Did Deep Blue predict what Kasparov was going to do or simply take the best move available in its memory? If you are familiar with your opponent you will probably try to predict what he will do but that is not the only way to play the game.
Yes. One's opinion of future outcome does not go into the pulling of the trigger. The trade is taken because the set up is there. The prediction of the set up's value was done prior to taking the set up into the live market. So do all the predicting you want, as researcher or back tester, or however it is one gets to the point of having rule sets he believes in. But when it is time to trade, take the trade because the rule set shows itself. Without regard to outcome. 1. I don't use indicators . I use price and volume. 2. As I said, I am not there yet. I am talking about a master level trader. I know I am not one myself.
Deep Blue uses mostly brute force attack. It chooses the best move by predicting, via live simulation, its opponents best possible moves, future board strength for all said moves, as far as it can calculate within time limits. Of course its predicting. Every time it makes a move, it does it because its internal number crunching predicts a stronger position or material gain. Its a prediction, because it cannot see every possible move ahead, otherwise it would simply absolutely know all the possible outcomes for every choice he had left. "...or simply take the best move available in its memory" <--- is just a way of avoiding using the word predict. That move exists in its memory because it predicted as many possible outcomes as it could and chose the one with the highest PROBABILITY of success. AKA prediction.
Another analogy for the "unthinking now" guys who trade on instinct and in the moment, and claim no prediction. When a martial artist starts his training, he is taught to block a left hook, maybe by raising his forearm/elbow to block. When a left hook is thrown, he thinks about it, and clumsily raises his arm to block. Years later, he has drilled this so much, that when a left hook is thrown, without thought, he simply reacts, raises his arm and blocks the punch. My point is, whether or not you methodically and consciencely think about it, or do it subconsciencely via trained instinct/reaction, you are still predicting that you will block the shot and avoid damage. Why else would you block the shot? Burning the once explicit thought process of blocking the punch into the instint/reactionary part of your brain doesnt change the fact that you re still predicting. Just the same BS semantics being played.
Why does this not surprise me? And why do neophytes argue with such conviction about things they know so little about?
So there's NO expectation of a future outcome based on research of how a setup has performed in the past? You'll go far as a trader with a limited ability to reason and an unlimited ability to deny you're wrong.
HAHAHA... AGAIN MON FRERE you omit some details! The biggie of course being STOCHASTIC functions and STOCHASTIC CALCULUS! You know, ITO, Stratonovic et al... You can google it and check it out. Such math came about ironically to deal with physics and not so surprisingly, the stock market. However, if these are unfamiliar, look at 1 period sma. That is a smooth continuous function. There is a lot of stuff that I look at that I'm sure are different then what you backtest. Just like the weather, maybe you don't look at the barometric pressure. These are the sort of differences where you get into leading type of indications, kind of like the precursors to rain (ie. thunder in the distance, cumulus congestus clouds). Hence, my forward tests are the absolute value of your backtests. So, here you have pointed out yet another difference between you and I. Whereas your rain drop is P now and past, mine is a vector bundle of independent now variables. To type slowly too, I guess, I look at trend lines. So when P is above my right trend line that is long weather. When P is below my right trend line, that is short weather. So here is the real T6 gotcha. My RIGHT TREND LINE is CONTINUOUS across several discrete data points. This is like a two for one thing. I plot a continuous function over a series of critical points. The result is that for every future P point that scrolls into the present, it is a comparison of the current P to the CURRENT POINT ON THE RIGHT TRENDLINE that is directly below my now P... MON FRERE! I get two NOW points! HAHA! Relax, consider some things. You can turn that negative EQ curve to positive yet... 30 years 359 days to go on our forward test... We're going to get there...