Mike, do you think you have found people to fool around with? - You did not produce a random walk from a set of rules. You mixed a random walk AND a set of rules. There is a huge difference in what you claim to have done and what you actually did. - It is not easy to look at a price move and conclude that it is random or it is not. Since you believe otherwise, please state the quantitative test for that. "Looking" at a move is not nearly enough. I am surprised you think you have proved or showed anything at all by mixing a pseudo random number function call with a deterministic set of rules. The real issue here that you are obscuring for whatever reason is the following: You have a profitable system based on deterministic rules and you add a randomizer of the entries/exits. If you get better performance on the average is it safe to say that the original system was worse than random? I do not think the answer is easy.
So to summarize a little, the idea of the random condition was more to try and isolate other possible non-random criteria (time for example), as opposed to trying to divine some magical property of the randomness. Is it possible to expand on what you mean by a 'non-random distribution'? I'm going to be off the Internet in Rarotonga for the next couple of weeks, but I look forward to seeing how things have developed when I get back!
To some extent, yes. I like your "divine same magical property of randomness" statement, and, to that I say: everything has been done before, and you can probably find it on the internet Suppose one wants quantify the effect of a "insert concept here-" based condition in the market. Suppose that condition, when "swept" via permutation of the pertinent variables, shows a cauchy-like distribution in the results. One might argue that the condition in question is unstable. The nice thing about a perfectly uniform random distribution with fixed bounds is that it can be used as a baseline for another conditions' utility. Does your added condition "behave" or is it unstable? Most poor systems have conditions that fall into the latter category. That's one of my conclusions that's open to debate. That said, there are properties of certain classical statistics problems that can be exploited and they're not magical, they just require some different thinking.
Profound statement because it is 100% meaningless: http://en.wikipedia.org/wiki/Distinction_without_a_difference
Hi Dave, Still having trouble with patterns? http://www.elitetrader.com/vb/showthread.php?s=&postid=2918703#post2918703 Maybe I should try to explain better... On second thought, nah, you sound like you're in intradaybills camp. Still believe this dave?
I see you won't give up with your idiotic statements. If you want to make sure that we know by now you are a idiot you are doing a good job. I though intrabill likes patterns and I do not believe, at least not any more, in edges. So how can we be in the same camp?
Coin toss enthusiasts, Check this out! http://www.mrexcel.com/forum/showthread.php?t=503375&goto=newpost
About coin flipping. I use it to test simple trading system permissions performance vs to the random noise. E.g. Pivot support or resistance can be tested to know, how strong barriers are that. You buy et fix price (e.g. support+0.1 ATR(7), sell buyprice+/- ATR(7). compare this win/lost rate to that when you buy and the midle of recent support/resistance range. You will be see that buying at support can be ~2% - 20% edge against the random noise (buying equally away from support and resistance). intradey pivot supports have the weakest barriers, only 2-3% higher winrate confirmed, which can be deminished by the slippae and commission!!! I found knowing volume is not an edge against not knowing it, when buying at suppport. However many books propagate this, but if you test it, you get no confirmation.