Matthew,
I understand that the trade drawdown issue will take a while to fix.
You have not yet addressed the other drawdown issue #1 in my original post. My total system drawdown still shows 73.51% for a short-lived and false drawdown on July 21st to July 24th. I notice the same issue is starting to affect other systems. For example, look at Broadsword Macro; it now shows a max drawdown that does not reflect the actual trades.
Your most recent Monte Carlo run for the EXLENCE system has included this phantom drawdown and now shows a significantly higher risk of 20% and 50% account loss. Previously the risk of 20% account loss was no higher than 20% and the risk of 50% account loss was 0%.
Chris
With regards Macro the individual trade is no correct and Matthew has assured me the drawdown numbers will be restored to their previous state. You will also need to check the intraday data on the equity chart is corrected. In the past I think these things are normally on overnight runs. This was apparently all because the price data format of JPY was changed.
I understand that the third class of ‘n/a’ cases is hard to solve - if it is possible at all. With respect to the computation of APD I wonder if it is not better to have a temporary solution of the problem in this way:
1. For the trades with known DD, compute the average DD per quantity (DD $ / abs(DD quant)).
2. For each trade with unknown DD, if the n/a is probably caused by the granularity of the C2 data, then estimate the DD as the (Num opened) * (the average DD per quantity that was computed in step 1).
3. In the computation of APD, use the estimated DD from step 2 if the true DD is unknown.
In the Magic Moment System this would change the APD from -0.15 to +0.23, which is a more reasonable estimate since the system was profitable.
Note that the limited granularity implies that the max DD is underestimated for all trades, not only the trades with n/a. For example, in the Magic Moment System there is a trade that closed with a loss of $500, but the reported DD for that trade is only -$312.50.
Don’t the bars include OHLC prices? Just find the lowest L (or highest H for shorts) and you can calculate the exact maximum drawdown for the trade, regardless of the granularity. The only problem would be when a trade is closed or opened in the same bar where the max DD occurred. In that case, it would be hard to tell if the order executed before or after the L (or H).
