Performance data in System Description for a Strategy

C2
1

I’m a quant developer who will soon be publishing my algorithmic strategy on C2. I would like my potential subscribers to see my performance metrics, which are derived from backtesting. My backtesting generates the usual metrics that a potential trader would be interested in, such as expectancy, win rate, average ROI/trade, profit factor, t-score/SQN, etc.

However, I notice in the Trade Leader FAQ that C2 does not approve of performance numbers being posted in the “System Description”. I don’t understand this, and would appreciate some clarity. I realize that C2 is focused on forward-testing (due to survivorship bias, overfitting in backtested data, falsifying results, etc.) but performance metrics derived from backtesting are important for subscribers to see. How else could they compare systems objectively to make investment decisions, when it could take a year or so to build up a large enough equity curve with the model account to attract subscribers?

Some subscribers may want to see more detailed backtest results. I don’t have a problem producing those (I have them on hand anyway), and could of course email/broadcast those to subscribers. Is this a violation of C2 policy?

I appreciate any thoughts on this.

2

Michael -

If you have backtests, and you want to post them in your strategy description, it is okay to do so, as long as you make it very clear that these are only backtests, that they are no guarantee of future performance, and that this data is not based on a C2 track record and is not verified in any way by C2.

Our general admonition against posting backtest results was caused by people posting ridiculous backtests which suggested that they implied forward results.

Finally, despite what I write above, if any material seems to violate regulations with which C2 needs to comply, I’ll bring it to your attention.

But if you use common sense, I’m sure you’ll be fine.

3

Thanks for clarifying Matthew.

Posting backtest results in the system description would not be practical (and would require approval each day, since backtests are run nightly). It’s only the derived performance metrics that would be of interest to most subscribers. It would seem that subscriber broadcast messages would be a more appropriate way to distribute those, and for anyone wanting detailed backtest results, they could be provided on demand. The usual disclaimer regarding forward results would of course be included in any such communication.