Can a System be reverse engineered just from viewing daily trades/account statements?

i think it all comes down to the specifics of the system and the cracker. if they both work in the same category - let's say break out on eMinis or turtleTrading - then the decoder has a chance.

the returns on the account will make the rboker wish he could do the same. he will try to find people doing that thing, might find them and start digging. it is very unlikely from my perspective to find out the exact system, but it could be that he comes close enough to hurt the performance of the first system. if i was a trader trying to decode i would write a simple version that covers all trades of the system plus many more and then let factor analysis try to figure out which factors were used to reduce my sample to the one i try the replicate. i think it is really a lot of work to do such thing and it must be a huge system to justify the expense. and it is an exclusive job - you could not crack when doing other things. for months.

easy way to get around the problem is to mix two systems. makes it from my perspective almost impossible to find a starting point for the cracking party.


peace
 
Quote from man:

i think it all comes down to the specifics of the system and the cracker. if they both work in the same category - let's say break out on eMinis or turtleTrading - then the decoder has a chance.

the returns on the account will make the rboker wish he could do the same. he will try to find people doing that thing, might find them and start digging. it is very unlikely from my perspective to find out the exact system, but it could be that he comes close enough to hurt the performance of the first system. if i was a trader trying to decode i would write a simple version that covers all trades of the system plus many more and then let factor analysis try to figure out which factors were used to reduce my sample to the one i try the replicate. i think it is really a lot of work to do such thing and it must be a huge system to justify the expense. and it is an exclusive job - you could not crack when doing other things. for months.

easy way to get around the problem is to mix two systems. makes it from my perspective almost impossible to find a starting point for the cracking party.


peace


Good post!

Peace to you too!
 
If it’s not some sort of a random system, then I think it’s quite possible.
I’m not sure one would need months to crack a system.
Just get as many factors as you can find – like prices, opens, closed, indicators – whatever (let say a thousand). Get trades. Combine them together. Run cross correlation between them. Then get most correlated indicators with your trades (let say 10). Combine them again together. You’ll get about a thousand combinations of those indicators. Again run correlation with your trades. Find most correlated combination. 3 hours.
If it's not the system, then should be close :(
 
Quote from man:



it is very unlikely from my perspective to find out the exact system, but it could be that he comes close enough to hurt the performance of the first system.


Thanks Man. What do you mean "hurt the performance"? Please explore. :)
 
Quote from CPTrader:

ETs,

Do you think that a System can be reverse engineered just from viewing daily trades/account statements?

For example, if you are a systematic trader and run for instance a manged account for a very market-savvy client, do you think they can review your historical trades and reverse engineer your system?

Could your broker also do this?

If so what steps can/should be taken to AVOID/ELIMINATE the possibility of a client (via a managed account) or even a broker reverse engineering your system?

Thanks!


I think that if you method is simple and straightforward, then people are either using it or have already tried it. If the method is somewhat complex, then people will probably not "crack" it. Add to the mix your occasionally inconsistent application of the method to further stump your trackers.

Is it possible that you are getting ahead of yourself with your paranoia?

Regards,

Thunderdog
 
Majority of the customers loose money, anyone who makes money sticks out like a sore thumb. I can guarantee there is an attempt by someone at the brokerage house to take a closer look. It maynot be the management, but some IT guy working late etc...

If you recall reading Mark Cook's interview in the mkt wizards book, he was doing really well trading options, one day he got a call from their compliance dept. who were inquisitive about his trading. Similar story by Marty Schwartz who complained about his broker front running his order.

If you're doing size, I believe the only way to get around this headache is to use multiple brokers. I hope one day I have this headache

:)
 
Quote from CPTrader:

Do you think that a System can be reverse engineered just from viewing daily trades/account statements?

assuming the trades are timestamped, absolutely. they may not come up with exactly the setup you're using, but it won't be very hard at all to come up with a system that duplicates better than 95% of what you're doing.

if you let people see the trades, and you are systematic, there's nothing you can do to stop it.
 
Quote from man:

ieasy way to get around the problem is to mix two systems. makes it from my perspective almost impossible to find a starting point for the cracking party.

my background is ultra-technical and i've done my share of reverse-engineering extremely complex systems. what you describe might slow down the average retail schmoe but it wouldn't slow down a pro for more than 15 seconds.
 
Quote from damir00:



my background is ultra-technical and i've done my share of reverse-engineering extremely complex systems. what you describe might slow down the average retail schmoe but it wouldn't slow down a pro for more than 15 seconds.

So what can you do to ensure an ultra-technical pro fails in a reverse engineering attempt.
 
Back
Top