1 (edited by kalimbo12 2014-10-14 16:47:37)

Topic: crazy SQN

How this could be?

http://s2.postimg.org/xglwnovid/mad_sqn.jpg

Re: crazy SQN

SQN takes into account the ProfitLoss of the closed trades. It doesn't reflect the Drawdown. Since your strategy backtest shows only winning trades the SQN goes to the sky. There is no limitation in the formula for a maximum value.

SystemQualityNumber = Sqrt(CountOfTrades)*AverageProfitLoss/ProfitLossStdDev;

I assume that the shown value is a result of high AveragProfitLoss and a very low Standard Deviation.

I agree it looks odd. What I can do is to limit the SQN to, lets say 100.

Re: crazy SQN

Its hard to say what is better. I think let it be as it is if its a true value.

Re: crazy SQN

there appears to be an area of large drawdown....... do you not have a stop loss?

Without a stop loss you can not depend on a System Quality Number. Stop Loss is a must.

Re: crazy SQN

Very important: limit drawdown by imposing a constrain in the acceptance criteria.
Also: do never believe a strategy with only wins. Your strategy is so obviously over-fitted that it will perform very poorly out-of-sample.
Cheers

Re: crazy SQN

nquental wrote:

Your strategy is so obviously over-fitted that it will perform very poorly out-of-sample.

Hello Nuno,
You are perfectly right. This strategy is obviously over-fitted. However, my concerns are not about the strategy but about the SQN formula.

Kalimbo, can you send me the strategy to inspect closer from where this SQN value is coming?