Monday, September 24, 2007

Why Quant may work?

In the previous post, we discussed how to construct a quant portfolio. Now let's try to understand why some thinks that it can work (ie it can outperform the market).

Well first, we must get the right factors though. If you screen for something like stocks that has hit 52 weeks high, or stocks with highest volume, or other funny factors, good luck. You have got the GIGO (Garbage in Garbage out) model. The model is only as good as the inputs.

What people usually believes as good inputs are like Low PER, Low PBR, High ROE, High cashflow, High OP margin, High EPS growth etc.

So there are roughly 400 stocks traded in Singapore and you only buy the top 50 with the lowest PER and highest ROE. What this means is that you are buying stocks that are cheap relative to all others and have the highest return potential relative to all others. And you do this every 6 mths, weeding out those that falls off the top 50 and adding new winners in. Theoretically, you SHOULD outperform the market.

But you don't. Murphy Law's works huh.

Well a few reasons. First of all, the data used are either historical or poor estimates. For PER, usually we get the 1-yr forward PER, which is basically the sum of estimates of all the analysts out there. And we know analysts are, well, like private bankers, GFN right? (GFN: Good-for-nothing). As for ROE, usually that's a historical no. so ROE may have changed, or dropped to below those of other stocks.

Second, to beat the market is a zero-sum game. You need to beat most of the other participants in the markets. This means you need to move faster than most other participants. Now when do you think these quant models were first used? Do you think you are one of the early birds using these models? The answer is NO btw. So investors have used this model since the last Ice Age, and here we are re-inventing the wheel and expecting to beat the market. That's not quite possible right?

But there is still hope.

The markets today, as with our world, has gotten very short-sighted. Thanks to MTV and instant noodles. Most people seek instant gratification. They are not interested in growing apple trees and waiting to eat apples years later. They are not interested in stocks that will only payback after 10 yrs.

So as we all know, the markets are unpredictable in the short term but follows earnings growth in the long run. The quant models, if used over long periods of time, should beat the market (esp if the rebalancing period is also stretched, so you don't get killed by transaction costs) bcos most other participants won't wait that long.
Post a Comment