themickey mouse degree(?) : the probem with psychology is just now - thanks to neurosciences- psychology is being ionvestingated with scientific tools and methods. So keep your mind open, BUT ask questions, a LOT of questions.
@MAESTRO I am really enjoying your wisdom, please keep up the good work. So, actually do you believe in any strategy with positive expectancy based only on historical OHLC without any psychological aspects? I am asking because building IA demands specific kind of data (ie. order flow) not available to everyone or not showing the whole picture for decentralized markets like FX. According to your ealier posts I can assume that there is no such possibility as markets follow RW and price is a result of events, not a reason.
Sorry, I had no time to post the past week or so. I got extremely busy with all these trials and experiments that are going on currently. To answer your question: I do not believe that price vs. time charts are enough to make any informed trading decisions. As the matter of fact, I have tremendous amount of data that suggests that any equity curve of any mechanical system based on the price data alone is pretty much a Random Walk. And, of course, as with any RW there are runs that are fairly long and very positive. However, an RW is still an RW! In this light my research has always been focused on finding the âhuman imprintâ on the flow of bid/ask/size/price/volume stream. Unfortunately, I cannot tell you that my results are 100% formal and could be presented in a form of an algorithm that mechanically could filter and utilize the behavioral patterns of market participants. Understandably, those patterns are too complex to formally model them. So, I resorted to the next best thing â utilizing the natural abilities of a human brain to spot patterns that cannot be successfully logically described using sequence of âIf â Thenâ statements. The notions of âperceptionâ and âintuitionâ were used to identify those patterns. Of course the problems of this approach are consistency and repeatability of the results. That is how the concept of Intuition Amplifiers had come about and developed into a variety of tools and methods to increase the correlation between the tradersâ intuitive responses and the upcoming price developments. Rather than trying to explain why a trader makes this or that decision using different forms of visualization of the market data we are focused on the increased reliability of that process and creating the methods of training the operators to perform with a significantly positive expectations. I hope I shed some more light on this topic. Cheers, MAESTRO
Thank you for response. I read most of your posts on ET more than once, this stuff is just incredible and eye-opening. In my opinion information located in your pdf paper "Intuition Amplifiers" is more than enough to recreate (to some extent of course) practical tool for measuring market emotions. I think it's all about searching for abnormal activity of traders and exploiting it. Event-based timing is such a great thing, mostly related to news trading at its origin i think. In news trading the most important rule is a shock generated from difference between forecasted data and real data. Many people state that direction of the asset during that event doesn't even matter. ROC in volatility should be high enough to create spike, which enables a kick of the order acquisition and take profit almost simultaneously. I was thinking if we can predict possible volatility jump using IA just before news reveal. It's important to be in queue to avoid requoting during order placing. In FX market volume doesn't really matter, so maybe measuring tick speed per minute could help solve the problem.
PM me if you are interested in experimenting with the IA based dashboard. I can set you up. There are a few ET members who are currently experimenting with some of my tools. It would be interesting to see what your experience turns out to be. Unfortunately only 3 out of 10 traders get it. But those who do usually become hooked on it. Let me know. It could be fun!
One of the most important types of Perception is the perception of probabilities. A typical trader often has an inaccurate perception of probabilities in everyday markets. In fact, several studies have shown that the probability notion is one of the hardest to grasp. For instance, if after flipping a coin that turned up heads 5 times in a row, we ask a child what will happen in the following flip, the most probable answer is that the child will opt for tails. The adult will have no difficulty in giving the correct answer to this simple question (that is, either heads or tails will turn up, meaning that both events are equally likely), but if the question gets more complicated an adult will easily fall into the so-called gambler's fallacy. An example of this is imagining that one should follow technical analysis indicators either when a stock is âdue to retraceâ or when a stock is âin a trendâ. Knowing that majority of traders is going to follow their perception it is easy to construct a tool that follows the prevalent perception thus creating a stable association with the upcoming price behavior patterns.
Here are a few classic questions to test your perception of probabilities: 1. A box contains green and yellow buttons in unknown proportions. I take out 10 and 8 are yellow. Which am I more likely to get next, green or yellow? 2. In a group of 50 people what are the chances that 2 of them have a birthday on the same date? 3. The last four days S&P 500 index was making advances. What is the probability that on the fifths day the markets will retrace? I have collected over 1,500 responses from traders over last few years. The results are astonishing! Cheers, MAESTRO
There are problems on probabilities that lead to apparently counter-intuitive results and are even a source of embarrassment for experts and renowned mathematicians. Those are the best market indicators ever! The probability of correlated perception is much greater than any probability observed to date in the markets! Once understood, those intuitive conclusions about other people perceptions could be very powerful. That is what makes such a reliable basis for the IA based tools.
Letâs look at the problem of finding the probability that a portfolio with two stocks has at least one winner. The probability is 3/4, assuming the equal probability of gain or loss. Let us now suppose that we have got the information that the randomly selected stock is a winner. Our question is: what is the probability that the other stock is also a winner? Using Bayes formula, we have P({WW} if {WL, LW, WW}) = 1/3. This result may look strange, since apparently the fact that we know the P&L of one stock should not condition the probability relative to the other one. However, let us not forget that there are two distinct situations for Winner-Loser, namely WL and LW. We now modify the problem involving the picking a random portfolio of a random hedge fund and getting the report that the first position reported to us is a winner. What is the probability that the other position is also a winner? It seems that nothing has changed relative to the above problem, but as a matter of fact this is not true. The event space is now {W*W, WW*, W* L, L* W, LW*, L* L, LL*} , where the asterisk denotes the position that was reported first. We are interested in the conditional probability of {W*W, WW*} if {W*W, WW*, W*L, LW*} takes place. The probability is 1/2. We conclude that the simple fact that the first reported position was a winner has changed the probabilities!
Consider a 5 stock portfolio assembled randomly from a 50-stock universe. In this universe there are four winners out of which one is enormous! Suppose that one of the stocks we picked is a winner. Let us further specify two types of portfolios: in one type, the winner is huge compared to other winnersâ average gain; in the other, the winner is average. Suppose that in either category of portfolios we are interested in knowing the probability that at least one more average winner is present in the portfolio. We are then interested in the following conditional probabilities: P(two or more winners if one of them is huge) and P(two or more winners if one of the winners is average). Are these probabilities equal? The intuitive answer is affirmative. After all a winner is a winner regardless whether it is huge or not. With a little bit more thought, we try to assess the fact that in the second case we are dealing with any of 4 winners, suggesting a larger number of possible combinations and, as a consequence, a larger probability. But is that really so? Let us actually compute the probabilities. It turns out that for this problem the conditional probability rule is not of much help. Can you find the right answer?