lol, thanks for demonstrating you made all this up and really know nothing. Black hugely contributed to the success of the GS quant group and Haug also did his part in JPM Chase. E.Chan really has nothing to show for whatsoever. How can you even mention those three guys in the post, lol.
so what is it then for? -how to extract data off free websites -the chapter on risk management is 12(!) pages, lol in the end he tosses in couple techniques (stationarity, cointegration, factor models, seasonal trading, high frequency trading) without ever describing any of those in the slightest detail. Its like a toilet paper roll printed off the investopedia website. A complete joke. So, this is the guy you compare with Black, Haug, possibly also Derman? LOL
How many pages do you think would be left if you tore out the MATLAB code and the cut-outs from his free blog? I'd rather have paid $4.95 for one month instead of overly paying for the book, and I recommend everyone else do so.
Regarding my idea, I did tests on some of my systems (see attachment), with several settings: last 5, 10, 20, 40, 80 trades. None worked. If I stop the system from trading when the last 10 trades (or any of the above settings) profit is negative, their overall net profit gets worse: from about a 50% of the original profit with a low number of trades (e.g.: 5, 10), to about 80% of the original profit with higher number of trades (simply because those requirement exclude fewer trades). Here's what I thought BEFORE conducting the tests, in a note I wrote: I thought my idea of a filter made sense, but since tests say it doesn't work, even if I don't completely understand why, I will not use it. Oh wait, I think I know why it doesn't work. The bad trades do not come in groups, and my requirement simply filters out random trades: but since the system obviously is a profitable one, it randomly filters out also a majority of good trades, thereby decreasing overall profit (the filters using a larger number of trades work better simply because they exclude fewer trades). Actually this is the proof that bad trades, at least as far as my systems go, are not grouped together, do not come in cycles, in trends, in predictable patterns. I will keep a record of the last ten trades just to know what risks I am taking with a system or another, but I will not use it as a filter. Oh wait, I was forgetting something important. If the profit of a system decreased by 20% and the number of trades decreased by 50% that would still mean the filter is good. So I'll check for that as well... Nothing. The trades decrease even less than profit. So it actually tends to get rid of the good ones even more than randomly (but not enough to implement a filter that only trades after good trades). Ok, so this filter is abandoned.
you could have spared yourself from all those tests by simply looking at your equity curve. You can run tests on serial correlation of your returns and then would clearly see whether your losses are serially correlated or not.
As my automated trading keeps on producing money, the same question keeps on popping up in my mind: how can it be possible? How can I, from home, with my very limited means, limited knowledge, limited capital, make money, pretty consistently, and without doing anything? Your commentary is contradictory I have worked at it for 10 years. During all these years, despite what others around me were saying, I thought I would ultimately make it and find a way to make money trading. Yet now that it's happening I still wonder, often - how can it be possible? What's the answer? Tell me. It's like a postman pulling in a guaranteed pension after a decade of work. Now sitting around and wondering how it's possible he has money coming in with no work.
Speaking from the perspective of a trader who has been successful for 7 years straight I am appalled at what you posted. You are taking a shotgun/spatter pattern approach to trading wasting 12 years of your life in the process. The answer to your question is you have blown out several times because your 33 systems do NOT work. You should, after all this time, been able to eak out a modest return enabling your account to grow sufficiently to cushion you from financial disaster. But you haven't done that! What you have done is have occassional runs of luck and then whoosh... it's all gone. When I make a trade I have a reasonable expectation of making a 4 to 1 profit or more. If I trade all day I know I will make 10 pts in the S&P per contract (because that's what I've done for 7 years). You have no clue what to expect because your so-called systems are flawed based on your results. I strongly disagree with using 33 systems when one would do nicely provided it was consistant and profitable. But again, you can't come up with one so you leap from one scatterbrained idea to the next. Your last idea is ridiculous and naive. BTW, what expectation do you have you will ever discover anything that works long term? Do you understand how markets work or are you merely looking for a minor/temporary discrepancy to pounce on for a few months? You need to FOCUS.
Thanks, very good remark. I don't feel like ten years of work is enough to be entitled to a pension for the rest of my life. And I was wondering if others felt the same, and felt this pension was going to last.
Regarding Ken (now on my ignore list): Yeah yeah... I'm doing everything wrong and he's doing everything right. I even believe that his one system is better than my 33 combined. His contribution? "You need to FOCUS". The real sense of his post? "Give up because you're not as good as me". Instead, I'll keep going at my own pace, and still make money.
Concerning the filter... On the other hand, it does make sense, if you are forward testing, to measure profit of the trades done in forward testing, and only invest money in systems that show a profit (it's the only way of knowing if you didn't over-optimise your systems), so that is how I ended up using that register of the last x trades I had created for another purpose (the aborted "last ten trades" filter).