio and thus the expected optimal annualized growth rate for any stable system.   Please note implicit in the relationship is an assumed normal or guassian distribution of returns.  As we all know Guass exits stage left when the real world of finance enters and one of those rascally characters like Levy or Mandelbrot enters.  I met Mandelbrot while acting as Chief Analyst for Starlab, Europe's equivalent to the MIT media lab  a few years ago, nice fellow.  We researched Artificial intelligence, materials science, genetics, nano-tech, even time travel...no kidding.

Mr. Market at the end of the universe.

Mr. Market is Ben Graham's personified embodiment of the stock market, even though he is irrational sometimes, he should be working towards maximum terminal wealth so that he can eat at the Restaurant at the end of the universe with the late great Douglas Adams. 

In order to do this Mr. Market needs to use optimal leverage in the equities market in order to maximize his terminal wealth and pay for dinner.

The thesis is that if Mr. Market is truly an optimal allocator of capital then the relationship for equities between the Kelly Formula and the equity market Sharpe ratio should approximate a leverage ratio of 1.00 for optimal growth.  A 1.00 optimal leverage ratio shows Mr. Market is an efficient gambler and leaves nothing on the table. Given the standard equity assumptions below it looks like the old boy is pretty good at the tables.

 Neat trick huh? Here is the spreadsheet. Download KellySharpeChan

Fortunately, this appears to be the case given the assumptions shown.  Of course it isn't exact as the assumption all shift around, but it is close enough to interesting. 

Thus information theory, Mr. Market and parts of Modern finance all seem to co-exist happily ever after until dinner time.  From time to time the market gets a little ahead or behind itself, but in the long run optimal allocations for maximum terminal wealth are being made.  Now the message is please stay away from leverage and assume that the co-variance matrix isn't stable.  Efficient market theory individual is bogus, but in aggregate may be semi correct. Not Nobel Prize correct but okish.  Scholes & Markowitz should still give back their prizes and Nobel winners like Dr. Amartya Sen should get more daylight as they ultimately help the human condition more.

Please feel free to weigh in with thoughts, critiques and considerations.  The nice thing is that the equity premium puzzle gets resolved by using variance as a risk proxy.

Even if you don't like the thesis above, the tool is helpful in identifying leverage for a "Stable Asset Allocation System" regardless of horizon.

If anyone is looking for a global macro L/S PM with hints of quant talk to this guy, he returned +19.8% last year with +50% cash (no leverage) and a 1.0 Sharpe ratio, has worked in 6 countries and has a sense of humour. CV on request.

Fairly effective predictions tools and equity new highs in 2018

Here is a collection of some of the predictions made in this blog.  The blog missed the full scope of the Chinese bubble burst. I thought -40%, it turned out to be -70% and was 6 weeks early from the top.  Sornette, Gott and yield spreadsheets are helpful tools.  They are available here or for the asking.

China shares bubble bursting (thanks sornette)June 2007

First article on house decline June 2007

Warning of mark to market problems June 2007

A handy free prediction tool made public July 2007

Suggestion CDO would lead to global economic crisis July 2007

Home values lead to $6.6 trillion in losses Aug 2007

Discussion of the panic of 1893 as a pre-cursor to today

Warning on Citibank debt Nov 2007

Anti-VAR and Model Rant, yes I am a recovering quant now value focused Dec 2007

The ratings agencies are SEC abomination Dec 2007

Commercial real estate should be avoided Feb 2008

Fair Value on the Dow 8,800 March 2008 (the simple prediction tool in Excel free download)

Value at Risk is more dangerous than useless, have alway thought so March 2008

Municipal bonds will take a hit March 2008 this is still playing out

Indications that housing could still fall faster.  April 2008

Civilian employment peaked in 1999 May 2008

Recurring indication of overpriced equities dow 8,500 June 2008 

CDS netting risk July 2008

We weren't saving enough(not a prediction just a pointed fact)  Aug 2008

Hints of deflation when inflation was fashionable.Aug 2008

The real scary stuff is in 3-5 years when Basel 2 is in place.

China could experience radical political change in 6-12 months

The sad reality is that equities in the US are -40% from the highs and will need 60-70% in returns to get to break even.  Based on historical average equity returns that equates to 10 years depending on dividends and inflation.  Breakeven is 2018

The point of this post is that active listening to the numbers and ignoring popular opinions or the "wisdom of crowds" could help spot problems. 

There is a fascinating cultural phenomenon I call collective myopia.  It is powerful, in that groups can seemingly perform the heroic, stupid, foolish or sublime all while being oblivious to the obvious.  Understanding collective myopia and being able to listen for it, see it or spot is a useful skill. 

Hopefully our regulatory, legal and political institutions will learn from current mistakes, but there really isn't anything new under the sun here.  Read Kindelberger, we have been going through financial crises for centuries and will invent new debt instruments and buy "things that only go up" well into the future because our neighbors do it.  We are social animals and take our cues from each other, as Kindelberger says, "nothing makes a person lose common sense more than to see a neighbor get rich."

What should you do?

Advice to get into Norwegian T-bills May 2008  this could look good over the next year or two.

 According to Morningstar of 9,900 US mutual fund managers 3 were positive over the last 1 year period, and yes I am up for the last year, but it is no reason to be smug, I am just as scared as the rest of you, for the near future is uncertain.

Tomorrow Oct 10th may be the day without money, but have faith.  Banking, money and credit are cultural phenomonon and will be back.  In 5-10 years we can learn to create new dreams, it is the human drama playing out in all its Shakespearian form...all the world's a stage.

Please note, I do not and am not allowed to discuss positions related to the fund I work with in public forums.  I also believe portfolio theory and VaR type models had a lot more to do with today's crisis than is discussed.

Beat the Market with Ed Thorpe

I have been studying for the CAIA Chartered Alternative Investment Analyst exam coming up.  It is a bit like the CFA but only 2 exams and focuses on alternative investments.  It may be geeky to admit but it is fun stuff.  I used to build trading models as a kid and play with neural networks and t-bond futures in college back in 1990. 

One of the areas that is new for me is Convertible bond arbitrage which is a great way to get back into math a bit.  The strategies and approaches are as I used to be a runner in the 10 year t-note options pit at the CBOT. 

Awhile back I reviewed Fortune's Formula which was all about Gambler's, gangsters and information theory.  It also had great stuff on portfolio allocation and risk which is a subject much like Taxation that few investors think about at their peril.

Successfully growing wealth via compounding returns needs tax deferment and minimal impacts from Siegel's Paradox (the fact that money lost beyond 13% has to work harder to get to break even.)   Risk management and portfolio allocation is sorely un-respected in the field of investing for the long term.  A simple understanding of portfolio auto-correlations and re-balancing approaches can add 100's of basis points to a portfolio at the fund or even asset class level every year.  Of course this mean thinking about 10-20 year horizons.

One of the Gurus of Convertible arbitrage was and is Ed Thorpe who runs a multi-billion dollar fund in New Port Beach, California.  He still publishes and is revered by Nassim Taleb and the Wilmott quantitative finance crowd for his pioneering work.

He published a book in 1965 called Beat the Market that is now out of print.  I found a copy via my grandmother in Iowa of all places and had it scanned.  You won't learn anything new in there, but it is interesting to see a pioneering piece of work and the historical bits such as:

• Spreads of 30-40% on warrants
• A capital gains rate of 25% considered long term after holding a position for 6 months
• Today we call them investment "ideas", back then they were "situations"
• and this view of the future: "We are misguided  when we exalt ourselves by insisting that the psychology of the marketplace and of man are unknowable.  The sciences of man are now emerging from the Dark Ages.  Economics and psychology stand today at Koestler's watershed just as astronomy did in the time of Tycho Brahe.  Our superstition, blind belief, and ignorance are being swept away forever by the scientific accumulation and analysis of data.  There will be predictability in the affairs of men."

Funny how the technology always gets better but the folly and fickleness of men never changes.  Glad a I took anthropology study psychology and sociology in my spare time. That scientific prediction of man is a little over the horizon.

If anybody is interested in a PDF of Thorpe's book Beat the Market for educational purposes drop me a line.

Back to breakeven for equities in 2012-2013

20% bear market declarations are pretty much useless, but they give journo's and TV types something to sell ads with.  Mathematically speaking, the hurt starts at 13% losses.

Losses over 13% require gains of greater than 13% for recovery.  The graph below highlights how this works.  The blue line approximates the intuitive relationship of losses equaling gains.  The red curve is the reality. 

Rule #1 in all investment allocation decisions is don't lose money.  Risk management is the name of the game at the asset and portfolio level.  If you are a European and invested in US shares you have lost 50% of your money after the currency conversion.  You would need a 100% return to break even.  
 
The acceleration point for losses requiring greater gains is around a 13% loss.  Think of it as inverting the power of compounding returns.  Currently the S&P is 26% below it highs.  To get back to the previous high point for domestic investors, the S&P 500 needs to gain a little over 35%.  The long term +80 year historical average return for equities is around 7-8%, so back break even is in around 4 years based on historical average returns would be 2012-2013.

The ex dividend annual return for the S&P for a dollar based investor since 1998 has been roughly .3%.  When one factors in an inflation estimate of 2.5% per year, one ends up with an effective loss of purchasing power of 25% over the last decade.  Welcome to the lost decade.Download returncurve.xls

P.S. I am not such a buzzkill in real life, I am just pointing out some facts.

P.P.S .  Found this at Wolfram Research.  This phenonomon is know as Siegel's Paradox.

CDS chain netting risk

In epidemiology, risk of contagion spread is sometimes modeled using a chain model approach.  The chain approach involves a continuous chain of risk from a single entity spreading through a network of contacts with various rates of success.  Break the chain, break the spread.  CDS exposure to netting failures can break the chain of cascading liquidity crisis with a buffer of excess capital. 

CDS Notional amounts get a lot of headlines, but few really understand the nature of the risk and why it so large.  Netting risk is rare and unusual.  Bankhaus herstatt is the only institution I know of to have failed due to a failed netting where the other side was actually good for the deal.

Settlement is boring in terms of risk, most people like to talk about models, VaR, Swans etc.  But netting is the plumbing that keeps it all going, failed CDS netting could lead to systemic risk on a significant scale.  This risk is heightened due to the following factors:

1. Large Notional amounts outstanding
2. Low Tier 1 capital in place
3. Liquidity challenged environment
4. Crude netting, reporting and tracking in some institutions
5. Sometimes subjective EOD or delayed agreements of an EOD
6. An environment where EOD's are becoming more likely or concentrated
7. An environment where a netting counterparty failure may become more likely
8. Counterparties offsetting risk elongate settlement chains and therefore could amplify netting risk

Here is a simple graphic model that shows nettting exposure growing as OTC (non exchange traded participants each try to minimize risk by offsetting position with other counterparties.  the transactions are identical in terms of reference credit and notional amount.


We can consider this a Chain of settlements that all have to fall in sync assuming "no excess cash buffer".  The Chain runs from A to D
Here is an example of exchange traded settlement as a centralized netting agent with the same behaviours of the institutional participants.

As we can see these two settlement systems have very different netting risk exposures in the event of default events.  The nature of CDS's which act a bit like extreme barrier options is that these settlement chains rarely get tested or stressed.  They haven't evolved robustness from a network perspective.  Most organically evolved networks are formed and behave structurally in something called a scale free network topology.  Basically this means the number of connections or relationships is logarithmically scaled.  Imagine the 80-20% rule in action.
Here are some crude assumptions with absolutely no empirical data behind them.
  My general hope is that the efforts of ISDA to accelerate standardized terms for netting and the efforts of the major banks to establish an automated netting facility happen soon.  I don't know what the capacity of the financial system is for a failed chain event, but it is certainly cause for concern.  We certainly don't want any grad students writing theses with titles such as "CDS netting failure, the day money disappeared."

If anyone has data or a model on the "capacity" for a network to handle these failure it would be interesting to see.  Assuming a 0.05% annual failure and clearing rate of 3 business days. with an 80-20% concentration of failure could lead to the system needing to clear $420billlion in a cycle.  Assuming the same 80-20 rule for institutional concentration means the top 50 institutions would need a liquidity buffer of $336b. 

Again these are made up numbers using the crudest of assumptions, but highlight the risk of failed netting even if the counterparty is good, that or a lot of bank holidays to stretch they cycle thereby increasing the netting capacity of the system. 

I won't even touch on the component of an institutional failure being used as the reference credit.  I call that recursive failure in the system and it is nasty indeed.

Asset allocation (Fortune's Formula) book review & tool

Occasionally one comes across a book that is educational, historical, about economics and that could be made into a movie.  Fortune’s Formula by William Poundstone is just such a book.

Fortune’s Formula by William Poundstone combines Claude Shannon's information theory with gamblers, hedge funds, card counters, crooks and history. 

What a lot of investors and fund managers don’t realize is that they are gambling blindly.  I won’t go into my issues with mean-variance asset allocation, but suffice it to say, I consider mean variance approaches and VaR as belief systems that are far too popular for our own systemic good.

Fortune's Formula includes great bios of Claude Shannon and Ed Thorpe and an introduction to something known as the Kelly Formula.  Ed Thorpe is a mathematician, hedge fund practitioner and in the eyes of some, including Nasim Taleb the rightful co-creater of what is now known as the Black Scholes formula or the (Thorpe-Bachelier) theorem. See Wilmott

Those of you familiar with LTCM are now thinking, “…an academic mathematician hedge fund manager, this is going to have a nasty punch line at the end of it.”  The punch line in this case is thankfully flat and deeply impressive, Billions of dollars managed in one of the lowest variance profiles for decades.  Thorpe has delivered the best financials met by a very small handful of managers.

Kelly’s Formula

Kelly is described as:

“…working at the esteemed Bell Labs, a chain smoker, liberal drinker, a lot of fun, the life of the party. He was gregarious loud, and funny, quick to loosen his tie and kick off his shoes.”

He was also brilliant.  He came up with the optimal “bet” asset allocation strategy assuming one knows the odds of playing the game and the payoff.  His approach to asset allocation provides the maximum return over a given set of time.

Please re-read the last sentence if it didn’t impact you enough.

The basic formula is:  Edge/odds=portfolio allocation (bet size) …what the heck does that mean and how does one use it?  If you are a gambler it clicks immediately, if you are trader or mathematician, have patience.  For those familiar with Markowitz asset allocation there is a nice section on Kelly vs. Markowitz.  I personally consider Markowitz a version of trend following and dangerously suspect in regards to the instability of correlation drift.

The basic premise of Kelly allocation is that in order to maximize the geometric return of a series of bets over time, one should go about allocating capital based on the average edge one has in “winning” versus the odds of winning that are paid out.

The greater your perceived information edge, the more you should concentrate your bet.  There are many Kelly sites and tools, some of dubious merit and intent. For those of you in the money management business, you may want to “tone down” the geometric return and perform a variance analysis relative to the risk profile of investors you have holding your fund.

Simple rules of thumb may be good sanity checks. Having grown up in Iowa and “played” futures as a teen, I can say don’t ever bet the farm on a sure thing, don't worry we still have the farm back there.  As a young runner, I used to watch smart traders managing there bets like this in the 10 year bond options pit at the Chicago Board of Trade.

Investing is a marathon event. Long tortoise, short hare. Investment clients may say they want the highest return possible, most wouldn’t stick around after the drawdown associated with a higher volatility approach even if they wanted to be “part of the plan” in the early days.  The pros, Fund of Funds etc. usually say they are comfortable with a certain volatility profile, but will bolt at certain drawdown levels.

True fact: most investors under perform their investment vehicles.  I have studied this using both hedge and mutual fund flow data. Investors, invest high and exit low.  In gamblers parlance, they make horrible jockeys and can be their own worst enemies.

A simple tool is provided for playing with the concept at the bottom of the post, I just lifted it off the web so make no endorsements for it or anything related to it.  This approach might make sense for a portfolio of investment, drug candidates, product launches etc., just beware of your own biases inherent in the estimates.  Like all models it recycles all too efficiently: put garbage in and you will get garbage out.

The nice thing about investing buy and hold in equities, is that unlike a casino, the odds are actually in your favor over time, assuming your horizon is great enough to overcome transaction costs. 

As a money manager I find this ties in nicely with Buffet’s approach of a few eggs in a basket watched closely.  The greater your information edge (remember to be humble) the greater your holding concentration should be.

To learn more I would highly recommend buying the book, the story is a heck of a read.  If there is a movie it will probably be horrible.

Even Warren Buffet makes an entrance into the book with his passing interest in nontransitive dice.    Warren is all Midwestern style, he speaks plainly, seems to listen a lot and gets results without a lot of fuss, still waters running deeply.

Central bankers on vacation, inflation gone mild

Government reports of inflation are derision magnets.  The freshly released inflation figure is easily the least popular kid in the economic class of statistics and can usually be found seeking solace next to his old punching bag buddy the quarterly GDP estimate.

Why are inflation numbers so unpopular?  My own opinion is that the anger and debate this statistic engenders is due to the fact that it can’t possibly handle the work load assigned to it.  Everyone wants inflation to tell a story about their current view on the economy and the future, yet it can’t do that. 

The whole problem with inflation is that it is a point metric describing a complex system. What?  Yes you heard me, it is a point metric, a single point trying to describe the state of a complex system. 

One Average I feel fine.

Inflation is limited as an economic barometer in that it is an average.  There is old introduction to statistics about a man with his feet in a block of ice and his head an oven, the old fellow replies that, “Oh, on average I feel fine.”  Inflation currently reports the US is feeling relatively fine.  Inflation is a point metric describing a complex system.

 

How it really works and why we should be nervous.

The system illustrated isn’t fine, eventually the output is going to be in for a wild ride.   One could suggest that variance in the system components may say more about future GDP change than inflation alone. 

 Tools & entrails

Most people assume the inflation number represents a flow metric driving another flow metric, namely future GDP growth (economic output).  One could argue that inflation metrics are better used when combined with component variance as an indicator of future likely GDP variance.  Most likely variance is tied to inefficiency and thus temporary GDP contraction.

Central bankers, economists and other dangerous folk try stabilizing inflation using basic monetary tools.  The assumption being that stable inflation is correlated with higher stable GDP growth.  The reason I call these people dangerous is that they are attempting to control something mostly outside of their control with blunt tools.  Most central bankers are oracles with only half the required amounts of entrails required for predicting the future ;).  What should one advocate as proper policy for the system illustrated above?

I believe central bankers to be well meaning people, but my hunch is that outside of regulatory oversight roles, the least active central bankers are probably the most effective ones.  Many interventions meant to stabilize the system may actually increase uncertainty and variance in the system, thus lowering long term GDP growth. 

The safest thing a central banker can say is, “I don’t know what is really going on right now, but I am going on holiday until it sorts itself out.”  The economic system eventually stabilizes itself.  This is the laissez faire economics schtick after all.  Unfortunately oracles who claim not to possess the gift of fortune telling typically get replaced, so we are all a bit to blame as well.  Please put down those monetary control tools, you may hurt someone.

Pattern recognition is psychologically wired into all of us.  Biology makes us prone to many errors of judgemen and beliefs in contol that flawed.

The working thesis is that low 1-2% stable inflation is correlated with growing GDP.  The things that create stable GDP and stable inflation are probably regulatory certainty and monetary stability.  Perhaps the politicians should go on holiday with the central bankers and let the rest of us get on with it. 

No more stimulus checks please, its beginning to economically look like Zimbabwe around here.  What next will they get up to next, perhaps adding a zero or two the currency to create a perceived “wealth effect”.  Bread and circuses folks, that’s what we are asking for, and what we are getting delivered.

The typical assumptions of dampening the economic cycle via active monetary policy is nice, but may be flawed due to the multiple feedback loops in the economy.  See or better yet play the beer game to understand how this works.

So the next time you see a central banker or politician, encourage them to take it easy and perhaps find a bit of sun somewhere far away.

If anyone has some pointers or research on economic component variance as a future GDP indicator or monetary policy inactivity as it relates to GDP growth please post a link in the comments.  I enjoy listening and learning.

Flaming VaR (value at risk)

One of my interests is financial risk.  I am fascinated by complex problems involving numbers and behaviour.  One of the most popular approaches to financial risk management is called Value at Risk or VaR.

VaR and the assumptions underlying it are some of the key elements of todays financial system.  VaR is most likely to be the key risk methodologies going forward.   VaR has been created in lots of flavors.

• Parametric
• Historical
• Extreme
• Vanilla
• Flaming

Ok, I am making the last one up, the good people in academia and Risk metrics haven't invented flaming VaR on purpose.  Flaming VaR has been created by market reality hitting bad models.  There is a tiny secret that people in the multi-billion dollar financial risk business don't want you to know...  VaR blows goats. The nice thing about VaR as a system of risk is it only fails 1-5% of the time, of course this is when you need it the most.  It is a bit like buying airbags for you car and being told they work for every accident situation except head on collisions.
Why does VaR fail and why should we be scared.
VaR fails because it assumes prices are log normally distributed. Common sense and a brief reading of financial history will tell you this isn't so. 
VaR scenarios are usually built using price data sets that are limited. Example: today people are pointing to previous US recessions looking for clues about the future.  Most of these commentators are looking back to the 70's or 50's and only the US.  Small hint: a statistical sample of 3-7 is pretty pointless for analysis.  An anecdote is not analytic.

VaR expresses risk as a dollar amount at risk over a given period of time with a confidence interval.  Right now some bright risk manager at some bank is dutifully reporting to someone, "sir, we have $5m at risk daily within the 99% confidence interval."  His counterpart is then replying "good Humphrey, because we have $10m of risk padding". 

Confidence intervals and statistics make most people roll over and play dead.  The certainty with which someone quotes VaR makes the other person think, "hmm ok, we are safe, this fellow speaks math and must be bright."  But lets look at that earlier VaR statement.  It basically says for that for a given 1 day period there is 1% chance that we are going to lose more than $5m.  Hey there are most likely going to be 3 of those days next years.  Now the belt and suspenders genius in the room multiplies the $5m by 2 and says we are OK.  You may have noticed some problems already, most financial losses take place over many days. 

VaR approaches are such that the longer the length of time the less accurate the sample.  huh???  Think about it, If I have a pool of 30 year data and there are roughly 7,800 days to look at.  If I want to sample a one month period I have 360 samples, for the smart kids in the class a rolling approach can be used.  Even the rolling period has the flaws that I am trying to understand maybe an extreme externally driven price driver (macro-economic shock etc) with a 30 year window.  Those shocks are rare and far between, and this is the problem with historical and even parametric approaches to risk modeling, they treat markets like the live in a bubble.  It is a classic system failure.  Hint Warren Buffet thinks VaR and Options theory is junk.

Some of the bright lads over Goldman got this wrong, but they are in good company with the Nobel Prize winner of LTCM among others.  They calculated that the chance their portfolio would blow up was a 25th standard deviation event.  To normal people 25th Standard deviation is statistically equivelant to 1 in over a trillion billion chances give or take, which equates to the sun blowing up  in 7.5 billion years before the portfolio would be stressed.  Now if this market worked in a bubble that might be right, but disasters happen, govt.s fall, things change, stock markets disappear every 150 years or so.  I am sure there is some poor quant looking at his numbers wondering how this could happen.  It happened at LTCM and it will happen wherever VaR approaches are used.  The problem is the model, read the Black Swan for a good insight into the problem.  Historical price data are useless for external event based shocks.  I bet there isn't a financial model out there that has the impacts for a second (post 1907) San Francisco Earthquake modeled into it, but that is a 25% probability by 2025 for such an event.  Historical data approaches that use market price data are flawed and doomed. 

It is a bit like driving a ship and predicting the landscape for the next few months by looking out the side port hole.  For those of you who like Monte Carlo forecasting, feel free to look out the port hole a few million times, the information won't be any more effective, but you will have greater statistical confidence to order the ship to go faster.

Imagine if back in 1905, you and I are planning to build the Titanic.  You tell me that after looking over all of the history books you discover that a ship that size has never sank.  You then proceed to tell me that given this fact, statistically speaking, my ship will never sink.  I then proceed to build a bigger and therefore safer ship.

Once you give a mathematician a framework, good or bad they will iterate (improve it) forever.  Nothing is more fun then honing a fine model to a pointless point.  Rarely do people challenge the framework, I mean look at all the data and the dancing around they can make it do.  I would advise using common sense or Gott's theorom to pull them back to reality. Expand the referent frame. Hint: there isbig money to be made betting against the VaR heads.  Go long gamma in the tails and spread the bets.

Moody's and S&P for example have securities called CDO squareds, that are rated super AAA.  These things were and for some fools still are considered safer than the US govt.  The flaw in this of course is that things that are themselves components of other larger systems at risk can't overcome the failure of the larger system.  Logically if the US govt were to have an issue, most likely US based securities may have even greater issues.  Common sense isn't a requirement with a math or economics PHD.  I think there should be a variable CS applied to each equation in risk a CS would be positive or negative indicating whether someone with common sense had agreed that the results, not the approach made sense.

The reason to be nervous about this  whole thing is that having a bad risk system is worse than no risk system.  Financial risk is a mix of human behaviour in the pursuit of return.  If we believe we are risk secure vs. exposed we exhibit different behaviour.  The current horrible risk systems; VaR, Basel II and others are far more dangerous than earlier naive risk systems usually based on rules of thumb and common sense.  It is safer to have everyone a little bit worried and driving more cautiously than have everyone told they are driving hummers and have Michael Shumacher skills when they don't.  Basel II is going to cause a financial disaster.

Prepare for more fall out in the coming 6-9 months as one or two more international banks fail and a few hundred mid sized US banks suffer the same fate.  The ratings system which used a mean variance approach to risk and past prices delivered this one. 

It turns out we are all collectively safer when we all feel a little more at risk.  Quite a few hedge funds in the statistical arbitrage, debt arb and mortgage back space will also be closing in the coming weeks and months as margin calls and collatoral requirements get upped.  The global macro guys and 130/30 equity guys should be ok as they were seeking real alpha with less leverage.

The sad thing is that the global fianancial system is implementing Basel II.  This means that after we start making it out of this issue most likely in late 08 to early 09.  We can get ready for the big one, which will probably be a 30th standard deviation event starting in 2011, just 1,000 days away.

Warning past correlations may be larger than they appear

The most dangerous tool in the financial modelers tool kit is correlation only bested by its kissing cousin co-variance.

In order to keep this blog from being drier than a popcorn fart in July, I will forgo detailed statistical treatises.

Most modern historical accidents relied on a blind belief in correlation statistics.  The belief that the non occurrence of co-incident events in the past surely indicates the un-likely event these events could bump into each other in the future.

Many financial modelers have had the good fortune to replace common sense with a PhD.  This has lead them to great salaries and recognition at least temporarily on Wall Street.  The recognition will continue for some time in the future but will alas not be conjoined with good cheer and great  salaries.

Most of the accidents in finance, the current CDO and CDO squared bubble are due to the dubious belief that past relationships are indicative of future relationships.  In a closed physical system this is true.  For the rest of the real world it is not.  As Warren Buffet has stated if the past was a way to predict the future, then the wealthiest people on the planet would be librarians. 

Mathematicians who aren't known for being famous with relationships in the real world are taken far too seriously when it comes to the financial world. 

For the mathematicians in the audience please wake up and realize the financial world is not a loosely coupled Hilbert space.

Being a recovering quantitative analyst I can assure you that the faith placed in financial math models is scary.  I have worked with 2 of the world's largest global banks albeit in jr. roles and wrote my masters thesis at a top engineering school on quantitative approaches for hedge funds. 

The greatest danger in finance is the ever shrinking *.  The * (asterisk) starts life in the world as a caveat or warning that a model rests on certain assumptions.  As money is made and people want to believe, the asterisk seems to shrink and disappear in importance.  After the inevitable accident the stunned financial modeler will bring back up the * in order to save his own asterisk.  The current asterisk that should be quoted in regards to CDO's and CDO's Squared is "Assuming the correlations in mortgages debt holder payback etc. holds steady."

Many a flawed Markovitz asset allocation has fallen foul of the correlation error.  As products are created that are ever more "Markovitz efficient", such as structured products we will all be at risk of correlations that are larger than they appear in the rear view mirror.

 

The next big Asterisk to read about will be RMBS (retail mortgage backed securities) and the inevetible failure of a few CDS (Credit default Swap) counter-parties.   This will lead to yet more banking problems.

Unfortunately the problems of the credit crisis are still unfolding.  Things will most likely be improving by 2009, but the nadir has not been reached.  A few banks are going to be caught out when "winning" CDS trades are activated by default events and the counter-parties are found to be troubled banks, iffy hedge funds or suspect trading vehicles.  Full disclosure I make a living in the financial markets and trade based on some of the information in this blog.

SEC creates the Ministry of Truth

Warning gentle reader, this is a personal rant.  The Securities Exechange Commission has a small power that makes it indirectly manipulate over $3 trillion in debt Securities.  This power is vested in them by their declaration to determine who is and who is not an NSRO (National Statistical Ratings Organization). 

The NSRO's are the organizations who gave AAA ratings to the instruments known as CDO Collatoralized Debt Obligations.  These NSRO's have mis rated businesses and municipal govt's for decades.

The NSRO scam is tantamount to creating a legal monopoly on truth, something still in vogue only in the bureaucracies of North Korea, Cuba and a few other less cheery holiday destinations.  The irony of these actions happening via the SEC is strong.

The SEC's act of anointing NSRO's is tantamount to the creation of a ministry of Truth.  Imagine if you will that a small group of companies anointed by the government told you which investments you could and could not make. You dear gentle reader might like to seek a second opinion.  Unfortunately that second opinion while readily available would not carry the weight and benefits granted by the NSRO's.

Anyone in a democracy wouldn't stand for a government monopoly on truth setting the price on investments and yet that is exactly what the SEC does via its creation of NSRO's.  This monopoly on truth is one of the systemic flaws that has lead to the current credit crisis. 

The credit crunch of 07-08 is a logical follow on to the credit bubble.  The nature of the failure is what is known as a normal accident.  Normal accidents, sometimes called systemic accidents occur when the very structure of a complex system leads to its failure.  If you are interested in these types of accidents read the book Normal accidents by Charles Perrow.

The global credit market is undergoing a "normal accident".  The very structure of limiting debt ratings to a few agencies helped facilitate this.  It is a classic case of unintended consequences, whereby a safety mechanism induces a failure.  Usually in banking risk arrives in the form of moral hazard, here it showed up in limited opinions and dubious math via a few limited groups.

It seems a bit flawed to limit legally recognized opinions of creditworthiness to an anointed eight firms.  One should have multiple opinions and none should be more equal than others.  I look forward to industry and academic papers which call for  the doing away of the whole NSRO system.  My greatest fear is that the SEC anoints a few more firms as NSRO's such as the recent granting given to Egan Jones.  Now we have 8 geniuses establishing truth for the world's debt markets.

In my opinion the whole business is rotten and risks greater future problems.  Every ministry of truth system has been found to ultimately rest on falsehoods which lead to their collapse.  As Karl Popper has indicated correctly, every system ultimately bares the seeds of its own destruction and this holds true for the NSRO's. 

There is a need for dissenting voices and the discrediting of the entire NSRO system for the shambles it is.  For the sake of the debt markets, please contact your local congressman and ask them to tell the SEC to give it a rest on anointing those debt market geniuses known collectively as our NSRO's.  The NSRO system needs to be eliminated for it is fostering the very risk it seeks to mitigate.

Calling all Quants.....

Warning this is for quants only.  If you understand the problems of leptokurtosis in options pricing or the nature of gamma risk read on.....

I have recently been reading the Black Swan, which any serious quant will have finished a long time ago.  Thinking about black swans etc. the inherent failures of Black Scholes variants to work as a good pricing mechanism got me to thinking.....

If one converts a vol point on a curve into a probability or likely hood event, then one can also construct a time likelihood etc. easily, ie. this is a in 10 trillion year event (See Goldman Sachs alpha fund et al for details.)   The post fact analyst and pundit always can say that common sense made it obvious such and such was a falsehood.

I think a more interesting way to step out of the fallacy of pricing extreme events or price variants is to step outside of the price box altogether and to posit a regime type question.  ie. how long has this phenomenon existed.  Then using the work of Gott, one could arrive at the probability of it not existing.  Here is an example instead of pricing AAA CDO's based on a finite time series of a few years and then assessing the shift outside of say 2 stdev's.  My suggestion would be to use Gott's theorem pose the question how long has this instrument been priced below/above X.  Then convert the time back into a probability to use in pricing the instrument.

Now this approach has flaws in that, the answer will vary on the way the question is framed.  I would posit however that the naive approach of only using previous price is also only a singular frame of reference to assign value. 

The major thesis is to not diminish VaR, and BS theory but rather to add an infinite number of ways to assess the framework of price and value and to free valuation from the context only of price and potentially put valuations into a wider framework of analysis.

Any comments...?

Scary financial stuff

demon of our own design is an interesting look at the world of financial risk.

The Author is a practitioner and academic who provides a welcome perspective on the area of contemporary finance and the environment of financial risk. 
The book is a bit heavy going for the lay person who may not be comfortable with options and and basis points, but for the market practitioner it is a must read.

The author put forth a very reasonable account of the foibles and mistakes that have been made in recent areas of finance.  He articulates the theory of Normal Accidents and how they relate to finance in a great way.  Normal accidents, are a phenomenon of design whereby a complex system, be it a nuclear reactors, airplane safety procedures or any complex system is designed to fail.  This design for failure of course isn't intentional but unfortunately due to complexity of interaction the system fails in an un-intended but wholly preventable way, 3 mile island, Chernobyl and about any financial accident you can think of falls within this group.

The author does a good job of explaining why there really is no such thing as a hedge fund and also posits a few interesting questions about the half life of a trading strategy and the usefulness of ever more esoteric instruments.  As he states, just because you can swap out a cash flow based on a state variable doesn't mean you should.

The clarity, candor and perspective in this book are refreshing.  My own personal concerns in terms of financial risk were only re-enforced.  The current credit crunch articulated in this blog will grow in size significantly but most likely not pose any systemic threat on a global basis. 

The current credit issue will be acute not chronic a $2-4 trillion mess at most.  Looming in 3-5 years will be a crisis which will ultimately be attributed to the Basel II capital accords.  This will be chronic and most likely caused by the safety mechanism of the Capital accords which are being put into place to mitigate risk. 
My sincere hope is that more people read this book and think about the dangers inherent in layering on ever more esoteric risk management systems with finer models thus allowing ever greater financial "innovation".  Gamma risk will look like nothing compared to Basel II model risk when it is fully realized.  Innovation is rarely a good thing in finance and banking especially when it involves ever more esoteric ways of "deriving" value by seperating an asset from the fundamental information or sources of it value creation.

a little thougt project

Happy Thanksgiving.  Here is a great video for a little food for thought for the holidays, if you really pay attention your head may hurt a little after watching this and you will forget about that belly full of Turkey you are carrying around.

Predict anything...kind of

 

Now here is a neat trick, using Gott's interpretation of a Copernican observation.  You can predict the likelihood of almost any event terminating based on how long it has been happening so far.  Gott made a famous observation about the Berlin wall falling and got it right :) 

I took Gott's math and converted into a handy predictor excel sheet for you.  This is sure to wow your friends.  To read about the estimates behind the end of mankind (no kidding) check out this new york times article. 

Gott is a pretty interesting astro physicist at Princeton university and deals with the anthropic nature of cosmology and universe. 

Me, I am just a sucker for good probability and Bayesian type party tricks. Feel free to download the Nick Gogerty predictor Download observation.xls to wow your pals with.  Please don't pull this out on a first date unless she is a math major.

Warning finance geek speak: If you are a quantitative hedge fund manager or event driven macro trader it may spark some ideas involving long dated Volatility and drift ;) as this allows one to tweak certain Volatility assumptions with event probabilities, such as extreme exogenous shocks or cessation of existing regimes.

Volatility Caveo : volatility beware

A lot of people don't realize how bad losing can be.  Behavorial economists, show that we psychologically feel loss about 2.5 times greater than gain.  Thus finding $10 and losing $10 isn't the same thing.  If you lose $10 it feels like a $25 loss. This is psychological.
In investments and fund management a mathematical truism is in place.  If you buy a share in a company or fund and it loses 50% of its value, that same share has to gain 100% from that point onward for you  to be at break even.  This relationship of the  "extra" effort required by an investment is shown  in the graph.   Basically after 12% the extra return required is climbs fast.  The first objective for an investor is to not lose.  Seek safety and think long term.  Another interesting fact is that most fund losses and points of maximum loss (peak to trough extremes) sometimes referred to as drawdowns are highly correlated with the volatility of a funds monthly returns.  Years ago I did work in studying hedge funds and found this correlation between volatility and maximum drawdowns held true statistically.  This makes sense as a hedge fund manager can't predict their own monthly or yearly returns.  If they could they would control for it.  Thus if you are selecting a fund, a recommendation is to look for a fund with a  monthly volatility of less than 25%.

Meeting Dr. Fractal and good dinners

Way back in the past, a long, long time ago when dinosaurs roamed the earth and I still had a bit of hair, I was involved in a financial modeling workshop on behalf of Starlab.

Here is a picture of myself with Dr. Keith Still (far left), Benoit Mandelbrot (middle) and me.

We were working together at conference in Texas in 2001 over a weekend.

One of my favorite things to do is have a good dinner conversation over a bottle of wine.  When I worked at Starlab, I lived on site and used to do just that.  I considered myself fortunate if I could rope  people together for a meal.  I found myself surrounded by PHd's in Math, Physics, Biology, Robotics, Artificial Intelligence etc.  This is a fantastic way to spend an evening, just listening to learn how each of them viewed how their world works or could in the future.  Some of my other friends were sociologists, architects, musicians, designers etc.  Most were from all over the globe, usually I was the token American :)  Great meals and great conversation. 

Today, I am lucky and so are my dinner guest in that my wife likes great conversation and is an amazing cook, this means better food for the guests. I do the dishes :)

Back to Mandelbrot, he is one of the most famous mathematicians alive today and I have a great deal of respect for him, he never sold out or allowed his name to be used improperly.  He discovered many interesting aspects of financial markets among his work in mathematics. To my knowledge he never allowed his "brand" to be used in a way that might mis-represent the truths he found in finance.  He was a nice and  bright man with a gracious wife. 

I had the pleasure of working with Dr. Keith Still in Brussels, again a bright man gifted in many areas of science and possessing an interesting ability to analyze and put things together.  He is one of the world's experts on movements in crowds and egress.  He has done work in modeling irregular heart beats, foreign exchange and a few other fields, truly a polymath, no pun intended.  He possesses a a great knowledge of modeling complex systems with multiple interacting agents.

One afternoon Keith introduced my to his mentor  Dr. Ian Stewart (Mathematician) and Dr. Jack Cohen  (biologist) I conjunction with Keith, Ian and Jack was a Stem Cell biologist.  We had a 3 hour lunch in an English pub in Warwick.  Great stuff, funny conversation and incredible opportunity to try and keep up with everyone.  We spent the whole time discussing morphogenesis and other fascinating topics.  A great afternoon. If you get the chance check out some of Stewart or Cohen's books they are great reads for the layman or sci-fi fan.

China shares will bust in 8 weeks before Sept 1, 2007

ADDENDUM: since posting this originally, I have gotten some feedback from the Economist indicating Shanghai a-shares are at a 45 PE.  I thus recant my prediction and state it thus.  A 40% decline is possible if PE's get to 70.

The economist made a surprising mistake this week.  The showed this graphic in an article about the potential for a bubble in Shanghai A shares.

The mistake is showing a "bubble" on a straight graph and not a semi-log chart.

The assertion by the Economist is that Alan Greenspan's assertion that china is at risk of a bubble is wrong.

The graph to the left is the wrong "bubble-o-meter"  and here is why.  Bubble are psychological and not mathematical phenomenae.   The ups and downs of a bubble are felt in percentage terms  by the participants. 

The graph to the left tries to articulate with a simple normalized data set that bubbles take  5 years and that the Chinese share bubble is below previous bubbles. 
This is a bogus presentation of the facts.  If we were to pick the lowest trough of the bubble roughly 4 years into the current curve "roughly equivelant to 75 in this chart and then calculate the rise , we would see a  renormalized peak  climbing faster than the nasdaq Bubble to roughly over  500 in  a much shorter time.   Again this would be obvious with a semilog paper and normalized series.  

As they saying goes, prediction is hazardous, especially about the future.  My personal prediction is for a decline of 70% in Shanghai A-shares from sept. 1,2007.  The political fall out will be huge due to the popular participation in the markets. Many "Average" people are now playing the markets and will lose out.  The political fallout may require a bail-out by the Chinese government to quell dissent.  If that occurs, it could require a major sell off of US treasuries held by the Central Bank.  This would cause a slight dollar decline, a potential excuse to let the Yuan rise and potential Rise in US interest rates.

The graph below is not semi-logarithmic, but shows how "normal people" see shares in terms of participation (turnover).  A blog about this is here.

For comparison sake, please have a look at the famous South Sea Bubble as well.

If anyone has recently been to China and has the "cab driver/doorman" with a winning stock tip anecdote as contrarian indicator.  Please post a comment.

something very interesting about time travel

Ok,  when I worked at Starlab on the Sr. Management team in Brussels we had a physicist on staff working on time travel.  I had to explain this to cynical VC's.  For any of you who have pitched an idea to a sceptical VC you can imagine the sell was tough.  Here is a video which highlights some work being done on building a time machine.  If the device works, it may be able to retrieve information from the future.  Have a cup of coffee then watch this, think about it.  Go for a walk, then watch it again.  To paraphrase the great Danish physicist Niels Bohr, "If you think you understand the implications of this, you haven't probably understood it at all."

new algorithm for a new day!

First off lets ignore the lame image here.    One of the more interesting companies playing around in the field of Machine learning (AI) whatever you wish to call it has opened up their development algorithms.  Numenta, the baby of Jeff Hawkins has opened their doors for developers to drive there development environment.  This is fantastic stuff.

Hawkins book "on intelligence" which I reviewed earlier is a fascinating read on consciousness and mind and the role played by the Neocortex.  I would urge anyone interested in AI or theories of learning to read it.  I used to work with early neural nets and genetic algorithms trying to predict bond and currency markets as an undergrad.  It was a hobby.  Needless to say, I was getting a lot of action as an undergrad:)

The theories postulated here and the potential for moving forward with machine learning is deeply interesting.  Hawkins HTM theory is interesting with its focus on the Neocortex and associative learning.  It treats this thin layered outer coating of the brains the place where the magic happens, converting qualia into biological associative bits so to speak.  This leaves the bulk of the grey matter out of the equation.  It will be interesting to see how the wetware guys think about this new silica approach. 

Hawkins got slammed a bit in a video presentation as just putting up a new recursive neural net.  I think that was mistaken.  I look forward to reading papers and stories about Numenta's approach.  A lot of researchers will debunk the approach showing it to be poor at older tests and data sets that neural nets excel at.  But in the long run, I bet with Hawkins.  HTM won't answer the mystery of consciousness, but it will start making things a lot smarter a lot faster.  The future just started getting here a little faster.

check out an older video of him here or this one from MIT

My Fund!

Years ago I developed an quantitive traded index product for money market funds.  Money market funds are the ugly ducklings of the mutual fund business, fighting it out for basis points.  The amazing fact is that people allocate 20-25% to them.  A huge opportunity.  The idea is to offer the transparency of an index with a few extra basis points from a traded international cash product.

The consumer gets a diversified basket of currencies, no leverage and an open approach to managed currencies that yields far better than the T-bills plus they are used to.   I designed the product from concept to marketing creative and graphics on my own.  Have a look at these pages:

The latest volatility spike in YEN and forex combined with low risk free rates in general may make this product more interesting.

Have a look at the pitch document I put together a long time ago.  This thing could make a mint for an asset management firm.  Read my prospectus and concept Download GO.pdf   .

kind of interesting

Now here is something I just happened to come across when reading about Feynman Diagrams.  The Ulam Diagram which is a way of plotting prime numbers in a spiral on a 2-d surface.  This could really do a persons head in.  So here is my thought take a wolfram CA  and various rules sets and plot the projection in a type of ulam diagram.  Treat the Cycle like and event cone with expanding diameters. 

I bet a prime number generating function emerges with some form of a CA rule set can produce a signature plotted on a complex surface that approximate a prime number generator. 

If CA's can be "proved" closed form (i know they aren't)  then there may be a whole new potential for a functional geometric proof in a "function space" which may be continuous but discrete. 

Ok all of you fanatics out there with a solution to Reimann's Zeta, fire up your copies of mathematica and create a CA phase space mapped to a variant of an Ulam projection. :)

Functional chips

There are silicon chips that perform logic functions that we are familiar with in our computers and phones, they come in 2 flavors digital and analog.  A very fast emerging world of what I call functional chips is emerging.  Functional chips are those that perform some physical function, chemical analysis, mechanical action etc. on very small scales.   

Functional chips are a huge and growing market. Everyday new things are happening in the space.  This isn't nano-tech per se, but can be huge financially.  Nano-tech gets overhyped.  Micro-tech will be bigger first and already is if one includes the logical chip market. 

Nanotech products will probably dwarf micro-tech products, but for now those nanotech products are mostly plain old materials sciences.  The real leap in nano-tech will be industrially scalable platforms with multiple subsystems integrated with interfaces to micro systems.  This is an eco system that needs to be built down from micro-systems and built up from the nano-tech world.  That means bridging 2 order of magnitude in variances and wildly different sets of boundary and physics issues.  Prepare to wait for lots of nano-tech platforms and products at least 7-15 years out.  We are still discovering amazing things with functional micro-tech.

This article on bubble logic indicates the potential for making microfluidic chips more functional and flexible.  When combined with mems technology the toolkit for biochip development and cheap sequencing just got better.

Genetics, pharmaceuticals and Chemistry will get safer, faster and smarter, very rapidly.  Expect more developments in microfluidics and mems technology for altering the chemical environments on chips, micro-environments, cryonics and centrifuges as well as variable pressure environments are all needed. 

Anyone for a disposable micro-spectrometer for on chip analysis?   This will then be proceeded by companies specializing in combining the sub assemblies on a chip custom made to order. 

Eventually these functional chips may be Generalized like VLSI with lots of sub-optimal "wasted" components.  These functional chips will most likely become a bigger market than today's logic chips, when used in massive arrays like factories.

Turing test in reverse

The Turing Test for artificial intelligence is a famous test for the testing of machine intelligence.  I have always thought the debates about intelligence about the test itself are fairly uninteresting due to the anthropic judgment inherent.  Many things can pass a Turing test in limited context, but none so far pass on an ongoing basis.

One interesting way to think of evolving AI is where the test gets applied in other contexts and where the technology is evolving in the real world.  Spam Filters are an interesting inverse function of the test.  A filter is established to try and ascertain the "human interest" level of a message.  If the message is viewed as interesting then the message is passed.  Who knows, Spam Filter related technology as they evolve may be a component of AI passing the Turing test in the future. 

One of the problems with a Turing test is that it is a soft anthropic limit.  It would be much more interesting to have measured "intelligence" scale similar to IQ but designed to assess "post human" intellect.  Something to consider as the singularity is getting closer...

Critical Mass

Critical Mass (How one thing leads to another)

Critical Mass, is an interesting book that posits the thesis that there may be a "physics of people"  The author, Phillip Ball is a very gifted and intelligent individual who is as comfortable discussing phase changes in complex systems as he is discussing the views of Hobbes and Locke. 

In critical mass, Ball articulates that we may be getting closer to be able to predict the responses or paths of large groups of people and or cultures.  He is careful not to put the thesis forward to strongly.  This choice was good.

Attaining a metric or science of people using a reductionist approach like physics would be ambitious on the best of days.  Ball does introduce lots of interesting past attempts at the "science" of people in the collective sense, including concepts ranging from the Leviathan, to emergence as a function of feedback, the dynamics of traffic (autonomous rule based agents), Adam Smith and economics, Critical states and finally Small world networks.

The areas touched on are well researched, but interesting to bring together none the less.  Many dangerous things have happened in the past when "science" is supposedly used to understand people.  For most science is then applied  "technology" and bad things happen.  Whether it is the rational "science" of divine right, eugenics, or Marxism, the concepts of people as being deterministic, naturally invites people in power to manipulate the rules to determine an end.  This is dangerous and has always lead to bad things happening as people are swayed by the "logic" of some argument. 

One leaves this book with a strong faith in laissez faire when it comes to political and cultural institutions.  The Hippocratic oath of "first do no harm" seems to apply not just on the scale of the individual but on the group as well.  The important thing to learn at this point is that there is no science, or governing mechanism of people in a collective sense. 

In constrained environments or for a given time and place people do behave as expected, whether it be a Skinner experiment or the marketers product positioning.  Critical mass as a book is OK and a bit vague in terms of really resolving anything.  It is an interesting read from it sheer breadth and depth and one can appreciate the authors scope and knowledge in selecting the content and thus allowing the reader to reach a conclusion.  Unfortunately, the collected parts of the book are more than the whole.  Critical Mass suffers the irony of being grossly mis-titled, when in the end it is a series of fascinating fragments waiting for cohesion.

Artificial Intelligence

On Intelligence is a well written book that offers a peak at how you think and how things may think in the future.

I started working with Neural nets when I was 20 trying to predict derivatives prices.  I learned a lot about neural nets and genetic algorithms even more about the difficulty of predicting various markets.

I have been fascinated by neural nets, genetic algorithms and machine learning.  All of which is derived from some of the earliest theory work done in control systems.  Thanks Norbert Wiener.

Jeff Hawkins book is great stuff, for 2 reasons, one is clarity of communication, the second for the key ideas themselves.  Basically the thesis is that human intelligence is a function of the neocortex performing a massively repetitive prediction algorithm. 

The prediction algorithm is looking at bits and pieces of things and then matching them up with an invariant memory model.  Invariant memories, the second key concept to Hawkins thesis are the ability to recognize things regardless of context or orientation.  For example you can recognize a dog, from any direction, size or species.  That is a great trick! and incredibly powerful computationally.

The repetitive nature of the algorithm (embedded in your physical neurons) combined with massive feedback and feed forward loops is the thesis for consciousness and learning.  Hawkins approach seems very interesting. He forgoes a bunch of the fairly un-interesting philosophical traps of consciousness and awareness thrown out by types such as Daniel Dennet.  Hawkins has fairly limited time for the Turing test as well.  As an engineer he is more interested in a solution than its "philosophical impacts". 

I agree with this.  Alan Turing was a brilliant mathematician and thinker, but the Turing test itself is too anthropic to be of any use to science.  It is interesting from a philosophy perspective but pointless from an engineering and science perspective. 

The human ego is fragile and many of us tend to respond to the potential of conscious machines with a strong state of denial.  The same type of mentality that was upset when we lost our place at the center of the universe, or as a unique species. Science gets in the way of our self importance sometimes.

If you are interested at all in the potential future of artificial intelligence or even potential storage architectures of the future, Hawkins book may be for you.  Even media as it gets richer will need to be able to sort and search images and complex data sets.  Hawkins start up www.numenta.com is probably going to create something very interesting in the next few years.  Keep watching.

Voronoi geometric brilliance

There is a form of spatial plot known as a Voronoi plot.  A diagram is shown here.  The picture was borrowed from the brilliant people at Wolfram research.  they make mathematica.  Stephen Wolfram is behind the firm.  If you get a chance take a look at his book A new kind of science.  I will make some comments on it later.  It is profound and a peak into a fascinating future of math, physics and complex systems understanding, that we have just started to understand.

The Voronoi tessellation or plot has many applications.  If you have a look at the diagram to the left, you will notice points and vertices which create closed spaces or cells.  This optimal (packing) plot is found in many places in natures and natural phenomenae where physical spaces and points of growth grow to define "boundaries". 

Like many things in nature that are beautiful, the beauty emerges from a set of rules, governing symmetry, growth and pattern while allowing for diversity, strength and emerging complexity across scales.  Imagine if you will each point to be the nucleus of a cell and the edges to be cellular walls.

One of the best sites to explore the Voronoi concepts is The Voronoi website.  the Voronoi concept is used not only in small physically bounded regions such as biological cells but also in astronomy to identify likely galactic clusters (see paper here)  and here as well

The concept also lends itself to many forms of optimization and problem solving for crude or abstract bounded set problems.  The nature of an auto clustering algorithm on a 2d plane is very powerful.  It is my belief that using iterative resolution voronoi plots, will be a way in the future for image recognition as it is the most logical to way to detect variance without use crude spectral analysis. 

These 2 images are an example of high noise dataset reduced using a Voronoi map see here.  The feature set is reduced significantly.  My guess is that the delta of various plot resolutions would very effective for measuring and detecting crude sets or image identification.  This is far more effective and probably efficient than edge detection and then spectral mapping. 

As a clustering approach, something feels right to me about the Voronoi approach and I wouldn't be surprised to see it showing up in a lot of places in the future.  As more analog data needs to be compared and contrasted in complex sets, this approach makes sense.  Now that is all good and well for understanding "static" data matches or comparing static data over a dynamic range.

The other interesting application will be Voronoi plots involving Cellular Automata.  Most CA models use simple regular boundaries, typically 2-dimensional squares.  Wolfram's first work involved this.  However, most phenemonae exhibit radial irregular boundary conditions that don't work for traditional lattice models etc.  The Voronoi method of modeling things allows for "noise" within the micro boundary and positioning of elements, but gives a quick solution to synthesis and modeling.  As of today no, language of Voronoi plots has evolved.  My belief is it will.  This is a rich field for exploration,  Voronoi dynamics, fractal metrics and perhaps even a voronoi excited state model yielding something equivalent to a Feigenbaum constant remain to be discovered.

And here for a big finish is a really cool app call the Bubble Harp  Scott Snibbe has made some fascinating works here.  They aren't toys, they are windows on a rich and fascinating world.  For fun install the bubble harp app and hold down your mouse as you create vertices.  the resulting action almost looks like a beating heart :).  The Voronoi concept is beautiful  and will be  profoundly revealing for years to come.

Network theory is cool and connected

Network Theory is fascinating stuff. The thing I find most interesting about the mathematics and applied mathematics of network theory, like graph theory is the fact that only in the 20th century have we really taken a look at networks.

Network theory is deceptively simple and yet powerfully revealing, quite a bit like a good bikini :)   Basically from a purely mathematical construct a network has 2 components Nodes, and connections.  That is it, the graph shows some basic networks.  The amazing things is the dynamics of networks.  In theory any causal event or chain of causal events can be modeled with network theory.  Social interactions can be model led with Network theory.  Yes you really are only 6 people away from all the people on the planet (six degrees of separation)

Most natural phenomenae exhibit a unique aspect or type of network in that the connections and connectedness of the nodes follows a power law, this type of network is known as a scale free network, now before you think this all theory and math, please realize that the connectedness of the web follows a scale free network model.  Most dynamic systems with independent components settle into some form of a scale free network.  Thus understanding them is important.  Whether modeling the spread of disease in a population of people, the transmission of ideas or the dynamics of some biological systems scale free networks hold the key.  During the summer of 2002, I read a few papers and books on the topic, two of the best, I would recommend are Sync by Stephen Strogatz and Linked by Laszo Barbasi.   Both of these men are fine mathematicians in their own right and great popularizers of science.  They have excellent skills at communicating and sharing some of the amazing worlds and phenomenae they have discovered.   

The more interconnected we as a culture and species become the more important it is to understand network theory.  I think network theory will eventually become as important as thermodynamics.  The thesis being that modeling individuals nodes or components will be useless, but understanding complex systems in aggregate will allow for powerful understanding and control of those systems.  Network theory and the emerging metrics and models associated with it will lead to whole new theories of "things" and relationships.  Much like thermodynamics allowed us to scale up from Newtonian reductionism and ultimately harness energy, Network theory could very well allow us to model and control some of the most powerful networks of all.  The list is long, but could include, neurochemical, social, environmental and communications networks.  The implications and impacts will be profound.  And it will have started with a Connection and a node.  Check out these books if you get a chance, I personally fascinated by social networks, memetic spread and of course financial networks.

One of my friends from Starlab is Keith Still, he works primarily in Crowd simulations and modelling.  He is considered one of the experts in this realm.  Keith is doing soem really interesting stuff in terms of modelling autonomous agents in an environment to assess their behaviour, anyone interested in crowds, safety or cellular automata may be interested in visiting his page or emailing him.  Keith is a great guy and very generous with his time.

Out there! Science

I used to be Chief analyst at a place called Starlab, best job I ever had.  I am a huge science and engineering fan. I use the word fan, because I am only a bystander from the sidelines.  I have a deep love of structures, order, pattern and relationships in things and people.

One of the more interesting projects I have been watching over the last few years is Black Light Power .   The thesis begs us to question classical quantum mechanics and the nature and state of some of the constants that make our model of the physical universe.  For those of you with a deeper Physics understanding, I would be fascinated to hear a critique of Mills Grand Unified Theory. 

The concept of a compressible state and fractional Quanta seems pretty wild.  The impact for physics and our understanding of a lot things could be profound.  So far the larger world at least Nature, Science etc. seems to be ignoring the work, albeit the less skilled popular press has picked up on it.  Almost free limitless energy tends to attract the press.  Anyway it is fascinating stuff and Dr. Mills has promised some major announcements in the coming months.  I am interested in seeing peer recognition and of course applications as the could be profound.