Saturday, March 03, 2007

Another Day - Another Idea

After getting the results of the ATR and RSI filters I wondered what would happen if I combined these two filters together into one. The good news is - it works - the better news is - it works better than the two filters separately.

show stocks where close is between 15 and 35
and average volume(90) > 500000
and Average True Range(10) < CMA(average true range(10),10)
and rsi(2) < 2

You will note that the RSI(2) is now "< 2" which is probably as rock bottom as it can get before it has to reverse direction. Now it can stay down there for quite a few periods and the price can go lower before it reverses direction. That is where the ATR less than 10 comes in. Remember that the ATR is being used as a proxy for volatility and in the words of Bollinger - low volatility begets high volatility. There was more to than that but that's what we are looking for - a stable base to build on. So by combining the low ATR with the low RSI we are minimizing the amount of time that the RSI stays on the floor - at least that's what we are trying to do.

This version's results were: 69% win rate, 3.95 Reward/Risk Ratio, and 181.94% ROI. Which is impressive. But what is really impressive is the longer term effect on stocks that have this particular set of starting conditions and I'm talking about net change over a period -

4 days - 1.56% and 20 days 3.36%

What that means is that not only is this an efficient short term filter but is is also an efficient longer term filter as long as the Market conditions warrant. Market conditions warrant" means that the market is also poised and ready to go up.

The reason for this is once the momentum is in effect it continues for quite some time normally - or at least 20 days.


Bullish Jim said...

That's a very cool filter.

I'm being dense here but would you mind explaining what the CMA portion of the equation is asking for?

I've been tinkering with this filter at stockfetcher but I can only backtest 4 months at a time. What time period did you test?

The default in stockfetcher is to assume 10% stop loss and 10% profit target. In general I've noticed that tightening the stop loss decreases win rate but increases ROI. Just curious what profit and loss targets you used.


Another day another dollar.