Protecting your strategies

achilles28 · Jun 29, 2012

Quote from amazingIndustry:

Dude, it happens all the time, most cases just never make it to the courts or public because many hedge funds and sell-side desks design very opaque contracts on purpose. Hedge funds have never been more active to "harvest" strategies. They are right now on a "hiring binge", on the surface saying they want to hire PMs and have them bring on board their successful strategies (with track record). Once the code makes it into the firm it is so fast copied or reverse-engineered that you cannot even blink faster with an eye.

The only protection against that are solidly programmed and obfuscated code libraries. I generally test fund's attitude towards that by saying that "I am happy to bring my strategies on board but I like to trade them in a black box type of fashion and that I am very solid coder and am confident the code cannot be reverse engineered. I also built in a permission algorithm that connects to an outside server and should I ever leave the firm and take my code with me (the one I brought with me not in-house written code) the permissions will be revoked and the strategy cannot be traded anymore".

Guess what happens then: I hit brick walls. In almost all cases funds used all sorts of excuses that such deal would not work out. No logical explanations given, only obscure excuses. Makes it very clear what they are really after. I am not saying there are no other funds out there. Some more trustworthy firms who look to seed traders mentioning they would not have any objections about my policy whatsoever.

This, in my opinion, is the only way to protect against outright theft. Do your own work and never trust anyone, no broker, no hosting firm, no hedge fund, nobody. If you have something that generates risk-adjusted stable streams of PnL then you are on most investors' most wanted list, well not you but your code.

Never. Impossible. Wallstreeters hold themselves to the highest ethical standard!!!

trend2009 · Jun 29, 2012

I said I decoded one of such strategies before, though not sure 100% identical, but at least I benefited a lot.

Quote from jcl:

Anyone can decode a strategy when he has the code. But not when he has only a number of trades. That was the problem discussed here.

OddTrader · Jun 29, 2012

I would think the experienced developers do Not need to get your codes physically.

My guess is All they need is to understand your underlying logics of the strategy/system, plus your actual/back-test signals and results/PnL. And time - probably years.

Quote from amazingIndustry:

Dude, it happens all the time, most cases just never make it to the courts or public because many hedge funds and sell-side desks design very opaque contracts on purpose. Hedge funds have never been more active to "harvest" strategies. They are right now on a "hiring binge", on the surface saying they want to hire PMs and have them bring on board their successful strategies (with track record). Once the code makes it into the firm it is so fast copied or reverse-engineered that you cannot even blink faster with an eye.

The only protection against that are solidly programmed and obfuscated code libraries. I generally test fund's attitude towards that by saying that "I am happy to bring my strategies on board but I like to trade them in a black box type of fashion and that I am very solid coder and am confident the code cannot be reverse engineered. I also built in a permission algorithm that connects to an outside server and should I ever leave the firm and take my code with me (the one I brought with me not in-house written code) the permissions will be revoked and the strategy cannot be traded anymore".

Guess what happens then: I hit brick walls. In almost all cases funds used all sorts of excuses that such deal would not work out. No logical explanations given, only obscure excuses. Makes it very clear what they are really after. I am not saying there are no other funds out there. Some more trustworthy firms who look to seed traders mentioning they would not have any objections about my policy whatsoever.

This, in my opinion, is the only way to protect against outright theft. Do your own work and never trust anyone, no broker, no hosting firm, no hedge fund, nobody. If you have something that generates risk-adjusted stable streams of PnL then you are on most investors' most wanted list, well not you but your code.

amazingIndustry · Jun 29, 2012

developers (as in programmers) would not know what to do at all with trading signals. Experienced strategists and quants may be able to use trading signals to understand the overall strategy mechanics. But today's working and profitable strategies implement so many strategy details that generally it would take a long time to reverse engineer an exact or close enough copy of a strategy from signals alone (without seeing any line of code).

The firm knows anyway that a strategy is momentum based with this type of risk mitigation algorithms and that profit taking algo. Those alone do not make a working strategy. Please notice I am not talking about some MT5 or Amibroker or Ninjatrader or what have you retail "stuff", I am talking about strategies that trade high frequency or intraday frequency on hundreds if not thousands of stock symbols, G10 currencies, options chains, ...

Quote from OddTrader:

I would think the experienced developers do Not need to get your codes physically.

My guess is All they need is to understand your underlying logics of the strategy/system, plus your actual/back-test signals and results/PnL. And time - probably years.

jcl · Jun 29, 2012

Quote from trend2009:

I said I decoded one of such strategies before, though not sure 100% identical, but at least I benefited a lot.

My bet still stands: $1000 for the first one who reverse engineers a relatively simple profitable strategy - no neuronal network, pattern detection or the like - based on 100 trades. Needs not be 100% identical, but should come close.

OddTrader · Jun 29, 2012

You're correct. I meant developers of strategies like this:

http://www.wilmott.com/messageview.cfm?catid=5&threadid=90608 .

Please notice I didn't even mention exactly whether they could attained 100% (or 80% or 120%) close to their target results.

Quote from OddTrader:

I would think the experienced developers do Not need to get your codes physically.

My guess is All they need is to understand your underlying logics of the strategy/system, plus your actual/back-test signals and results/PnL. And time - probably years.

themickey · Jun 29, 2012

Quote from jcl:

My bet still stands: $1000 for the first one who reverse engineers a relatively simple profitable strategy - no neuronal network, pattern detection or the like - based on 100 trades. Needs not be 100% identical, but should come close.

To reverse engineer doesn't take hours, nor days, it may take weeks or a couple of months or so, from my experience. The $1k reward is not even worth contemplating.
As I speak I'm working on something, all evening and no headway yet, not even a glimmer, but I know I will crack it.
Just takes logic, bit like detective work.

jcl · Jun 29, 2012

Quote from themickey:

To reverse engineer doesn't take hours, nor days, it may take weeks or a couple of months or so, from my experience. The $1k reward is not even worth contemplating.
As I speak I'm working on something, all evening and no headway yet, not even a glimmer, but I know I will crack it.
Just takes logic, bit like detective work.

Yes, the $1k is of course no compensation for work, just a sort of incentive. If I claimed on a forum that I can do something, and some other guy doubts it, I would be motivated to prove him wrong - even without money.

But for how much would you do it?

trend2009 · Jun 29, 2012

the real incentive is that you should provide a system that worth decoding. if you have a mechanical system that has 75% win ratio, and the average win at least not bigger than loss, I will take your challenge.

It is idiot for anyone to spend months on a garbage system just to prove it is a garbage.

Quote from jcl:

Yes, the $1k is of course no compensation for work, just a sort of incentive. If I claimed on a forum that I can do something, and some other guy doubts it, I would be motivated to prove him wrong - even without money.

But for how much would you do it?

gmst · Jun 29, 2012

Quote from trend2009:

the real incentive is that you should provide a system that worth decoding. if you have a mechanical system that has 75% win ratio, and the average win at least not bigger than loss, I will take your challenge.

It is idiot for anyone to spend months on a garbage system just to prove it is a garbage.

Yes this is a good and valid point.

JCL should provide 100 entries/exits of a good system with PF at least 1.7, to make it worthy of decoding.

If this is true, I will also take an attempt at decoding the system.