In the above figures... notice how if you only run 30 trials, the region of interest takes up a huge chunk of your total number of trials... I would say using a 1000 monte carlos is a better choice
One thing about markets that needs to be carefully considered while trading systems is.........markets change over time and thus systems based on some anomaly also go out of favor. Today vrs 10 years ago, more and more trading is being done by algorithm bases strategies. Upto 70% of today's trading is algo based vrs 20-30% a decade ago. This also shows how much markets have changed and big firms have created a new way i.e. HFTs to trade. In many a ways for a retail desktop based individual trader, mere knowing the market behavior is not enough. Markets are becoming more and more efficient and especially for common retail traders.
One of the biggest issues with purchasing a trading system (aside from reliability of performance) is the user's lack of faith in the system during drawdowns. If you can't see the code and see why the code is doing what it is doing and see exactly why the drawdown is happening... you will bail on the system and turn it off at the worst time. Then when price rebounds you will freak out and turn it back on (selling at the lows and buying too high). Loss of faith is one of the main reasons traders fail with black box systems.
This is just a guess: 5 seconds ATIT: 30 trading days 5 minutes ATIT: 300 trading days 5 hours ATIT: 3000 trading days (6-7 years) 5 days ATIT: 20 trading years It's pretty obvious from this why high frequency trading has become so popular. Who wants to die before their system is fully forward tested ?
No, like I mentioned...just a rough guess. I need to develop a formula from ATIT or trading frequency...or a combination of the two perhaps. Which one is more relevant ? With the ATIT approach, one would require the same forward testing time doing 10 trades with an ATIT of 5 minutes vs. 20 trades with the same ATIT.
Oops, I didn't see the "guess" part. I would think the relevancy of ATIT vs frequency would depend on the specific strategy. For some, they are basically the same thing.
They are mathmatically well correlated only for SAR systems (stop and reverse)......systems that are always in a position during the trading day. The ATIT would then be inversely correlated with the trading frequency (TF). The higher the TF, the lower the ATIT.
Position Change Divided By Time. In that manner, the count or frequency is not affected by scale-in, scale-out strategies.