Hi Grace A. Zaccardi; Thanks for continuing the discussion.
About short and long term stock price movements, and the independence of their predictability...
An example showing how it is that short term stock prices can be predictable, while long term stock prices are unpredictable. Suppose stock prices are determined by the throw of a coin. In the event of a "head", then prices move DUDUDUDUU. In the event of a "tail", then prices move DUDUDUDD. Heads make the stock go up by one price level, while tails make it go down by one price level. In the long term, the price movement is unpredictable. But in the short term, it is highly likely that a D is followed by a U and vice versa. You could make easy money making the spread.
If, on the other hand, long term stock prices are predictable, then short term prices also have to be predictable, but not by much.
If stocks did move as mentioned above, then it would be possible to make money (on average) by buying on the bid and selling on the ask. Daytraders can do this. The problem with making money this was has to do with the probability of getting a fill. If you are on the right side of the move, the probability is low. But if you are on the wrong side, the probability approaches unity. (G) Consequently, it is difficult to make money from the spreads on stocks, but it is possible.
Alternatively, one can consider a trading technique that doesn't require buying at the bid and selling at the ask. With these trading techniques, it is easier to get the fill, but the problem is that you lose the spread with each trade. In addition, on fast moving stocks, you have the same problem with fills as when you are trying to make the spread. If the market moves against you, you will likely get filled, but if it moves with you, there is a good chance you won't get the fill.
Most of the testing by the academics is silly. There are tons of people who make good money trading the market short term. The people who do this have no reason to go around publishing articles on how others can do this because they have no reason to increase the competition among traders.
While it is not possible to establish that a sequence is truly random, there are numerous tests that scientists regularly use to establish that a (finite) sequence is not random. I prefer the nonparametric statistical techniques, as they apply to all distributions of random data, but there are plenty of techniques. If it really were the case that scientists could not determine the difference between random and non random sequences, I guess there wouldn't be much science. Arguing that something can be believed only to the extent that it is proved with infinite sample sets is kind of silly, don't you think? That is an option God has, but man is restricted to the finite world. And anyway, how are the academics supposedly testing this if they don't have any infinite data sequences?
There are plenty of human abilities that seem impossible. Most people can't describe how to turn a tree into a paperback novel, but people do it all the time. Since we observe the paperback novels for sale in the grocery store, and since we are told where paper comes from, we conclude that it is possible to turn a tree into a novel. But very few of us are capable of performing that trick. Short term trading is similar. If it were not possible for short term trading to return profits, why is it that there are jobs advertised for short term traders?
The orders that enter the market are, by and large, not at all random. They are entered by humans, and those humans have reasons for doing what they do. A market maker who is trying to sell a lot of stock (for instance) does not do anything so simple as hold the market at a particular price. If he did so, the price would quickly drop below the price he wanted to sell at. The first drop might be because of random sell orders coming in, but the more likely consequence is that daytraders will sense that there is a major sell order being filled, and they will sell short in front of it. The whole thing is an intensely complicated game with varying amounts of information available to the various parties. There is no way that someone from academia would be able to make money in that environment without going through the same learning experience that confronts all new traders, but very few academics have trading experience, much less successful trading experience.
Re tails of distributions, some explanation... If micro prices truly are random, then by the strong law of large numbers, macro prices have to be normally distributed. (An example would be day to day price changes.) But when macro prices are compared against a bell curve, they end up with fat tails. If you are a mathematician type, the above will be a good explanation. If not, it probably seems a bit mysterious.
-- Carl |
