Financial Mathematics Text

Thursday, August 15, 2013

On why #themarketis up (down) today

So I'm suggesting that a new hashtag be used: #themarketis. The idea is to use it to explain why the market is up (down, etc) on a given day. Why, you ask? Allow me to explain.

Almost every day there's a "news" story on the market's price movements. Many of them offer explanations (rationalizations) on why the market price has moved. The explanations sound plausible but I think in most cases they are, at best, simplistic hypotheses which would not stand up to a robust statistical analysis.

The claim might be that the market is up/down due to concerns over a new Fed Reserve policy or instability in a particular country. Or perhaps the market is up/down because some bit of economic data came out that looked good/bad. Alternatively, a company might come out with a good/bad earnings report. There are plenty of examples; I'd guess you can probably find a headline of this sort on a daily basis.

The question becomes how do the journalists make who make these claim actually verify this information? Do they take surveys of some sort? How do they test which one of these hypotheses happens to be the better explanation?

My guess is that there's not any research at all into it. Furthermore, (and this would be my hypothesis), the bulk of the stock market price moves are more or less noise. (To borrow a phrase from Steve Keen, they are endogenous to the system). In either case, all of these hypotheses would need to be tested against the others to see if they have much validity. Most of them are probably nothing more than just so stories.

But the markets do exhibit a lot of noise. Using data from yahoo finance, daily returns for the S&P 500 exhibit a standard deviation (volatility) of 0.97%. Most of the price moves that occur (that are allegedly caused by this or that factor) are within 1 standard deviation of the mean. Here's what the volatility of the S&P 500 looks like:

S&P 500 daily price movements

As you can see there's some pretty violent swings. I might be willing to buy that some significant outside event is causing those price swings but for the rest, I remain skeptical.

And stealing an idea from Benoit Mandelbrot, here's what they would look like if they were normally distributed:

S&P 500 if normally distributed

Well, that's not too exciting...
But the broader point is that wide price swings are pretty common (and the price movement in the stories aren't typically all that large). Here's the historical frequency of price movements up or down larger than a particular threshold:

S&P 500 Frequency of Price Movement

There's about a 1 in 2 chance of a price movement greater than 0.5%, a 1 in 3 chance of a price movement outside 0.75% and a 1 in 5 chance of a price movement greater than 1%. So these types of movements are pretty common.

But these price movements are frequently given causal explanations related to market perception of economic and financial variables, etc.

Now I think it could be objected that a good portion of this endogenous price moves are a result of people's fears and expectations (or their expectations of others fears and expectation - see Keynes on higher order speculation) and that's probably a fair point. But that still raises the question of to what extent the particular hypothesis I'm positing (as opposed to some other hypothesis) is actually contributing to that volatility.

So #themarketis up (down) because . . .

Since I don't find the explanations in these news articles to be much more than speculative guesses, I think people can come up with more interesting explanations that are at least as plausible as the ones postulated in news articles. And for that I suggest the #themarketis hashtag. Here's how to use it:
Today, #themarketis up because because the Da Bears won yesterday!
 #themarketis down today because I got stuck in traffic on my way to work.

Perhaps some interesting hypotheses will be formed from this. So feel free to leave your best guess as to why #themarketis up or down today.

No comments:

Post a Comment

Some common OpenID URLs (no change to URL required):