If the problem is computationally difficult, then it is a mistake to not count the value of computation.
Suppose you believe that a reasonably smart person with some spare time can make money on the stock market. There is still a question of how much money, compared to the amount they could earn doing other things. If we pick some reasonable value for intelligent, market understanding thought, then we can define a new value α∗ which is α - the value of the time it took to find.
The value of thought time, compute time and data are all constant. The returns are linear until you get to the point where you disrupt the market, destroying the pattern. Thus α∗ is much harder to get if the amount of money you have to invest is small. (Especially if you don’t bet the farm.)
There are different reasons why alpha exists and persists. One misconception that seems to be common is that alpha could only persist because of secrecy. Now that kind probably exists, but finding and exploiting it is catching lightning in a bottle. It’s awesome if you can do it, but it’s probably not worth the effort. There are easier games.
Some types of alpha do indeed require a lot of compute to find, but you can speed that up by renting compute in the cloud (which does cost some money). And you can search more efficiently by having educated guesses about where to look, which also seems like something rationalists might be good at. And yes, how much you can take advantage of that depends on your capital. Seen from the other side, this is one of the advantages of trading though. You can earn faster by saving more money. It scales in a way a normal day job doesn’t, which may only give you small raises that sometimes don’t even keep up with inflation.
But you can do this kind of trading more efficiently by cooperating with other agents. You can share information. The deal works like this: you tell the others what your edge is, they reciprocate and tell you about theirs. This is a cost to you. The information wasn’t free, and sharing it diminishes its value, because alphas diminish the more they’re traded. However, you didn’t have the capital to max it out yourself anyway. Your friends probably don’t either, so all of you can trade it for a while. If you have a dozen trader friends and you each had one edge, now you all have a dozen plus one edges, with the benefits of diversification. Thirteen edges for the price of one. Secrecy isn’t always the best move.
(This is also one of my motivations for writing this sequence. More trader friends means more edges for me!)
Other edges persist because people, for one reason or another, find exploiting them distasteful. E.g. a skewed risk profile. In some cases, it may simply be constrained by capital. It could be a very good deal for a small trader, but a large trader with different opportunities wouldn’t bother, because they can’t trade it with sufficient volume to be worth the effort.
If the problem is computationally difficult, then it is a mistake to not count the value of computation.
Suppose you believe that a reasonably smart person with some spare time can make money on the stock market. There is still a question of how much money, compared to the amount they could earn doing other things. If we pick some reasonable value for intelligent, market understanding thought, then we can define a new value α∗ which is α - the value of the time it took to find.
The value of thought time, compute time and data are all constant. The returns are linear until you get to the point where you disrupt the market, destroying the pattern. Thus α∗ is much harder to get if the amount of money you have to invest is small. (Especially if you don’t bet the farm.)
There are different reasons why alpha exists and persists. One misconception that seems to be common is that alpha could only persist because of secrecy. Now that kind probably exists, but finding and exploiting it is catching lightning in a bottle. It’s awesome if you can do it, but it’s probably not worth the effort. There are easier games.
Some types of alpha do indeed require a lot of compute to find, but you can speed that up by renting compute in the cloud (which does cost some money). And you can search more efficiently by having educated guesses about where to look, which also seems like something rationalists might be good at. And yes, how much you can take advantage of that depends on your capital. Seen from the other side, this is one of the advantages of trading though. You can earn faster by saving more money. It scales in a way a normal day job doesn’t, which may only give you small raises that sometimes don’t even keep up with inflation.
But you can do this kind of trading more efficiently by cooperating with other agents. You can share information. The deal works like this: you tell the others what your edge is, they reciprocate and tell you about theirs. This is a cost to you. The information wasn’t free, and sharing it diminishes its value, because alphas diminish the more they’re traded. However, you didn’t have the capital to max it out yourself anyway. Your friends probably don’t either, so all of you can trade it for a while. If you have a dozen trader friends and you each had one edge, now you all have a dozen plus one edges, with the benefits of diversification. Thirteen edges for the price of one. Secrecy isn’t always the best move.
(This is also one of my motivations for writing this sequence. More trader friends means more edges for me!)
Other edges persist because people, for one reason or another, find exploiting them distasteful. E.g. a skewed risk profile. In some cases, it may simply be constrained by capital. It could be a very good deal for a small trader, but a large trader with different opportunities wouldn’t bother, because they can’t trade it with sufficient volume to be worth the effort.