I had a case where I had a short-term day spike of 35% in APEI. The stock then fell back to almost the same level within the next week. In my ranking systems, it's normal that stocks that have large spikes also lose some of their score.
I usually only use score as a sell rule, but want to test an additional sell rule. What would happen to the overall performance if I sold stocks on the close of the day on "spike" days with 30% or more? Is that possible?
In general, I have found that, a positive push like that for highly ranked stocks is bullish. Did a simple test, you can do similar tests with your strategy.
Buy stocks that is ranked > 90% jumps 30%, hold for 10 days. That gives a positive outcome. If you that same with poorly ranked stocks... disaster!
Interesting, and thank you! It gave some good results, but why use "previous close" and not next open (or next close) so you can see the behaviour of the stock 10 days after the short-term spike?
We are analyzing the previous close. If you use next close you will miss a full day in the back test.
Example if you buy on ROC(1)>30 and sell with NoBars>1 on previous close. You are checking how the stock would react the day after a 30% gain.
If you buy on ROC(1)>30 and sell with NoBars>1 on next close. You are checking how the stock would react the day two days after a 30% gain.
At least that is how I understand that the back testing is working...
Here are the results:
What I have seen is that the same day as the stocks spike, sometimes the rank also falls sharply, and falls into sell territory. Like today I have the AEYE, up 22%, rank falls below sell.
Is there any way of testing a strategy where there is a spike of 20%, but at the same time the rank falls below < x, or is that impossible since the simulator uses weekly data?
Here are the settings I used:
