Benchmark Stats for a Good trategy

[vickymatch]

Hi Guys,

I am new to this forum. I do some quant modeling on stocks markets and have build some strategies for nifty. However, I am not sure how to test whether a strategy is robust or not.

For example, I have one intra-day strategy, which over the course of last 10 years give following stats

Number of Days Traded : 26%
Win % : 53.2%
Reward/Risk ratio: 1.5

Sharpe Ratio (for 1 year period): 1.65
Max Drawdown: -8%
Max Drawdown Duration : 194 trading sessions

For this strategy, i have considered intra day brokerage of 3 bps, impact costs have been assumed to be 10 bps.

Could anyone please tell how does this start looks like from benchmarking point of view. Also, what are the ideal stats we should look at and if there are any other stats we need to find before making a decision.

Thanks,
Vikram

 

[vickymatch]

Re: Benchmark Stats for a Good Strategy

Please check the historical chart for the strategy as well. Eagerly waiting to hear from some of you.

 

[SavantGarde]

Do a Time Weighted Return.....that should give you the true picture of the strategy....!!!

While you are at it...check other parameters like Expectancy and win/Loss Ratio...!!!

SG

 

[piyush08]

Do a Time Weighted Return.....that should give you the true picture of the strategy....!!!

While you are at it...check other parameters like Expectancy and win/Loss Ratio...!!!

SG

What's time weighted return ??

 

[piyush08]

Re: Benchmark Stats for a Good Strategy

Hi Vikram,

I am a novice in this area and try out backtesting various strategies during free time. From what ever little I understand -

a. Brokerage used is ok. Impact cost wont be 10 bps. In Nifty futures impact cost is max 2-3bps. that is 1-1.5 points.

b. The profit curve looks solid to me... while before 2008 most of gains were made during bear moves or sharp corrections.. the performance after 09 has been consistent.. that matters most.. near term performance.. that means the strategy still works.

c. While I do my analysis i tend to look mostly at overall profit in terms of nifty points and the shape of the curve. The stats of 53% hit ratio and 1.5 win-loss seems good to me.. most strategies as it hardly have win %ages greater than 55-60%.