I think a way that you can think about it is from a risk perspective. One rule passed down among traders is to not risk more than 2% you account size in any single trade. For a $50k account that would be $1k. Now let’s say that your system gives you 2 times your risk (2R) with a 60% win rate, and let’s say that in average you make 10 trades a month… you can run those numbers and see if they make sense for you.
There’s usually an inverse correlation between win-rate and accuracy. If you increase your risk you can be right more often but your rewards are smaller. Alternatively you can be more “selective” with tighter stops and thus lower risk and is likely that you’ll be wrong more often but that the winner will be bigger…
I believe that you can have the skills to build a system that can detect patterns from historical data, but notice how, for example, from the previous calculations the problem of trading is less of a “technical” problem and more of a market, risk management, money management, statistical kind of problem.
Then there’s the emotional part of, you still have the power to stop your system… if it has had 4 or 6 losses in a row… are you still keeping it up? Loosing is inevitable in trading, but how much loosing can you handle? Have the market conditions changed? How do you measure that?
> but notice how, for example, from the previous calculations the problem of trading is less of a “technical” problem and more of a market, risk management, money management, statistical kind of problem.
I agree. It just happened I built such system already which models strategy over significant period of time, and accounts statistical/risks indicators, transaction price, sleepage, etc. I built it for forex eur/usd pair (probably hardest market to beat), while results were positive, I decided that they are not strong enough, and I keep wondering since then if predicting patterns in commodity futures is easier, and I should try to relaunch my effort..
I’m probably biased but I prefer futures, even for forex, just because is only one market vs all the extra markets a big spreads in forex, 6E for euro dollar or M6E for even smaller positions/capital/risk.
There also the element that there are a lot of things you don’t learn until you run a system live
I run it on test account, if I subtract commissions and potential sleapage, it looked like gambling with odds slightly in my favor, in short, my deep learning pattern recognitions were not strong enough on that market.
There’s usually an inverse correlation between win-rate and accuracy. If you increase your risk you can be right more often but your rewards are smaller. Alternatively you can be more “selective” with tighter stops and thus lower risk and is likely that you’ll be wrong more often but that the winner will be bigger…
I believe that you can have the skills to build a system that can detect patterns from historical data, but notice how, for example, from the previous calculations the problem of trading is less of a “technical” problem and more of a market, risk management, money management, statistical kind of problem.
Then there’s the emotional part of, you still have the power to stop your system… if it has had 4 or 6 losses in a row… are you still keeping it up? Loosing is inevitable in trading, but how much loosing can you handle? Have the market conditions changed? How do you measure that?
Good luck!