Protecting your strategies

[logic_man]

The best analogy I can think of to explain why this is impossible is that of target shooting. You can miss the target any one of a million ways, but you can hit the exact center of the target only 1 way. For a firm to try reverse-engineering your strategy is like trying to hit the exact center of a target when they have no idea whether the target is in front of them, off to the side, above them or below them. It's simply too risky to even bother with. If those guys want to trade their capital, they can just go out and hire traders, not reverse engineer their customers' trades.

Any strategy sufficiently sophisticated to be profitable in the world's most liquid markets cannot be reverse engineered using only entrances and exits. When that guy from Goldman Sachs got busted for trying to sell Goldman's program trading software, he was selling the code, not Goldman's entries and exits. You could say that's because he had access to the code and that's true, but it was probably more risky to steal that code than it would have been just to give over the entries and exits to a buyer who could then crack the underlying algorithms.

A POS strategy that might have a hot streak for a few months but is otherwise unremarkable and destined to fail might be able to be reverse engineered, but there is no value in doing so.

 

[Pekelo]

Quote from logic_man:

The best analogy I can think of to explain why this is impossible

Impossible is such a big word, not to mention you are over generalizing. Some strategies are easy to replicate others are not so much.

Also, there is a possibility that a super-duper complex strategy can be substituted fairly well with a much simpler strategy. It doesn't need to be exactly the same. For example if you original strategy makes 40% and the much simpler strategy (but based on the original's entries) makes 32% that is still an acceptable substitution...

 

[logic_man]

Quote from Pekelo:

Impossible is such a big word, not to mention you are over generalizing. Some strategies are easy to replicate others are not so much.

Also, there is a possibility that a super-duper complex strategy can be substituted fairly well with a much simpler strategy. It doesn't need to be exactly the same. For example if you original strategy makes 40% and the much simpler strategy (but based on the original's entries) makes 32% that is still an acceptable substitution...

Yes, the ones which are easy to replicate don't work. That's why they are easy to replicate.

There was another thread where I recently did a back of the envelope calculation of how many possible permutations of even the most basic TA indicators there are and the number is in the trillions (hundreds of indicators times thousands of possible optimal indicator values times thousands of ways to combine the indicators). Being able to reverse engineer against that level of complexity is so difficult that it's probably easier to build your own strategy from scratch.

Also, I doubt that you could make alterations to a strategy and have it still hold up well. The problem would be that, over time, the simpler strategy would have an entirely different set of trades. Strategies are kind of like snowflakes in that way, where they have a unique set of trades associated with them that no other strategy will copy exactly.

It's definitely true that there are funds who will try to hide their buying and selling activity by various means. But, I think they do that because they don't want anyone "plagiarizing" their orders, not that anyone could reverse engineer the strategy by knowing when they were buying and selling.

I'd worry more about my house being struck by a giant meteor than I would worry about my strategy being stolen. I guess it wouldn't surprise me if brokers ran reports to see which of their customers were the most profitable, but I doubt they could ever take any action to derail their profitable customers' profitability or copy it.

 

[gmst]

Randomizing some entries is probably an interesting idea, but it will work only in some circumstances. E.g. this will not work in case of breakout trading, since you must enter at the time breakout is happening for a positive edge, either going in before or after the breakout might cause some bad things for the trade. If you are lucky, you will just lose commissions.

As per the argument that its not possible to reverse-engineer strategies, I can't agree with it. It depends.

Reverse engineering is especially easy if your strategy has an important time of day component. E.g. lets say your strategy trades on an average 65 days in a year within first 15 minutes after 1 hour of ES market opening. So, your entry times would be within 1030 to 1045 EST, on 65 days in a year. I think its trivial to approximate the logic behind this strategy. So, even though the reverse engineer might not be able to pin point exact combinations of rules you are applying, but he will be able to fairly accurate resemble the logic of your code that results into those entries/exits. Now lets say after reverse engineering, he validates his entries/exits against yours for next 20 trades and finds 95% success rate. Also assume this is a fantastic strategy with a PF > 3 and very minimal DD. If your brokerage/clearer has huge amount of money, and this is something really unique, whats going to stop them before they start front-run you with 1000 or 5000 contracts? I say, nothing, if they are sufficiently well capitalized. And this is on ES, what if you are trading TF - much less liquidity. This action of the broker/clearer might also change market dynamics that causes this strategy to be profitable in the first place.

 

[jcl]

Quote from themickey:

It would be very easy to reverse engineer, I have done it. However not illegally. It can be done without too many smarts.

With all due respect, I doubt that you or anyone else can reverse engineer a serious, non-trivial strategy.

But you can prove me wrong. I'll send you a list of trades from a profitable system, and if you manage to reverse engineer the strategy and replicate the trades, I'll pay you $1000. Up to the challenge?

 

[gmst]

Quote from jcl:


Of course this is different when your trade platform directly sends your code to its manufacturer. It's interesting that multichart does that. Maybe that's the reason why multichart crashes so often.

Kindly note that I am not accusing multicharts of any wrong-doing here. But yes I do say that they do send strategies along with the whole dump when they crash. They might be using it only for investigative purposes to sort out the crash thing. But surely, I am concerned. Especially if they do so when the program discovers some very high PF and high reliability strategy.

Maybe one solution is to have 2 computers - do strategy development on 1 computer not connected to internet and use 2nd computer for order execution. Do order execution not using multicharts auto-trading but using broker's API. Yes, you are still susceptible to your broker stealing your strategy, but at least the strategy development platform guys - they can't steal your strategies. Happy to hear what other folks think of this solution, or if they can offer some other suggestion that is practically possible to implement.

Probably another solution is to use .sef files for all the strategies that you have completed. However, it is very time consuming if you want to edit the strategies.

 

[Stan_MultiCharts]

Quote from gmst:


5. your strategy development software:
Now this is interesting. I never thought about it before, but I was doing some work on multicharts yesterday and it crashed and wanted to send an error-report. When I checked the files it was sending, the dump contained all my strategy files. Is it possible that developers of multicharts or any other trading software for example have encoded deep inside the trading software to upload the strategies files to multicharts servers, whenever they find say any trading strategy on ES with over 200 trades over 5 years with a PF > 2, verified by out of sample testing? I am purely hypothesizing here but why can't the developers do this. This way, the software will just send the developers the top strategies of their best traders. If something like this can happen, what counter-measures if any can a trader take?

Aside from stating the obvious - that MultiCharts does not send your strategies - here are a few points on what really happens and why that doesn't make sense for a software developer.

1. You can choose not to send your logs.
2. MultiCharts sends the contents of your Logs folder only.
3. MC doesn't know what symbol you trade, with what broker, and or any specifics about your strategy. Also, there are no servers to handle such information. It's useless without it, even if developers wanted to trade it.
4. There is a reason a company is a software developer and not traders - they are good at making computer programs and not trading. Revenues come from sales of the software, and not from trading itself. There is a reason why engineers don't drive race cars, and professional drivers don't build them or design them. Each is good at their own thing.
5. Reputation of a company is built on traders' trust over decades and can be destroyed in a day - if anything like this was remotely true and the word got out (if for some strange reason some developer was doing this), sales would drop through the floor and potentially put the company out of business. Who in their right mind would jeopardize a stable business model for no gains whatsoever? It doesn't make sense.

 

[gmst]

Quote from sbokov:

Aside from stating the obvious - that MultiCharts does not send your strategies - here are a few points on what really happens and why that doesn't make sense for a software developer.

1. You can choose not to send your logs.
2. MultiCharts sends the contents of your Logs folder only.
3. MC doesn't know what symbol you trade, with what broker, and or any specifics about your strategy. Also, there are no servers to handle such information. It's useless without it, even if developers wanted to trade it.
4. There is a reason a company is a software developer and not traders - they are good at making computer programs and not trading. Revenues come from sales of the software, and not from trading itself. There is a reason why engineers don't drive race cars, and professional drivers don't build them or design them. Each is good at their own thing.
5. Reputation of a company is built on traders' trust over decades and can be destroyed in a day - if anything like this was remotely true and the word got out (if for some strange reason some developer was doing this), sales would drop through the floor and potentially put the company out of business. Who in their right mind would jeopardize a stable business model for no gains whatsoever? It doesn't make sense.

Stan, Thanks for stepping up and making a comment.

1. However, it happened 2-3 days ago and when I clicked to show the file that MC is sending, along with 100 other things, I could find all the files in this directory also that multicharts was sending.

C:\ProgramData\TS Support\MultiCharts\StudyServer\Studies\SrcEl

As you know all these files can be opened up in a notepad and everything revealed. So, I am not sure if you are correct here. Or, if my understanding is flawed, I request you to kindly clarify what it means.

Once I discovered above, I didn't send the logs.

2. I think multicharts does sends the information from quotemanager, so it does send the symbols that I have opened on my platform. Doesn't it send the quotemanager information ?

3. I agree with point 4. and 5. you make. That is why I am not accusing you of anything. Rather say that I am happy if you can put facts here and make all the doubts go away.

Cheers,
GMST

 

[gmst]

Quote from OddTrader:

"Are these really top 10 largest CTAs?"
http://www.elitetrader.com/vb/showthread.php?s=&threadid=213943

My understanding is: All CTAs are using individually/ separately managed accounts for their clients, and All successful CTAs, while managing many billions of dollars for many years, apparently do not worry much about how to protect or hide their trades by any special means.

Others can easily duplicate or analyse their trades without any problems.

I'm just curious what protection is for these CTAs?

Good question. We have come CTAs on this board. I hope they step up to give their point of view.

 

[logic_man]

Quote from gmst:

Randomizing some entries is probably an interesting idea, but it will work only in some circumstances. E.g. this will not work in case of breakout trading, since you must enter at the time breakout is happening for a positive edge, either going in before or after the breakout might cause some bad things for the trade. If you are lucky, you will just lose commissions.

As per the argument that its not possible to reverse-engineer strategies, I can't agree with it. It depends.

Reverse engineering is especially easy if your strategy has an important time of day component. E.g. lets say your strategy trades on an average 65 days in a year within first 15 minutes after 1 hour of ES market opening. So, your entry times would be within 1030 to 1045 EST, on 65 days in a year. I think its trivial to approximate the logic behind this strategy. So, even though the reverse engineer might not be able to pin point exact combinations of rules you are applying, but he will be able to fairly accurate resemble the logic of your code that results into those entries/exits. Now lets say after reverse engineering, he validates his entries/exits against yours for next 20 trades and finds 95% success rate. Also assume this is a fantastic strategy with a PF > 3 and very minimal DD. If your brokerage/clearer has huge amount of money, and this is something really unique, whats going to stop them before they start front-run you with 1000 or 5000 contracts? I say, nothing, if they are sufficiently well capitalized. And this is on ES, what if you are trading TF - much less liquidity. This action of the broker/clearer might also change market dynamics that causes this strategy to be profitable in the first place.

All of these scenarios stack hypothetical upon hypothetical and then draw a conclusion that has zero chance of happening in the real world, yet is logical given the premises.

How about 1 high-profile instance of this happening? Let's stipulate that it "could" happen in the same way my house "could" get struck by a giant meteor in the next 10 minutes, but now let's hear about at least 1 example involving someone who's been profitable at least 1 year in a liquid market who's had their strategy stolen and reverse engineered. If there isn't one, either this is the perfect crime and therefore must happen all the time without anyone ever getting caught, in which case it's one of the greatest conspiracies in market history, or it must just never happen and, hence, no examples to speak of.