S&P Options Directional Trading System Update

P/L tracking chart....


UPDATE ON THIS (original post below this):
I added the "Expiration Friday" tweak into my backtest (used data from March 1992 through July 2006). Out of the total of 85 short signals generated by the system, 38 of the signals, using a maximum 5 day holding period, would have occurred over an expiration Friday.

Of these 38 signals, only 11 would have been unprofitable after the 5 day holding period.

So, I implemented the additional tweak of holding longer than 5 days if expiration Friday was one of the days within the 5 day window to test whether it was beneficial to lengthen the holding period a few days if this was the case.

By doing the following tweak if the short signal trades over an expiration all 38 out of 38 signals that held over an expiration Friday would have been profitable:

1) If the trade is profitable on day 5, close out the trade for a profit.

2) If the trade is not profitable on day 5, continue to hold it another day.

3) If the trade is profitable at the close on any succesive day after day 5, close the trade at market close.

By doing the above, all trades would have been profitable with the longest holding period being 10 days (5 days more than the original 5 day holding period)

So then I thought, what if I just change the basic system to "10 day maximum holding period" for all trades. Doing this drastically reduces the overall average P/L significantly. In my mind, this proves that at some level, some options week markup/markdown shenanigans do occur, and it is worth taking into consideration when developing short-term trading systems. Let the Efficient Market Hypothesis cultists put that in their pipe and smoke it...

I guess I'll see whether the 2 outstanding signals from last week will fall into the 38 for 38 profitable bucket if I extend the holding period out 5 days beyond the original 5 day maximum holding period making it 40 for 40, or if the very low volatility that is typical of August in the markets will give us our first losers. (Side note: I checked and of these 38 signals analyzed in greater detail, only 2 times did they fall in the month of August--years 1995 and 2000--both much more volatile times than what we've experienced the past couple of summers--OK that's enough digging into the details--these signals are either going to work or they aren't at this point as I don't want to unjustifieably "fit" the system to any more parameters)

ORIGINAL POST
Above (click to see the chart bigger) shows how the 2 short signals that were generated last week are doing. Again, the system trades the S&P500 via options contracts and currently has it's max holding period set at 5 days. But I'm thinking about "tweaking" the system to hold for 2-3 extra days if the 5 days overlap an options expiration Friday. I haven't added this tweak yet into my backtest results to see how it affects the long run average.

As the system currently stands, backtest results short signals are correct over 60% of the time, and long signals are correct over 70% of the time. (I'll post more backtest stats on these signals this coming weekend.)

As the chart above shows not a lot going on--basically flattish to slightly negative. See how it plays out over the next 2-3 trading sessions.