Two things come to mind - if I'm teaching grandpa to suck egss here feel free to say so ;-) Optimization - you need to ensure that you only ever use parameters for trading going forward which you have derived from preceding data. Out of sample testing that derives parameters from time x and then tests a data set from time x-y is not valid. You also need to ensure that you don't overoptimize. Look for graceful degradation when making slight changes to numeric parameters. Abrupt degradation on slight parameter changes is a warning sign. The other consideration is whether your simulation accurately reflects the fills you will get in real life. if you assume that you hit the bid or take the offer every time you only need to consider market size. But if you enter and exit with limit orders you'll either be in a queue or get filled pro rata, and that makes a HUGE difference to results.
...have probably seen the same systems ..... ...these are about ......I've probably seen a couple more on the pure equities side ....
1% a day is very possible! I do know quite a few traders do that and even 100% a month. But like a few people answered liquidity becomes a issue at some point! One example is a guy I know who trades only $250K and makes a few mil a year doing it. But once in a few years he blows out the account and starts over. So following his example I would just keep a limited amount of capital that is safe and you don't mind losing. Thats what I am working on myself. Just a note this guy has been trading for 50 yrs and has warned me about going to big and told me how many great traders he has seen fall because they used too much capital.
Supposing you do 1% a day, would you keep on working till you hit 10+% a day? Is 20% a day impossible?
Well, GS have been unhappy internally about the return generated by Hull (not the person, the group). Rumor around Chicago is that Blair is itching to end his retirement and get back to trading, there have been persistent rumor that Blair would try to buy Hull Trading back, but these are just that, rumors. As famous as some of the trading firms are, such as Hull, DRW, Peak6, Wolverine, Stafford, etc, etc, they are not hedge funds, so their actual capital size is usually quite small (some are actually as small as $50M), problem with market making is that it is not capital intensive. So when ibanks take over MM operations, they usually try to pump capital into the trading operation, hoping for more returns, but they are almost always disappointed in return generated.
Again the assumption that one must and will compound. Still can't understand this assumption. Make 'em pretty, Chris
Yes, I optimize on back data and then apply it to out-of-sample data.. some times the out of sample results are better than the in-sample dat that i tuned the parameters with. Also, I plot the parameters and the performance is indeed smooth around the optimal parameters a so no worries there. About fills: avg trade is 20 mins and since the parameters are stable w/ regard to performance i dont think that'll be an issue. I have really-really fast order execution thru my broker.. capable of handling several thousand per second.
1% is my baseline.. so I can stop spending 12hrs a day crunching away.. and then I'll increase performance at a more relaxed pace.. my goal is to generate a decent amount of money each month and for me to be able to leave the corporate world and never have a 'real job' again. So, I'll keep working until until my ROI starts decreasing.. time spent vs. return increased.
Hmm, I guess this is a guy making directional trades.. highly leverged and not hedged? I'm very sensitive to risk and have failsafes to shut my system down if starts drawing down more than the expected amount.. hopefully that won't happen.. the performance should degrade over time and I re-optimize at the end of ever day.