Hmm. There is no right answer obviously, but I think profit per trade value might be a more important metric in some sense. Your t-stat, which is a product of Sharpe and the square root of number of trades is important to determine if the alpha is real but for strategy selection you might care more about trading less per dollar earned.
Van, interesting comments that I will have to consider. I'm confident that the stops are not a factor. Obviously, I tried to avoid curve-fitting, but this is always a concern. If I understand correctly, you think I have too many losing trades. Wouldn't fewer losing trades indicate curve fitting? You stated, "If these presented results were truly robust the amount of trades to produce these results in given time period would need to be X times higher". Why would shorter term trades make it more robust? Can you explain a little more about why you want to have longer draw-downs?
Can you explain a little more about why you want to have longer draw-downs? it comes from experience I know just one robust strategy,but it is robust in small time frame and larger time frame.So if i get more time spend in a losing period and you don't imho there is a chance it is curve fitted. Why would shorter term trades make it more robust? Wouldn't fewer losing trades indicate curve fitting? Law of large numbers,but this is in relation to profitable trades and how profitable they are.The is a relationship between these numbers(positive trades/losing trades) to the amount profit that can be extracted safely that even robust strategy does not break above. Again it comes back to no losing months.Not something i would like to see. I only give you my point of view,don't want to sound negative.I am sure there is many ways to slice this which i don't know about.I am by no means math wizard Profit comes from trends if system is trend following or next bar signals or something,doesn't matter what is the signal to open the trade.It is about the distance to make that money,number of ticks etc and markets spend about 55% of time in consolidation.If you have no losing months in 5 years period and that many losing trades ,yes that many losing trades it is difficult for me to accept this.Regardless what risk/reward you apply it is still close to 55% if it was same target/stop loss.Markets got to be so efficient that even time stops longer term give similar ratio(winners/losers) I was thinking to myself you use time stops to get to this level as you need to take full advantage of trend days. My best wishes,you making money that's good,don't increase size.As i wrote earlier i would want longer period spend in a draw down,at least 5 months total.There is big difference coming from a draw down 5 times than not having any on monthly basis. Best regards
So in a very real sense, the best strategy for you is not derived mathematically, its the one you can sleep most soundly with and wake up tomorrow and trade most diligently. Go for it I say.
It would be foolish to trade any of the four. A backtested 'strategy' losing hundreds of trades over 5 years with only a $2,300. draw down does not exist. You might want to put your Etch-A-Sketch away and try another platform.
Your post becomes more ignorant as it progresses. Your first sentence may very well be valid. Your second sentence is nonsense - any high school kid could create backtested strategies with many small losing trades and small draw downs over a 5 year period. Not sure if your last sentence is a feeble attempt at humor or just you trying to be offensive.
Your answer is appreciated. However, I do not quite understand what you are telling me. Do you mind expand on why this is so? Best regards,
All 4 systems look curve fit. Reason is 'profit to drawdown' ratio is way too high. $45000 a year profits, but drawdown just $2,400 Thats is almost 20:1 ratio. Even 3:1 on a yearly basis is pretty much 'holy grail' for a low frequency trading system.
Why were the strategies tested on 5 years of data and not 10? There is no way I would trade a system whose test period did not include a bear market phase.