“ That means the expectation value based on the ensemble mean of the first coin toss is (0.5x$50 + 0.5*$-40) = $5”
This reasoning is wrong. This is not the expectation. In fact, the expectation is zero. Please see: https://medium.com/@mikeharrisNY/gamble-expectation-and-the-ergodicity-conundrum-ffe3a523347b
Another gem by N. N. Taleb. The fact that under extreme time pressure subjects answer questions in an IQ test goes against the mode of operation of successful scientists and politicians who take their time to study the problem, listen to alternative view points, do their research. A measure of “fast thinking” in the sense of pattern identification…
You are not predicting, you are over-fitting. See this article why. Predicting stock prices for profit goes far beyond fitting a model. You have to assume a strategy and calculate the Sharpe ratio.
Excellent article. Based on CVX chart I’m looking at on my system, it seems that your validation sample starts around 01/2016.
How would you implement this model to actually trade the stock? Can you outline the basic steps in your reply? Thanks.
Thanks. F_0*(b^n) grows rapidly but Dr. Peters uses a small sample n = 52. As I show in my second article, after a sufficient sample the wealth drops near zero.
Dr. Peters asserts that he has found a game with positive expectation but with negative time average. Therefore, as he claims, Chernoff and Moses statement is violated. In this and another article I tried to show that this was a bad choice of an bet example to claim that it violated Chernoff and Moses. In fact, when sufficient samples are considered…