Question about backtesting and asset selection

[Cjacket23]

Hello all,

I have a trading strategy that I have found works really well on a few timeframes. It is basically a long-only trend-following intraday strategy, though it can be modified to be long-short. This strategy works really well for some stocks/futures/crypto/etc, but it does not work very well or is highly unsuccessful for others. With all that being said, my main question is how does one find out what the securities that it works well on have in common? What is the best way to approach this? I’ve been brainstorming a lot and any help would be appreciated.

As a note: I do most of my initial backtesting on TradingView because it is relatively quick, and then move the strategy over to QuantConnect for more robust testing. I also have access to Bloomberg though I am very much a novice on that platform.

Thanks in advance for the help.

 

[lindq]

Hello all,

I have a trading strategy that I have found works really well on a few timeframes. It is basically a long-only trend-following intraday strategy, though it can be modified to be long-short. This strategy works really well for some stocks/futures/crypto/etc, but it does not work very well or is highly unsuccessful for others. With all that being said, my main question is how does one find out what the securities that it works well on have in common? What is the best way to approach this? I’ve been brainstorming a lot and any help would be appreciated.

As a note: I do most of my initial backtesting on TradingView because it is relatively quick, and then move the strategy over to QuantConnect for more robust testing. I also have access to Bloomberg though I am very much a novice on that platform.

Thanks in advance for the help.

You may want to start by building lists created using fundamentals from a stock scanner. Then run them through your system.

As one example, when I used to trade equities from the long side, I found that a strong EPS helped to enhance returns.

Try growth, beta, volume, dividends, etc. etc.

 

[globalarbtrader]

Hello all,

I have a trading strategy that I have found works really well on a few timeframes.

This result is almost certainly overfitting as there is unlikely to be any statistically significant difference in the returns from your various timeframes. Average across as many timeframes as you can [allowing for costs] regardless of whether the backtest 'works' or not.

It is basically a long-only trend-following intraday strategy, though it can be modified to be long-short. This strategy works really well for some stocks/futures/crypto/etc, but it does not work very well or is highly unsuccessful for others. With all that being said, my main question is how does one find out what the securities that it works well on have in common? What is the best way to approach this? I’ve been brainstorming a lot and any help would be appreciated.

As a note: I do most of my initial backtesting on TradingView because it is relatively quick, and then move the strategy over to QuantConnect for more robust testing. I also have access to Bloomberg though I am very much a novice on that platform.

Thanks in advance for the help.

This result is almost certainly overfitting as there is unlikely to be any statistically significant difference in the returns from your various systems. Trade as many instruments as you can regardless of whether the backtest 'works' or not.

GAT