Selection Bias

[peterbirks]

David Schwartz is the FT's Saturdday columnist in the Money section who writes about technical trading, resistance points, double tops, and the like.

This week he's said that he has been encouraged by the rebound in the FTSE 100. But what Schwartz then says that the heavy Footsie weighting of BP has "distorted" what Schwartz presumably thinks is a "truer" real FTSE 100 chart line.

I'm distinctly unhappy about this analysis. Schwartz never muttered about the heavy weighting of BP in the FTSE 100 when the bP share price moved from the mid-500s to the mid-700s.

Similarly, if funds have to invest somewhere in the FTSE 100, and if they pull out of BP, then this would serve to artificially boost the "non-BP" FTSE 100 line.

There's plenty of reason to argue that the FTSE 100 is artificially weighted in favour of certain companies which means that its movement is not a true reflection of market senntiment, but the point about technical trading is that you can't bring these factors into play after-the-fact. Either chartism works or it doesn't.

Scwartz doesn't actually call himself a technical trader any more (not in the FT column anyway), perhaps because in the equity market it's been historically unreliable. But technical analysis is notable for seeing "trends" in hindsight. But Schwartz is a short-term trader (where technical analysis in heavily traded stocks is not an invalid strategy) writing about fundamental matters. It makes for an odd mix.

_____________

Comments

[real-aardvark.livejournal.com]

Never thought of that before -- I've always seen technical traders as the equivalent of alchemists, with artificially chosen stats substituted for lead. (As an interesting aside, the medieval alchemists got the relationship between lead and any other metal precisely the wrong way round. But there you go.)

It doesn't seem unreasonable that chart-based analysis, with all its talk of support points and hand-wavy patterns, makes sense when you're a short trader. My question is, why does it make sense? For it to make sense, there is obviously a qualitative difference between short-term and long-term trading; that is, mandelbrot-style equations do not hold, for a given point. (That is, you peer down the microscope, and the landscape changes.) But this division is clearly not analogous, in any way, to the division between quantum effects and everyday physics.

I'm left with the conclusion that the only reason technical trading happens to work in the short term is because the mass of traders behave as though it works. It's sort of a subtle distinction, but since rational human beings like, er, myself who are outside of the markets listen to guff like "X has broken through the psychologically important 5,525 level" and think "loons!", it might be the case.

[peterbirks.livejournal.com]

Hi Mr D: The "short-term" technical analysis in currencies works on the principle of a large tail wagging a small dog. So I guess it's roughly the same with shares. Effectively there are a few traders who are trading for real purposes, which might not be related to fundamental investment (to take a simple recent example, a big rights issue might require people to buy sterling if they want to take up the issue, which strengthens sterling).

Since this is common knowledge, the movement tends to be magnified by people buying sterling on the strength of this fundamental requirement.

Now, move to the second level. You now know that 90% of a move has been "on the strength", while only 10% has been "fundamental". That 90% has no real need for sterling, and has also probably been bought on margin, which means that it will be sold, and sooner rather than later. Much of trading therefore is about the "closing of gaps", whereby you try to estimate how many of those people who bought "on the strength" have so far sold it back. If you think there are still a good few more who will have to sell, then you go short. If you think that that particular seam has been exhausted, then you start looking for new news and a new trend. If you find one (say, another potential long position), then you become part of the original "second" level, getting in with your long position (on margin), on the ground floor. Now other people start trading on the basis of your trades, and start trying to guess when you have sold back at a profit (or stopped out at a loss). These "resistance levels" are just short hand for saying ("that's where he'll take his profit, or perhaps he won't" and "that's where he'll have put his stop loss, or perhaps he won't".

Technical trading, on this basis, actually does work, but the more complex bits of Elliot Wave Theory, moving averages and the like, are really just twists on the same thing. "If 1 person follows it, it's shit. If 10,000 people follow it, it works".

PJ