March 22nd, 2014, 9:44 pm
QuoteOriginally posted by: JeriotI suppose the motivation is to get some guidance from the right supervisor who is more experience and have a better inkling of what works and what doesn't - hence the desire to start looking for a job again.Trend trading works in all timeframes. As long as there is not too much hot money flowing into it, and as long as there is not too much upheaval, your competitors will only trade trends so far as they still make a fair return. So then you put the labor and capital and technology into it, and try to execute more efficiently, like any mature industry. You may find small innovations that allow you to make 1.25 times the fair market return, rather than 0.75 times. Being faster works, as do other microstructure advantages such as exchange memberships, whatever. Other than that, I think it is a changing game. Systems that continually find new patterns probably work, but I don't know if you can expect stable profits over a long period. It probably helps to know your competitor, who is changing whether daily or yearly.QuoteOriginally posted by: JeriotI don't think I really understand what you mean hereThink about a chess game. You have perfect information. In theory, whoever goes first probably wins. If you assume your opponent doesn't have perfect information - or, more accurately, assume that he does and just can't execute by using it perfectly - you could try to win faster, by guessing his patterns or predicting where he will be shortsighted. But assuming the reality that both you and your opponent have perfect information, the challenge is to have a mind or machine that can actually process it and execute. I am not saying I have perfect information for any market. I am just saying I have enough that I have this same problem of using what I know.I read a story for the layman on Deep Blue. It said something like Deep Blue can't work through all the gazillions of permutations. But what they did, is come up with a way to rate certain setups. If you have only a few pieces like a bishop and a king and a knight, and your opponent has even less, you can chase your opponent all over the board all day. It might take a lot of computer just to calculate all the different paths your opponent could run away. But a general evaluation would say this is a good setup to get to, even if we don't test every alternate scenario beyond this point. So I guess they did that also for configurations with more pieces. Maybe they ran through every possible move for some setups, and then said all these alternate but similar layouts have this same general quality.So in low-frequency algorithmic trading, or discretionary trading, there may be too much to think about at once. So like any decision, you have to try to distill out some key indications. Even if everyday life, I have lots of rules which I don't remember the origin of. Or, for example, you may have a friend who lies a lot. So long after you have forgotten all his individual lies, or even after you have forgotten everything he ever said, you might remember to not believe him. So you have to learn, and forget. Like learning that a certain chessboard configuration is good, and then forgetting why. So you need some way of making discoveries, and then having a means for storing and reusing them. And also for modifying and invalidating them, and organizing them.
Last edited by
farmer on March 21st, 2014, 11:00 pm, edited 1 time in total.