http://thequantinvestor.blogspot.com/2011/04/how-to-completely-outperform-market-by.html
Like any strategy it can be improved on, and in this post I'll try to do just that. Here I'll add another important consideration when buying a stock, the volume backing its price movement. To do that, let me introduce what I call the Volume Moving Average Ratio (VMAR). What this simply is, is the ratio of the 20 day volume moving average to the 75 day volume moving average (you can use other lengths of time as well):
VMAR = VMA(20 day) / VMA(75 day)
What this tells us is how much the recent average volume is compared to the longer term volume. I used the simple moving average but an exponential would work as well. In theory, a VMAR >> 1 means a stock has significantly more investment activity than usual and, correlated with a price gain, indicates that perhaps its price movement IS warranted.
Now we could potentially cover a huge combination of VMARs and hold times. I'll leave it up to you, the reader, to find the best combination. But now I present to you two scenarios of short (1 or 3 month) hold times with VMARs greater than 1 or 2.
What we see is impressive...enormously impressive. In fact a strategy of 'buying high' backed with volume can make you a very wealthy person. Of course I make no promises, and any strategy that worked last decade may not this decade, and you should consider that. Also, don't forget commissions, as with any trading strategy, can easily take 1-2% off every trade. But these are the facts, and the numbers, and of that I am 100% certain. So here you go...enjoy!
1 Month Hold (VMAR > 1.5)
10 Year performance Avg. Return Stocks meeting criteria
NYSE 756% 1.7% 1947
Nasdaq 597% 1.5% 4653
Amex 1531% 2.3% 885
1 Month Hold (VMAR > 2)
10 Year performance Avg. Return Stocks meeting criteria
NYSE 4381% 3.2% 330
Nasdaq 161% 0.4% 1507
Amex 1936% 2.5% 317
3 Month Hold (VMAR > 1.5)
10 Year performance Avg. Return Stocks meeting criteria
NYSE 314% 2.9% 1005
Nasdaq 539% 4.3% 2905
Amex 918% 5.7% 573
3 Month Hold (VMAR > 2)
10 Year performance Avg. Return Stocks meeting criteria
NYSE 1068% 6.1% 182
Nasdaq 248% 2.3% 1000
Amex 918% 5.7% 236
Neat Idea, Troy. But what is your "criteria" that stocks need to meet and do you have any idea why this works?
ReplyDeleteThe criteria are just what I have layed out... 52 week high and a certain VMAR, thats it.
ReplyDeleteWhy it works, clearly a company at a 52 week high with significant volume backing is doing something right, at least as far as the investors are concerned. The short-term hold also probably is a bit of a momentum play.
Volume requires both buyers and sellers, no? If investors as a whole think that a company is doing something right, then shouldn't the volume be low because investors don't want to sell their stock in this wonderful company?
ReplyDeleteYou talk about "perfecting the strategy" but what I see here is tuning two parameters to maximize returns with no control for overfitting. You've basically guaranteed that the 4381% return that you've claimed here is greatly exaggerated. And then there's the question of how many other variables besides VMAR have you've tried this with.
So believe me I know all about overfitting, as would anybody who has a Ph.D in a hard science. But overfitting is a much overused term, especially for people who don't understand it. The fact is that I haven't fit anything, only spanned to VMAR parameter points. If I maximized return for a space of VMAR then I would 100% agree with you, but I didn't. As for the 52 high buy-strategy, thats a single parameter with no change, no fitting there.
ReplyDeleteAll I've presented is another way to further qualify the purchase of a well performing stock. As for other variables, of course I've investigated them. Isn't that the point of any quantitative investigation, to find underlying measures that quantify the data?
Dr Lau,
ReplyDeleteI'm not sure if you're still active with this blog but I'll give it a try.
I was testing this strategy and I can see the great improvement made by the VMAR.
However, we're looking into the future here because there's no way to know what the VMAR is before the end of the day.
If you do the same test with VMAR of the day before, the returns decrease dramatically.
Also I was wondering if you did some forward testing and found a way to implement this strategy in real life, what the slippage could be etc...
Thanks for your insight,
Jo