The word "commissions" is typically used to designate an expense, not an income. I suppose what you originally meant to say was "management fees", or perhaps "incentive fees". Otherwise, the implication is, indeed, as Handle123 pointed out, that the profits are derived from the order flow rebates from exchanges/counterparties.
Nobody gets your point here and yet you still come across as being incredibly arrogant. Could it perhaps be an issue on your end? Or is that an impossibility? And as just pointed out while writing this post, it was you who confused terminology. Perhaps a more humble attitude would have spared you some embarrassment.
A hedge fund manager who can earn a fat performance fee on billions of dollars will make more money faster than trading his own account. Even the best hedge funds rarely average more than 20% per year. Averaging 20% a year on your own money is a respectable feat, but it won't make you wealthy very quickly. However, if your multi-billion dollar hedge fund can earn 20% a year and you receive a 30% performance fee on that 20%, you are going to make a tremendous amount of money for yourself.
Are you bored or something? There is always something to backtest. You know, you come across as someone who walks into a bar and says, "Barack Obama". When everyone looks at you with a question mark on their faces, you think to yourself, "Is it not abundantly clear that I've just ordered a "Bloody Mary" and a "Cosmopolitan"?
It's HF market wizards, chapter 5. I recently re-read this. For me the most interesting part of this story is that he is probably more secretive than anyone else in the book about what he does, and I found the description of what he did pretty intruging. So he's doing some kind of non linear analysis, or linear analysis on derivations of observable data like price, but on effects that work across markets, and also on effects that have persisted for some kind. So quite different from the normal data mining approach I'm so skeptical about. By the way I'm not sure what "a hundred thousand dollars worth" of data actually means, or meant in the past. Nowadays you could get that data for free. The normal model for getting data is a subscription model, which means you might be paying up to $500K a month to get tick level data from every exchange in the world, but the value of that data is hard to quantify. GAT
how could you find what the author does intriguing if all you know is that he applied some "non linear, or linear analysis on derivatives of price". Thats like Ferrari revealing its new concept car and telling the press that it has an engine and 4 wheels.
Because most people who do that kind of thing fail badly, through overfitting. And they normally fit specific to a market, and they also have to refit their models frequently. So this guy is doing something genuinely different, and making a success of something that most people fail at. Personally I find that interesting. There's also enough of a hint that I can speculate on what they might be doing, compared to say Jim Simons where I have absolutely no clue and I probably couldn't replicate it anyway. So it's more like Ferrari revealing a concept car with no engine that still runs. GAT
You can get unlimited amounts of free daily data from www.quandl.com They have a lot of intraday data as well at low cost.