I've been playing with your standard deviation formula. I like what you said about the sell signal: "If it turns down, a sell should be generated by tsf 21< 89. If not, it rides that balloon up."
Here it is again, for anyone else reading:
Enter long: 100*std(c,21,1)/(hhv(c,144)-llv(c,144)) < 5
Exit long: tsf(c,21) < tsf(c,89)
Some systems I've come up with -- or should I say all systems I've come up with? -- don't get me out if they don't go up. The sell signals work fine if the stock does go up, but otherwise, they don't kick in. This gets back to what David Coburn was talking about -- making sure you don't hold on to losers. Your system is a good one because it has a dual-purpose sell signal in a way, one that gets you out of the trade if it doesn't go up. I guess that must be an obvious element in good systems, but I hadn't really thought about it much until I saw your system (I've got a LEARNER plate on my bumper as I speed around these boards, bumping carelessly into people's systems). Or maybe most systems rely on simple stops, not versatile sell signals.
The only question I have about the Time Series Forecast exit is: Might it get you out too soon, as with AMTX before its big jump up, or MU?
|
