After years of studying ES, I have a trading strategy that I backtested (manually) for 200 days and then forward tested for about 75 days now. Not every day signals a trade, the total is about 170 back test and 51 forward tested. Both segments perform at 90-92% win rate with a positive expectancy of 1.30 ES points per trade. Very simple system, only 2 parameters to be met, so no curve fitting. I am not offering the setup, for sale or otherwise, but would like feedback as far as the testing is concerned. Sufficient?
What is the minimum I would need to trade one ES contract with your strategy? And what is the Max drawdown I should expect? I won't even ask you how much it makes. If it makes anything i will be happy.
It may be possible to tell this from information you have, but it isn't possible to tell it from the information given in your original post, above (and I'd respectfully advise you to be wary of anyone trying to). We'd need to know further information, to work out the potential statistical significance of your results. Systems with win rates over 90% are particularly high risk in the sense that they can have long periods of apparent profit before the ship dramatically goes down - and there are reasons for that. They tend to need a lot of results, to achieve statistical signifcance (in general, the closer the win rate is to 50/50, the fewer the number of analyzed trades will constitute statistical significance.) For a start, it would be helpful to know the sizes of the average and biggest losing trades, the MAE and so on.
90-92% win rate sound good, but..: What is the MDD% of the tests? What is the P/L% of the tests? And: how much is the initial capital requirement (if this matters)? And: is this a single item trading system or basket/porfolio trading system? (I guess single item as it is for trading only ES it seems) And: is it a long only system, or can it also go short? Ie. can it, or does it, also do reversing the position?
I have seen systems that have done well for 2 years, but when back tested out for ten years, fail horrible. In my own back testing, I need sample size of 3,000 to really know what I can expect.
Just Apply a variant of Kelly sizing. Adjust it daily. If it fails you stop to allocate. By Kelly Sizing I mean, fractional sizing adjusted to performance. You could get more backward data. But it's good you have forward tested.
Thank you all. The setup goes long or short or no entry based on my long observations of how ES behaves. 1.75 target /3.25 stop 2 losers in a row max. 20+ winners in a row. I plan to ramp up contracts trade based on the setup's returns. When it gains enough profit to add another contract I will and so on. My question is do I add capital or let it grow organically? Assuming of course that this is an accurate indication of what the setup will do. Additionally, in order to have such long winning runs, the setup must be significantly better than 50/50 or I happen to be in a run of historic odds.
That actually answers most of what I wanted to know, in my post above (and answers it better than I thought it might, too). I've never had a trading method with anything like a 90% win-rate, myself, and was expecting the losses to be enormously bigger than the wins, but it seems that they're under twice the size of the wins, which is clearly very encouraging. Sounds a good aproach, to me. If those figures hold good as it progresses, it's going to grow quickly enough anyway, without adding capital? Well, that rather "begs the question", I think? Doubtless there's a question of whether it requires a market behaving in a particular way, that might turn out not to be such a typical way, and this kind of consideration? Have you backtested it on different samples from different years, for example? But it would be churlish not to comment that it does sound very encouraging indeed.