RSI Lag Trading

I've mentioned the concept of lag in previous posts and an article in the latest Futures and Options Trader e-magazine stimulated my otherwise semi-comatose brain to explore a relatively simple, but practical systems application. Subscriptions to the magazine are free and I always look forward to the unique market perspectives frequently found within. Go here to signup and download the current (December) issue. You don't have to be a futures or options trader to benefit from the frequent trading nuggets of wisdom that are profiled.
The article in question was by Steve Lentz and Jim Graham of OptionVue and they employed the RSI lag to time bear call spreads or bear put spreads. They don't mention the period of the RSI they used for their model, but I optimized the thing in TS and found RSI14 on the short side and RSI11 on the long side work best for the IWM. The system as they define it has a good track record, although the drawdowns are a bit onerous, and the average holding time for winners is 37 days, which, if you've followed me for a while, is about 35 days more than I'm comfortable with. So I reversed the signals and morphed the concept into a short term (2-5 day) equity trading system (no options). You, gentle readers, are welcome to pursue the original intent of the article, but the above system trades more frequently and manages risk a little better, IMHO.

TS 2000i code is shown below and is optimized for the IWM. I added a fixed bar exit which is variable for the longs and the shorts to accommodate the added market momentum of the down moves. That is, downside moves are typically of greater range than upside moves so what takes 5 days to go up may only take 2 days to go down. The optimization test for this system seems to support that notion. FYI: I use the "Then Begin" and "End" conditions to assure that the system doesn't fire until the initial Lower Low and Higher High non-conformation with the RSI has transpired.

Blogger has refused to let me post the code in open format as it won't recognize the > symbol without screwing up the entire code so no cut and paste today.