Strangling 96 out of 100 and Online Spreadsheet Woes.

I usually use online spreadsheets to run my data when I am working on a new strategy and trying to see the future possibilities based on past performance. This time I have had to use a PC based system as there is just too much processing involved. The online services just cannot handle it. As a result I don't have my spreadsheets with me today, so my numbers are going to be rough generalizations. I also have some more adjusting that I need to do to my entry and exit criteria to get the accuracy that I would like.

Actually, on the entry and exit. This particular strategy does not conform to the typical style of back testing that you might see when trying to buy or sell a stock based on indicators. In those cases the indicators need to be overlaid and entry exit prices determined by using some complicated formula and a charting service that has some level of backtesting capabilities. While there area  number they are not cheap and the results are not always exactly what would have actually been traded... for a number of reasons.

Due to that my backtesting has typically been done by hand so that I would see what had gone on and applied my rules accordingly, I got to see exactly where the plan worked and didn't rather than just getting a number at the end of the test. This was necessarily very time consuming and, if a plan did not pan out, it was often frustrating.

This is different. I am not concerned with entry and exit points. I am not overly concerned with the price movements that the underlying security exhibits. I really don't care if it goes up or down or nowhere after a trade entry, only if it goes past my set move allowance from trade initiation to completion. The test is a win or loss determination for each trade based on the particular opening price and the particular price at option expiry (could be Friday's opening or closing depending upon the style of option traded). Each week there are one or two trades made, the allowance is applied for the time period and the result is spit out and tabulated as a total number over whatever time period I am testing for.

Also, unlike many backtesting plans my window tested can go back as far as I have data available. I can easily get free data for about 18 years, run that through my spreadsheet and produce a win rate in moments... just not online. If I want to apply a normal loss allowance I can do that, or I can use the actual closing value of the stock at the expiry of the trade to determine what the un-stopped loss could have been, sort of a worst case let it all ride scenario. I expect to do both methods to see what would have happened.

At this point I am seeing a greater than 96% win rate over all traded timeframes (1, 2 and 3 days). Well above my estimated 80% requirement. This is without any adjustments, just applying an allowable move percentage and making a sweeping assumption that I would have to use that allowance each and every time. In more volatile markets I could use a much wider allowance so the percentage I started with is really a minimum. I would rather not try to guess where the allowance could have been greater so I will just consider this as a worst case scenario. I have always liked to skew the results against me in order to keep the expectations more realistic.

Given such a high win rate and using a loss allowance of 4 times my option value (I may change that for a number of reasons but it is a good place to start) and applying this over the past since 1993 produces some very large numbers. I tend to think that the numbers are large enough that they can be discounted as, more or less, unworkable in a single security option. I expect that, at some point splitting the account into multiple security options would be necessary to not be trading all of the available open interest... that and switching to or adding the larger and more expensive index futures. This would also serve to mitigate any potential single loss. Of course there is always the removing of cash from the plan to use for personal uses along the way. I am not letting these numbers get me too excited as this is a longer term income strategy and I have yet to see what premiums can be sold in various option chains after applying reasonable move allowances to their respective historical data. I also have not accounted for income taxes in any of my math yet.

Even though past results do not guarantee future results I can at least be certain that if a stock or index fund exhibited 96 winning trades for each 100 trades and held this average for 18 years using a single move allowance value I can be pretty sure that some similar result may be reasonably expected to continue, particularly if I can apply varying move allowances to suit the volatility of the market in real time.

Jeff.