Did I try to characterize Taleb's approach ? Who can in one line ? What I want to refer is what he says in his book "fooled by randomness" (see review below from a mathematical site below) WHICH I AGREE WITH: people want to see patterns everywhere although they have no really mean to distinguish between patterns and randomness. I have an OBJECTIVE and RATIONAL model so I can say when market really exhibits patterns (deterministic behavior) but what if I haven't this model : I WOULD HAVE NO REASON to believe in patterns because I am not a believer I am a rationalist. No proof no belief this is a PRUDENTIAL attitude. This prudential attitude comes from statistical decision theory and must not mislead to think that when you have no statistical significance it is a proof that the thing doesn't exist. It means that you have no proof TODAY but that the proof could come TOMORROW if it ever comes. As for ME it comes from my model but for the HUGE majority it is based on vacuum. Let's say that you trade "sucessfully" patterns you would conclude that patterns exist. First as Taleb also says in his book this can be an illusion that you are successful (I already explained in another thread the Levy's law about persistency of chance or bad luck which would make people believe that they are good or bad when it is only due to randomness) and secondly I would add that it can come from money management as market's rules allow you to stop your bet before the ultimate result unlike Casino's bet. http://plus.maths.org/issue20/reviews/book1/ "There are many ways of being the fool of randomness. One, as here, is to fail to predict the rare event. Nothing can be more certain than that the unexpected will happen sooner or later, but lulled into a sense of security by the periods of relative calm between, people forget to allow for it. Another is to see significance in some random pattern. Taleb explains with crystal clarity why the more often you look at some fluctuating quantity (the value of your share portfolio, for example), the less meaning your observations have. Yet he sees traders who watch prices move up and down in real time on screen - the changes are so small as to be completely random - and think they are learning something. "
Our analytics engine spots patterns all the time. The real issue is what happens after a certain pattern shows itself. Is there any statistical significance/advantage to seeing these patterns. http://www.trade-ideas.com/DOSA/
Let's precise in the mind of Taleb (and myself and statisticians in general) what Patterns would mean at least in its restrictive sense. Patterns musn't not be confused with statistical anomalies - If there wasn't any statistical anomalies statistical decision theory would be useless - and since Taleb bases his trading on anomalies he surely doesn't mean that you cannot trade market's inefficiencies -or what he and people commonly think as inefficiency whereas I have an other point of view on what efficiency really means - Patterns can exhibit with or without statiscal anomalies. In the case of no statistical anomaly that is suspect to believe in pattern like head and shoulder (since any simulation of random walk would show frequently a fake head and shoulder pattern which is not a true pattern in this case). So with statistical inference it is alone difficult to assert the existence of "true" patterns. Only statistical inference with an hypothetic deterministic model (I claim that I have found one but since I don't want to oblige everybody to believe me let's do as if it doesn't exist ) can prove it decisively. So as long as a deterministic model is not found and get an academic recognition will the existence of true patterns in stock market difficult to prove. Anomalies and local or short term inefficiencies are much easier to prove since it involves only statistical inference approach. But for statisticians it is just temporary and wouldn't be considered as inefficiencies since it wouldn't last (from the official point of view because this has been a debate especially from Mandelbrott).
Professor Harry...Can't we simplify the thing and say "All TA is true except for the conclusions we draw"> From long and bitter experience I 've finally found that after all the TA analysis the thing to do is the exact opposite ...amazing how well it works out...statistically off course.
Not from "professor" Harry but from professor and economist Lemennicier (in his book "Art of reasoning") Inductive Inference Contrary to the deductive inference, an inductive inference goes from particular to general, it generalizes starting from particular facts and analogies. Induction is, however, paradoxical; we must use a syllogism to pass from minor premises or particular propositions to generalization. Let us take the example of the sun. Since you were born, you saw the sun rising in the morning... You conclude that the sun will rise tomorrow, as it did every day until today. This induction can be written in the following way: What the sun did in the past, it will do it in the future. Since always it is observed that the sun rises the morning. _________________________________________________ So the sun will continue in the future to rise in the morning. Thus, the major premise (what arrived in the past will be rpeated in the future) supposes a regularity, a uniformity or a repetition of the events observed. However, we cannot pass from the minor premise to the major premise; all the observed facts, even the most regular or the most uniform, cannot be used as proof to establish the truthfulness of the major premise. An identical difficulty is observable with the statistical and econometric analysis. To prove it, the following example: It was observed that the prices of the actions on a stock exchange market follow a random functioning (minor premise). Any statistical law discovered in the past will be also observed in the future (major premise). _____________________ So the future prices that will be observed on the stock exchange markets will follow a random path. Actually, the statistical analysis of past data does not bring any obviousness or proof making it possible to justify a regularity of this statistical law (random path of the prices on the stock exchange markets) in the future. Referring to what we said on the impossibility of prediction in economic theory, we know that if the prices already reflect all available information and what we can anticipate, then the variation of the prices can only represent what is unforeseeable! Since what is unforeseeable follows a random path, the time series of the price changes follow random processes at every moment. The regular observation of those on the stock exchange markets results not from the past observation of this statistical law, but from the economic theory of the impossibility of prediction on the financial markets.
That's why I said : "No proof no belief this is a PRUDENTIAL attitude. This prudential attitude comes from statistical decision theory and must not mislead to think that when you have no statistical significance it is a proof that the thing doesn't exist. It means that you have no proof TODAY but that the proof could come TOMORROW if it ever comes. "
TA provides a framework upon which opinions, actions, and non actions can be executed or formulated. without a framework, the markets appear chaotic to the human brain. best, surfer
OK. How about turning the question around a little. is there any reason why it wouldn't ever work? Now that would make for some interesting replies best regards Natalie