I am backtesting 3700 stock, all start from their listing in exchange until 2 weeks ago. With Reward:Risk ratio 2:1, i got 27129 Win Trade 43640 Lost trade, or 38.3346% win rate, if i trade every trade opportunity i will got enormous return that even excel cant handle the number. Is that real?? i mean how to make its more close to reality beside doing real time trading?? what kind of backtesting i need to do to make its close enough to reality, so i can get profitable result when doing real time trading
Did you factor in commissions and slippage to your backtest? Strategies that backtest well usually fail with live trading...for a variety of reasons.
%% Good list\hard to trade that many , but many of them are in an ETFs, for sure. FOR trading[benchmark] spy,SPXL, use all data, especially bear market like 2008+ bull market1999................................................................................................ [Reality=MOST full time Pros cant beat the spy benchmark over 10 years; some do.]
So... you believe you have a method that works. Next thing to do is trade it sim. If that appears to work out, then trade real money... but SMALL. You will learn things you hadn't considered. If trading small is successful for you, then scale up. If that works, go buy yourself a FERRARI!!
This happens to everyone all the time, until they fix tons of issues and bugs. There can be many reasons for this, including bad data, wrong assumptions, not accounting for liquidity, not including bankrupt companies, bugs, overfitting, etc etc. Also, "starting from their listing" means that you're testing 30+ years of stock history? Did you account for stock splits, dividends, mergers, and other corporate actions? Did you include stocks that no longer exist because companies went bankrupt or got acquired? And in live trading you cannot buy $100 million worth of some stock. Sometimes you can buy several shares at the price you want, or without causing the price to go higher and not be able to buy more shares, so you have to account for liquidity as well. There are so many reasons for such numbers that your question(s) cannot be answered until you spend months or years finding and addressing every issue, including test trading on paper and live.
No i did not factor in commission and slippage, but its annual return are 10 times and its in daily timeframe so i think slippage and commissions not effect too much yes i also think its impossible to trade every opportunity, but even with 510 trade every year it still got some huge annual return yes i will start small, i just wonder if there are some aspect i might not count when doing my backtest , thats why i ask here. i got the data from yahoo finance, wonder if its good enough for EOD data, liquidity is not a problem, if i limit my trade with stock that have volume $100M/day my win rate increased, so i think its fine. if the average trading day is 7 days is that really matter the stock splits and dividens?
No. You cannot get fills as a retail trader. Imagine you try to fill multiple basket orders. What do you do when you are not getting fills? Go market? Always MKT? Back testing does not really reflect MKT orders unless you are doing something closer to TICK data. Right now you are playing in a fantasy land in excel. Try to get it to work with 100 stocks in FORWARD testing. IMO backdating using large scale intervals (anything over 2 minutes) is not a good reflection of realty. PS: about once a month someone posts this "super shotgun" un implementable approach.
Backtesting does not include the "psychological anxiety/money pressure" (and how it alters your behavior/discipline/hope) you'll experience when trading for real. Trade small at first and work your way up. No need to be in a hurry to get rich. It will all still be there when you are ready.
If you are using Yahoo EOD data, and your backtest references LOW as an entry point, then your backtest will not be accurate. Yahoo daily lows for equities are often not accurate. Try running your backtest on ETFs that represent market indexes, and compare those to your basket of equities.
Most likely you are peeking into the future somewhere without noticing. Also, there's no good reason NOT to account for commissions in a backtest. Slippage is more complicated.