For another, more extreme, set of implausible curves, look at how all the Monte Carlo curves for Goofiz Foliage’s Future ATM (which has a very high Realism Factor) lie far below the historical equity curve.
Yes, I noticed that too. The same is true for Pinnacle Trading (an option system). But intuitively I expect that this cannot be explained by the ‘destroyed risk management’ hypothesis, because destruction of risk management should make the simulated curve worse, not better than the historic curve. That this does’t happen in these two cases is odd too, because I would especially expect it to happen in option trading with their straddles and struddles and what they have.
I don’t know enough of option trading. Would a straddle / butterfly etc. most likely involve both a winning and a losing trade? In that case, simulation with replacement might select too many winning trades, i.e. more than would be possible under the risk management of the system. This would explain the excessively profitable winning curves. Such a simulation should also return sequences with too many losing curves, lying below the historic curve, but these would not be visible in the Pinnacle graph because of the scale adjustment to the largest profitable simulated curve. This adjustment shows the historic curve as a flat line on $0, and the losing curves should be between this one and $0, which makes them indistinguishable from the historic curve.
Thus my conjecture is that Pinnacle Trading also has a lot of simulated curves below the historic curve, but that these are invisible because they are covered by the historic curve.
Jules
Sorry Alan, I didn’t read you correctly. The system you mention is one where the historic curve is above the simulated curves (as we discussed earlier) while Pinnacle has the opposite pattern. But that opposite pattern is strange too, and I just gave a possible explanation for it.
But it is strange that the same hypothesis would explain opposite patterns. Perhaps the Pinnacle pattern has nothing to do with destroyed risk management, but is only a consequence of the graphic visualization (scale adjustment and curves that cover other curves).
But this raises another question: How many simulated curves are hidden in the graphs of other system? A grey area below the historic curve may perhaps as well reflect 1% of the simulations as 99% of the simulations of which most are indistinguishable.
Jules
Jules,
You are not the only one who compliments me about my nice curves. The ladies love them too. 
Tarek
I guess these ladies didn’t compare your curves with the excessively more attractive simulated curves? 
hahahahaha, nice joke Jules.
