Tuesday, October 5, 2010

Trading Backed by Statistical Data

I believe that trading based on quantified, back tested results is very good approach to adopt. Obviously, past performance is no indication of future results, but back-testing a strategy at least gives you a good idea of what might happen. I am looking a JC Penney ($JCP). It has risen 37% in 18 trading days (up 16 of 18), it's 2 period relative strength [RSI(2)] is at 99.97 -  an obscene level that historically cannot be sustained for much longer. I went and looked at how many times JCP had a RSI(2) level above 99.95. That was the only filter I used. I wanted to test what would happen if I sold short the morning after it reached that RSI(2) level, held for 5 trading days, then covered at the open of the 6th day. Here is what I found:

RSI (2)TickerTradeDatePriceEx. dateEx. Price% chg
99.96JCPShort11/13/1986       4.87 11/20/1986        4.49 -7.71%
99.98JCPShort2/19/1987       5.27 2/26/1987        5.52 4.57%
99.98JCPShort2/7/1991       9.20 2/14/1991        8.78 -4.59%
99.97JCPShort9/21/2009    32.88 9/28/2009      31.44 -4.36%
99.96JCPShort3/24/2010    32.86 3/31/2010      31.99 -2.67%

The stock moved down an average of 2.95% over those 5 days. It was correct 4 out of 5 times. This is just an arbitrary example of how to use back-tested data to make an informed trading decision. Knowing this information can give you confidence so that you are not just going in there blindly.


Chris C said...

I guess I should eat crow on this. Selling short would have lost 18%. It's pretty large, but not a account-killer. With proper position sizing, the hit shouldn't be so bad. The one un-predictable nature about this instance was the Ackerman share acquisition. Tough to forsee possible buyouts. Stats can be wrong, but in the long run, I trust the numbers. Don't get complacent; use risk management.

Post a Comment