Hello, I was wondering if there was any interest in starting a discussion on momentum investing/trend following. Obviously quite a bit of research has been done in this area and I was wondering if we could share ideas. For example, at its simplest we could use a simple ROC over some time period. Then we could use multiple ROC's, say a short-term, medium, and long-term. I believe IBD and stockcharts SCTR ratings are based on multiple ROC's. Then we have slope of a linear regression line*correlation which Andreas Clenow presents in his book Stocks on the Move. Also, we have distance from a moving average, slope of a moving average, and also a moving average ratio of MA(c,shortterm)/ma(c,longterm). I've tested almost all of these as the positionscore variable for rotational trading, and in my experience the simpler option perform just as well as the more complex.
But I'm fairly simple, and I know what a knowledgeable community this is. Would anyone else care to share some ideas for more investigation for momentum rankings?
Kindest Regards,
Anthony R
This question is fairly subjective, but I tend to get creative with my metrics, going as far as "inventing" my own in some cases.
I think the trick with all of this is understanding the fundamental concepts. Every indicator is measuring something, and is not just some line, or number.
Think: What do you see & how do you test for it?
I'm having fun using a Keltner band on oscillators.
Interesting avenue to pursue Possum, thanks for mentioning that. I'm finding more and more that it pays to watch other signs such as market breadth and VIX readings to get a read on the overall market.
Stating simply from my personal experience with technical analysis of markets:
The most interesting findings are those instances where RSI is making new extremes, while price appears to be somewhat "frozen" (or trapped).
It's always interesting to have some discussion.
You are correct, IBD uses 3 periods for momentum. I personaly also believe that it's better to keep things simple. Why use 3 periods if 1 period is giving you almost the same results. I believe in that it's best to add as little variables as possible.
I often notice that performance does get better with adding variables. I have seen momentum systems with 6 to 8 variables score 40% CAGR and when taking out half of them, you end up with 20%. I however believe that the 3 to 4 variable system will more likely be more robust than the 6 to 8 variable system.
In the end, most important is that it all makes sence. For example, if you have a momentum system where you pick the 10 best stocks, but now you add the distance from a moving average. Now your performance gets slightly better. Have you filtered out 2 or 3 bad trades ? Or does it actualy make sence ? Because for me, stocks with positive momentum will always have some distance from the moving average. Therefore, that filter... in backtests it might have filtered out some bad apples, but in the future, I have no basis at all to believe it will work just the same in the future.
So do I have anything to share to help with your momentum system ? Unfortunatly not.... Small stocks perform better, but you allready found that out I think. In my opinion, the best thing you can do to help your momentum system is find something uncorrelated.
Might anyone be interested in discussing divergence of price-to-momentum? Of course, I'm speaking of price/oscillator patterns.
Generally, in the past, this has been separated into either Type-1 or Type-2. I must confess that I've not been heavily involved with keeping current of recent trends in this field (looking to learn something new).
I used to think that Type-2 was largely a "trend following" condition. My more recent study has questioned the validity of my original premise.
Any thoughts, or desired conversation of the topic, are greatly appreciated.