A Tale Of Two Filters

I often switch my filters around to see what happens - for example if I'm requiring a close to be greater than the EMA(140) for example I'll change the filter to require the close to be less than the EMA(140) and back test the two of them to see what the differences are.

I did that this morning with a crossover filter that I've been working on and came up with unexpected results. Here are the two filters side by side - the original one is on the left - the mirror image one is on the right -


And here are the results of the back test -


The crossover pattern is one of many "trader's observations" that I've made over the years and I almost always take a trade with this pattern on any time frame but I usually require that the EMA 4 be above the EMA 21 - it just seemed to me to be a better way to do the deed.

Consequently this morning when I switched the pattern to have the EMA 4 below the EMA 21 and then switched the open/close appropriately I was very surprised to see a better ROI albeit a worse win percentage. The really strange thing was both filters made 41 selections over the same period. I quickly brought up the trades page and reviewed the selections and was relieved to see that they were totally different sets.

Then I changed the test period from the current period 11/06 to 3/06 to the very prolific period 10/06 to 2/06 and the results once again matched the expectations.


This time there were 42 and 43 selections.

Another result that I use frequently is the net change over 30 days and once more the inverted filter (4 < 21) won the challenge. The difficult period was only a few percentage points apart but in the fertile period it had 4.35% to 3.33% advantage. Both of these numbers are respectable but the inverted filter seems to test better in the long term.

The crossover pattern appears to work either way and ultimate results depend on the market conditions which is always the case. If you use the crossover pattern the best place to put your stop is just under the lower of the two moving averages (either EMA 4 or EMA 21) because if its going to fail it will do so almost immediately. I always plan for two or three up days out of this approach and then play it day by day after that.

This is not an invitation to speculate in the stock market - it is a deep, dark, dreary place filled with money eating demons.

For anyone interested in shorting a stock - the crossover pattern works just as well on the dark side as it does on the sunny side - just make appropriate open - close changes.