September/October, 2006 Post Roundup

This blog is more of a note taking intrument / reminder system for myself to collect and consolidate the thoughts in my head than anything else.  I've posted about one post a day, and it's getting a little hard to keep track of them all, so I thought I'd begin a tradition to summarize the previous posts in one concise spot for future reference and easy browsing.  With that, here is a summary of everything so far:

I started with an explanation of CASTrader and the purpose of this blog.  Later, I followed up with a post on the philosophy of CASTrader.

Kelly and Shannon.  Every trader in CASTrader uses the Kelly Formula, which features prominently in William Poundstone's book, Fortunes Formula, which I reviewed in Part I of a series.  In Part II, I looked at whether the Kelly Formula is the best method to manage money.  The Kelly Formula has a lot of caveats and pitfalls for those trying to use it, but I concluded that because of the design of CASTrader, it was ideal.  In Part III, I discussed some other tidbits from Poundstone's book useful for CASTrader, particularly the Kelly Formula itself.  I also discussed Claude Shannon's discovery of the constant proportion portfolio, which I have experimented with and have used for about 3 years, well before the book was published.  I later followed this up with a more detailed post on what I call the Shannon Method and how I will use it as a kind of market maker on the commission-free CASTrader dark market (more on this below).  I also posted some examples of Kelly Math with a drawdown equation to illustrate what many find so uncomfortable with using the Kelly Formula.  I also posted a bunch of links for further reading on the Kelly Formula.

Alpha.  When you get right down to it, alpha is really the reason this blog exists at all.  CASTrader is being created to hunt for the elusive alpha.  All About Alpha discussed a theory he dubbed "peak alpha," (which would imply the world is running out of alpha) and since alpha is of primary interest to CASTrader, I explored Peak Alpha theory at length.  Of course, alpha is zero under the Efficient Market Hypothesis, which is why I'm more of an adherent to the Adaptive Market Hypothesis (AMH), which predicts there is alpha in the market after all.  The good news predicted by AMH is that alpha is in perpetual supply, but you have to innovate to find it.

Dark Markets and Partial Diversity.  In "The magic alpha of dark liquidity and partial diversity," I explored two concepts.  First, I explored the growing use of dark markets and dark liquidity - trading that never sees the light of day in the NYSE, NASDAQ or other public venues.  Big investors have many reasons for going dark, including saving exchange fees and other frictional trading costs.  Since CASTrader will be a collection of strategies, it makes sense for them to trade on a dark market.  Something magical happens when you get a bunch of diverse strategies consolidated into one "net trade" via a dark market.  I have dubbed this phenomenon "partial diversity," and I believe this is a source of alpha boost.  Previously unattractive strategies can become attractive.  So far, I've posted Part I and Part II of the design of the near frictionless dark market to be included in CASTrader.

CASTrader is born.  In a four part series, I examined the good, the bad, and the ugly features of the backtested performance on the DOW from 1915 to 2006 of my first version of CASTrader I dubbed "CASTrader I."  My basic conclusion was that CASTrader I is able to find tradeable information in the market based on price action, especially in recent years, but has definate room for improvement. In "Don't worry, be happy," I outlined how I will create CASTrader II, and how it will probably be able to go live slowly with real money hopefully by January, 2007.

Trading Patterns.  One eventual goal of CASTrader is to load it with as many trading methodologies to mine alpha via partial diversity as possible, so in a series of posts, I've put together a sort of link digest of various methods that point out miscellaneous statistical anomolies that might be exploitable here, here, here, here, here, here, here, and here.  One recent growing phenomenon is using computers to mine the web for tradeable news.  In trading on the news, I posted a roundup of links.  Behavioral finance is certainly a rich area for exploration, with link roundups here and here.  Also of interest is the relationship of stock performance and earnings.  Even the (voodoo to me) world of technical analysis is fair game for inclusion in CASTrader, and CASTrader I used it exclusively mainly because it was easy to program.  In Part I of a series, I asked "Can technical analysis beat the market?" and reviewed academic studies that addressed that question specifically.  In Part II, I summarized the lessons of these articles, discussed the implications of datasnooping and the Adaptive Market Hypothesis for CASTrader.  In Part III, I discussed how CASTrader will use technical analysis.  Lastly, there's a post roundup on the Halloween Effect.

Data.  One thing CASTrader will eventually consume in abundance is data.

Miscellaneous.  Though I try to organize my thoughts, you can't find a category for everything, so here are some purely miscellaneous links.

Measuring Performance.  Performance isn't just about measuring annualized returns.  A lot more goes into it than that.  In Measuring Performance, I looked at a roundup of various performance measures and presented my own simplified eyeball method that I will use with CASTrader.

Machine Learning.   CASTrader will be an example of an application of the field of machine learning, so this is an area this blog will also focus on.  In Direct Reinforcement Learning, I provided some links and advantages/disadvantages of an alternative method of machine learning than the one adopted by CASTrader.  In the Creativity Machine, I looked at a company that uses neural networks in a creative way.  Neural networks are another alternative to the Complex Adaptive Systems (CAS) approach of CASTrader.  Machine learning and machines in the markets linkfest is self explanatory.

Inspiration.  Finally, it's helpful to look at some of the people who demonstrate that this sort of investment philosophy can be done with rich success.  In Super Quants, I profiled three quants that have "made it."

That's it - some 40+ posts.  Even though my desk is a disorganized mess, I can now say my thoughts and my blog are not.  Back to coding.