I develop algorithmic strategies for a living, and my first reaction to reading your post was skepticism. I'm skeptical for two reasons. (1) because your methods are so unconventional in an industry where convention rules, and (2) because of the time frame of your success, which happened to be one of the more impressive market recoveries in history.
I can't tell you how many people I've worked with who fail to isolate the source of their pnl (myself included at times). This is key. It's important to benchmark your strategy against other stupid ones that you know don't have edge. When someone shows me strategies that worked in 2009 and 2010, I immediately make them prove their strategy was not the equivalent of being long equities.
Doing this will truly help isolate whether or not luck is involved. When you say that the number and size of your trades justifies the strategy's validity, that's just wrong. You could do 1000 trades in a day: buy 10 RUT futures at the beginning of the day, sell 10 at the end, and just scratch 1 lots for the other 998 trades. In a bull market like 09-10, that would have made 400k, and would have nothing to do with Machine Learning or its applications to HFT.
I make all traders benchmark their work against a series of other strategies that I know have no edge, even though they, at times, can appear to have edge.
Now, I'm not saying you didn't have legitimate edge, but you do your readers a disservice by omitting relevant stats and discussions like that.
> I'm skeptical for two reasons. (1) because your methods are so unconventional in an industry where convention rules, and (2) because of the time frame of your success,
I was also in this business, and there's nothing unconventional about his methods. It would, in fact, closely describe the methods of more than one shop I'm familiar with. (Except they WERE able to overcome the declines). And the 3-6 month indicator lifetime looks eerily familiar.
And these places are anything but "convention rules" - it's "creativity rules, before our competitors get creative enough".
> When someone shows me strategies that worked in 2009 and 2010, I immediately make them prove their strategy was not the equivalent of being long equities.
Assuming the OP is telling the truth, there is no equivalent "long equities" strategy that would make 1500% profit over 6 months (%3000 annualized), with a max drawdown of 20% ($2000 on $10000 - but his max drawdown was probably closer to 5% than to 20%). You are welcome to demonstrate that there is.
Sounds to me like you are doing low frequency strategies; it's a completely different ballgame than HFT. He's done 400,000 trades, half of them long, half of them short. It might have been luck, and he might have been riding something underlying the equities, but this is NOT equivalent to being long equities. He might have found a way to get non-linear leverage (rather than prediction). But that's also worth a lot of money in the right hands.
> buy 10 RUT futures at the beginning of the day, sell 10 at the end, and just scratch 1 lots for the other 998 trades. In a bull market like 09-10, that would have made 400k, and would have nothing to do with Machine Learning or its applications to HFT.
That may be (I wasn't trading in 2009-2010, and don't remember the movements or the required margins), but that would have had much higher volatility (and days with much more than $2000 loss) than the OP had. (Assuming, of course, he is telling the truth)
Given a max loss of 2k, we already know the Sharpe Ratio was pretty good.
2009-10 was more than just a huge rally, it was also a period where vol and skew were massively mispriced. I know this is high frequency, but like I alluded to, you need to make sure that what you're doing isn't replicating the pnl profile of low frequency strategies.
So, how did you perform relative to vol sellers? From the market bottom to the end of 2010, the max daily loss for a vol seller was about 3x average daily pnl, and >80% winners. So your returns do sound better, but not incredibly.
But, even if you failed to perform as well as vol selling did over the same period, that doesn't negate the strategy's validity. If returns were not correlated, then it's safe to say that you weren't just inadvertently shorting vol.
So, start there, work out a regression comparing your daily returns to someone selling vol. Do the same with moving average strategies. Mocking up a simple market making back test versus an ES beta is hard, but that too would be a something to test against. I don't expect you to do any of this, and I'm not going to bother to either. I'm just saying that a complete discussion of this subject would include that information.