Right On!! Prices are not coin flips. Prices are predicated by decision, and decision is not random. That is not to say we are not wrong more then right, but one does not throw a dart at his next trade. Further more unlike a random game of chance, what occurs in one measurement of time may directly influence the outcome of the next bar. As Chasinfla illustrates, one may merely look at a long term daily, and empirically conclude that trends are not random. The failure of yabz's analysis is that the depth of price change is not noted. Up, down, up, down, down,up,up does indeed seem random. But the coefficients assigned to those days can be wildly variable. +1, -3,+1,-8,-4,+3,+1 and voila, you have a trend in prices, with a "normal" distribution of up, down days. Trading is more akin to blackjack, then coin flipping. At certain times (full deck), the out come is virtually random. With only higher cards left the predictive outcome will trend, to the players favor.
"The market went from March 28 to July 29 without three consecutive daily rises in the S&P 500 ($INX). Then, it had four within three weeks (ending July 30, July 31, Aug. 8 and Aug. 9). Four months of a dry spell, then four streaks in three weeks -- sort of like life itself, but highly non-random." Victor Niederhoffer (beloved by all on ET) MSN 8/22/2002
prices are probably not random, IMHO. if a good discretionary trader was to play stone-paper-scissors with an average trader...he would probably win despite the fact the result should be random... to say that because a series of closing prices fails the runs test that price movement is random is stretching the test too far... how about testing the range of daily prices? the time of day of the high/low? the size of gaps, the location of the high/low, the location of intraday pivot points, the day of the week of highs etc etc etc... i don't think we can prove randomness (i'm no mathematician so laugh if you want) only that trading based on some types of information is more random than trading on other types of information. Q1
Not trying to be picky, but I hope you don't mean that the markets are random and then you need to "predicting something in the markets with a higher than random probability"? If the "markets" are random, then no amount of predicting will mean anything. One thing you should understand is the idea of a "hidden variables." Understand the seminal work that was done by Bell when he set out to prove EXPERIMENTALLY that very concept in Quantum Mechanics. Clearly, in trading markets profitably, there _ARE_ "hidden variables." nitro
If the "game" is truly random with no hidden variables, and the number of trials is "large" then I can tell you with extraordinary accuracy what you are likely to see. That doesn't mean that one player won't win, only that I can give you the probability of your prediction before the game begins. Honestly, stop talking about random markets, otherwise REAL traders would pack and go on to other careers. nitro
the markets are not truly random, they are chaotic. When you use simple linear statistics the markets look random, but that is only because the statistical model you are using is too simple. There is an entire field called Econometrics that has attempted to predict the economy and the markets using linear statistics and all the Phd's in the world have yet to find a perfect model. And if they ever did find that model, it would only work for a short time until the use of that model by large funds would alter the market and the model wouldn't work anymore.
Even if price movement was 98% random/unpredictable- a theoretically viable assertion, though not one I make here- the 2% 'clarity pocket' still in play offers a window of opportunity wider than a football field for skilled traders. Random walkers think they are right because they are mostly right. But because they are not 100% right, they are dead wrong. p.s. it is quite possible that the late great bull market provided such an extraordinary bias towards long positions that it was like a coin with a 75% heads bias rather than an even distribution of heads/tails flips. Random tosses of a biased coin, with the masses properly oriented towards the heavily favored side, probably goes a lot farther in explaining the temporary success of the short term trading masses than any collective sense of trading ability ever could.
For those of you interested in a more analytical discussion of whether the markets are random, take a look at the recent thread entitled Random Trading, posted 8/17/02. http://elitetrader.com/vb/showthread.php?s=&threadid=7793 In it we concluded that for 102 equities on the NASDAQ markets there was a downward bias in the ratio of no. of down days/no. of updays in the last 100 days which in a perfectly random system would be 1. The actual ratio came out to be 1.32 +- 0.1 A number of you have pointed out (Silk in particular) that you could use this information in a trading system. If so, lets here more from you on how this can be done.
Having spent countless hours watching the behavior of ants and their movement, I can truly say that ant movement is entirely random. I can't figure out what the hell they are doing. You would think they would go for the fruit, but they go for the water in the sink. You would think if I killed them, they would spread the message and never return.....Wrong!!!! I just can't figure out those ants and what they are going to do from one day to the next. However, if I watch ant movement during a very brief interval, and the is ant is headed towards the sink, or away from the bug spray, I can pretty well predict which direction he will continue to move in for a brief period of time. The study offered in this thread as proof of randomness was done on a closing to closing basis, not on intraday movement of an index of a stock. We can pretty well predict what will happen to the price of stocks if there is a surprise rate cut once it is announced minutes before the market opens, and if announced during the trading day we can pretty well predict what will happen on the initial movement and where momentum may take it, but where it ends up is an unknown. We can pretty well predict what the price of stocks will be near expiration date of options, as many do based on maximum pain numbers. There are other such studies. I can pretty well predict that if I yell fire in a crowed theater, people will run, and I will go to jail. I can predict that if I say a girl's ass is fat I won't get any of it . So, while many things are indeed random---ant behavior, women's moods, GW Bush's thinking, many things are not.