Backtesting Frustration (Part 7)

Today I resume my series on backtesting frustrations by talking about the frustration of “flips.”

I mentioned this in Part 1 with regard to recalculating results with different TF values. A trade that is down 25.5% with $0.26/contract may only be down 24.5% at TF $0.16/contract thereby evading the trigger of SL and ending up profitable. Simply recalculating the results, I thought, would overlook these flips (from loser to winner) altogether.

In taking a closer look at the put credit spread results I see that 188 out of 1,093 trades originally hitting the -25% SL show a loss smaller than -25% with $40/trade added.

I can now proceed in a few different ways: 1. Redo the 188 trades with TF $0.16 ($0.06)/contract; 2. Assume the SL was not hit and use either 7 DTE or Exp PnL values instead; 3. Assume these 188 trades closed for zero gain/loss.

Without doubt, the first option would be most accurate and also the most time-consuming.

I therefore started working with option #2 until I realized a major problem with the assumption. A trade that evades SL on one day may still hit SL on a subsequent day. Being so close to SL, another down day would probably trigger the unprofitable exit. With the market showing recent bearishness this seems quite feasible.

Furthermore, if the subsequent down day is big then the loss might end up being much greater than initially recorded.

Option #3 was intended to be a more conservative form of #2. In [falsely] thinking most of these trades avoiding SL would flip, it occurred to me that the market may not recover enough for full profit to be realized. To be conservative I could just call those zeros. Even a zero is much better than -25% for overall performance.

Hopefully I have made it clear that I can’t assume enough to go with option #2 or #3.

I then considered a fourth option: look at the chart. If the market is bottoming on the day the SL is hit then I can proceed per option #2 or #3 depending on where the market is at 7 DTE or Exp.

Still though, if the “Furthermore” (look up four paragraphs) happens then I may be looking at a much larger loss; just leaving it as before plus $40 would be inaccurate. This would be an argument for redoing all 188 trades. While it may not seem like lower TFs could translate to larger losses, there is a lack of granularity when testing on an EOD basis.

In stepping back and considering the wider perspective, it seems like a chance occurrence whether a flip or larger loss will occur. Unfortunately, I feel I must retest in order to have any possibility of knowing for sure.