RSI 2 vs 14 - How About 8?

I picked this up over on Bullish Jim's site * There's a great article on Trading Markets about the use of the Relative Strength Index as an indicator in trading. Interestingly, their studies show that the RSI (14) (which is shown in the top left corner of every single chart I post!) is basically useless as a short term indicator. However, their studies have shown great predictive value in using much shorter period RSI, specifically the the two period RSI.

Well I took that as a challenge since I have the technology to prove or disprove such a contention. I tested the RSI 14 and then the RSI 2 and, truth be told - neither one of them is worth much.

So I'm sitting there thinking about the RSI 2 (which did test a little better than the 14) and I said - wait a minute - I'm checking for an exit from below the 20 what if I checked for an entrance instead? So I tested for right after the RSI 2 slipped below the 20 line. The thinking being a short term RSI doesn't spend a lot of time at either the top or the bottom so you should look at them going into a condition rather than exiting it. Long story longer - the results improved but still were nothing to write home about. (I rechecked the 14 in the same manner and got worse results - which I expected - for obvious reasons).

I made one more adjustment - I changed the hit line to 10 and then checked stocks where the RSI 2 slipped below the 10 in the last day. And I obtained good results. Nothing that I would use in any shape, form or manner - but OK results.

I thought about that for a minute or two and I said if 2 below 10 gives good results I wonder what 8 below 20 gives? So I checked that out too. And that gave the best results of all. So how did I come up with 8? Simple - 14 minus 2 = 6 plus 2 = 8. Nah - I typed in 8 and didn't feel like changing it so that's how it became 8. I did try a couple of others afterward but 8 always tested best. I tried this filter at higher prices and lower prices without any real success. I have found that to be true that not all filters work equally well at all price ranges. That's intuitively obvious so I won't bother discussing it further here.

Here's the filter:

show stocks where close is between 15 and 35
and average volume(90) > 500000
and RSI(8) crossed below 20.00 within the last 1 day

The results were a 67% win rate, 127% ROI (annualized). And that's based on a 4 day hold period with 25K being invested in each selection. The selection is the top volume stock of the day from the filter output. The stop loss (fixed 8%) was triggered 2 times out of 62 selections. This filter won't find one every day but when it does it is a good one.

You can improve the hit rate a bit to nearly 1 per day if you change line three to RSI(8) < 20.00 but you will give up a little bit of win percentage and a little bit of ROI (which is unimportant).

I'm going to explore this a bit more later and combine RSI's and vary the entry and exit criteria a little and see what we can come up with without screwing the pooch.