My system performed poorly in December and now also late March. Both time periods coincide with strong rallies during a supposed downtrend. The biggest strength of my model is that it is adaptable and dynamic. It first determines the status of the market which I have called healthy or unhealthy. I hesitate to call it uptrending or downtrending b/c it's more sophisticated than that. It then uses variable parameters to calculate the signals, depending on the market's health. Both in Dec and late March, in hindsight it's easy to see the system was too slow with the switch back to healthy status. Of course, you also want to avoid the whipsaw action that's prevalent during unhealthy times, so the balance has to be just right. That is the key. With that in mind, I think I've accomplished the goal. The adjustment I'm making requires very little change to the model. I'm not messing with the code that calculates the buy/sell signals nor am I messing with how it handles the transition from healthy to unhealthy. I'm only changing how it handles the transition from unhealthy to healthy so that we are not shorting heavily into real strength. Here are some side by side examples of how the new technique compares to the old. First example, from December 2010 - the only month my system took a loss since the inception of this site. (click to enlarge, if necessary)
It's obvious the changeover back to healthy on the 24th makes a dramatic difference. Here is example #2 from last month...
Again, the switch to "healthy" on the 23rd allows the system to catch the Buy signal on the 23rd and the 28th. The percent gain in both these cases is significant to say the least. The examples go on and on in the past, but I will refrain from making this post longer than it already is. We probably won't see the benefits of this tweak for awhile since we are already "healthy", but I have confidence we'll be able to buy in near the bottom the next time the market is down. Have a good weekend...and here's to a gap up next Monday! :)
How are you determining the healthy and unhealthy market. The 13/34 EMA?
ReplyDeleteGood question! I like 13/34 and I still look at that, but I've never found it to be that useful. For market tops and the flip from healthy to unhealthy, I use the rolling inside-outside reversal technique using 10 daily bars on either the SPX or Nasdaq (which ever breaks first). It's described in-depth here: http://www.investopedia.com/articles/technical/04/031004.asp
ReplyDeleteI was using that same technique to find bottoms and flip the system back to healthy, but it's too slow during volatile times. I'm now using something that flips much quicker. It's a moving linear regression line(period 8) and a moving VWAP line (period 5). After the lines go positive it requires an up day for confirmation. This step avoids many whipsaws, not all, but many. Here's an example on freestockcharts. I also illustrate the top finding technique from above. Any further questions, just ask.
http://img688.imageshack.us/img688/9651/markethealth2.png