Tuesday, April 18, 2017

Paul Tudor Jones On 'Imperfect Information'

As the outro to yesterday's "Understanding Algos: "Why This AI Poker Player Is Crushing It"" I took this bit from the linked piece:
...More technically, poker is an example of an information-imperfect game. Sandholm and Brown go deeper in a paper presented earlier this year at the AAAI-17 Workshop on Computer Poker and Imperfect Information Games in San Francisco. The technique described in the paper for such problems is a variation of what's known as "endgame solving."

In a information-complete game like checkers or chess (or Go), the computer's strategy is to decompose the larger game into a bunch of smaller games. Iteratively solve those problems and you should wind up with a solution to the whole problem. This is endgame solving as it's usually understood. Checkers was definitively solved in this fashion.

"In imperfect-information games, endgame solving is drastically more challenging," Sandholm and Brown write. "In perfect-information games it is possible to solve just a part of the game in isolation, but this is not generally possible in imperfect-information games." Imperfect-information games have to be solved as whole entities, not through decomposition....
and went spinning off with a couple earlier references to 'decision-making under conditions of incomplete information' but forgot one of the best.

From our 2016 post "Global Macro: Paul Tudor Jones Interview at Institutional Investor":

...What’s so special about macro hedge fund managers?
I love trading macro. If trading is like chess, then macro is like three-dimensional chess. It is just hard to find a great macro trader. When trading macro, you never have a complete information set or information edge the way analysts can have when trading individual securities. It’s a hell of a lot easier to get an information edge on one stock than it is on the S&P 500. When it comes to trading macro, you cannot rely solely on fundamentals; you have to be a tape reader, which is something of a lost art form. The inability to read a tape and spot trends is also why so many in the relative-value space who rely solely on fundamentals have been annihilated in the past decade. Markets have consistently experienced “100-year events” every five years. While I spend a significant amount of my time on analytics and collecting fundamental information, at the end of the day, I am a slave to the tape and proud of it....
... MUCH MORE Oops, that link has died. Fortunately the II story was backed up on one of our servers and the interview will continue after my 2009 verbiage....
...Continuation of Paul Tudor Jones interview at Institutional Investor:
What's your take on the next generation of managers?
I see the younger generation hampered by the need to understand and rationalize why something should go up or down. Usually, by the time that becomes self-evident, the move is already over. When I got into the business, there was so little information on fundamentals, and what little information one could get was largely imperfect. We learned just to go with the chart. Why work when Mr. Market can do it for you? These days, there are many more deep intellectuals in the business, and that, coupled with the explosion of information on the Internet, creates the illusion that there is an explanation for everything and that the primary task is simply to find that explanation. As a result, technical analysis is at the bottom of the study list for many of the younger generation, particularly since the skill often requires them to close their eyes and trust the price action. The pain of gain is just too overwhelming for all of us to bear!
If interested, the whole interview is worth a look.

Additionally, yesterday's "Why Markets Are Inefficient: A Gambling ‘Theory’ of Financial Markets for Practitioners and Theorists" touches on the concept:
The purpose of this article is to propose a new “theory,” the Strategic Analysis of Financial Markets (SAFM) theory, that explains the operation of financial markets using the analytical perspective of an enlightened gambler. The gambler understands that all opportunities for superior performance arise from suboptimal decisions by humans, but understands also that knowledge of human decision making alone is not enough to understand market behavior — one must still model how those decisions lead to market prices. Thus are there three parts to the model: gambling theory, human decision making and strategic problem solving....