Defining a baseline test is necessary for all test runs so that we have something to measure the test’s parameters against.
The baseline test needs to be comprised of something common amongst all future test runs. Since future tests will primarily manipulate the exit triggers, the only truly common element are length of trade and the entry delta. Every test will test the Front, Second and Third Month. And every test will test roughly the same variety of entry deltas. The permutation of these variables creates 36 trades.
For the baseline we choose to hold these trades to expiration. This makes sense as a first attempt as any adjustment to the trade should probably be better (in either risk or reward) than holding to expiration if it is to be worthwhile. I would assume that most test runs with substantially improve the ‘risk’ component when compared to the base line, while possibly improving the reward. When holding to expiration, the risk is that we loose 100% of the margin associated with the trade.
First we look at the the Expected Return vs the Delta of each trade
Note that there may be a parabolic relationship with the data, and a local maximum appears to be around a delta 20 to 25 trade. The Delta 30 trades and above have negative returns.
Graphing the sum of the winning trades against the sum of the losing trades (in other words the Profit Factor) we can see a what might be another parabolic trend. There appears to be diminishing returns with trades that aim for higher profits.
Another way to look at the data is to look at the Sum of Profits Vs the Sum of Losses over Delta splitting out the trades by their duration. This view shows that there may be some tailing off for profits in any linear relationship to the delta when the delta’s get higher than 30; but the losses keep declining in almost a linear fashion. This again shows the diminishing returns.
I also like to look at the Number of trades VS the Profit Factor to ensure that trades with a high profit factor are not due to a very low sample size.
We can see that the Front Month trades found the highest number of trades. This is probably due to two factors. First, my tests aggressively look for more trades if targets are met early. So if a test exits within a few days, it can find another trade for the same test cycle. This would not be the case for this baseline test run as all trades are held to expiration. The second explination may have to do with the dataset itself, or the available trades. Further back in time, there appears to be fewer strikes available in the RUT. Few strikes, especially for back months would make it harder to find condors with a specific delta and width. The good news is that all scenarios found I high number of trades. This graph will become more important in the future when applying different limitations to the test to ensure a sufficient number of trades occur.
It is interesting at this point to highlight that trade with the excessive PF. A test I ran to determine whether there is a Daily bias found an annomoly with low Delta (5-7) and very low duration (8-15 days from expiration) trades. I added a very specific case to test this scenario, and it seems to perform well with an Expected Return of 5.45%, and a Win Ratio of 94%.
Finally, I want to look at the Win Ratio. Some traders may not find it useful, but I do when it is used in conjunction with other metrics. This chart shows us what we should expect: there is a nearly linear relationship between the delta’s we choose and the chance of the trade being profitable.
What is very interesting to note is that the rule of thumb for delta’s does not seem to be applicable. The rule of thumb uses the delta as an estimate to the chance that the trade will be profitable if held to expiration. If our Iron Condor has short strikes with a delta of 25, each short strike has a 25% chance of not being profitable at expiration. Combined, the strikes have 50% chance of being profitable. However, the test show that historically with the RUT, these trades would have a 70-80% chance of being profitable.