Why would I do that? Because you expected a handout and didn't get it? Boo-hoo
Protecting your strategies
- Thread starter gmst
- Start date
Why would I do that? Because you expected a handout and didn't get it? Boo-hoo
I don't think anyone here ever suggested a limitation of 1-candlestick chart. It seems that everyone on this thread who wants to believe that a system can be copied must always fall back on some limiting assumption. As another poster suggested, omitted variable bias.
In the example of copying a trader's breakout system, the only reason that was possible was because you "knew" that he only traded one chart. Without that piece of info, there is no way to know which other variables cause him to buy that specific breakout bar.
We are unable to reproduce what you reported. Please come to our Live Chat (through our site) between 6:30 am and 4 pm EST to show you did that.
Quote from achilles28:
Good points. But how exactly does one align IT compensation with performance, to ensure retention? Imagine a strategy that does 100%, a year. And the few programmers who run it make 250K a year. Whatever bonus the firm could offer would be arithmetic, juxtaposed to (theoretically) a geometric rate of return. In layspeak, its always more profitable for the IT guy to steal, "lose", transfer or sell the strategy. But like the other dude mentioned - nothing ventured, nothing gained. Sometimes its better to risk diversion and theft, to kill it for a few years... This part bugs the heck outta me. On the bright side, a few technical strategies are well-known, and the market doesn't change. Even if the classical edges are passed around like a mexican whore, it's not the end of the world...
Key individuals should always have at least profit sharing, if not an equity stake in the company. Think about it this way. If the programmer were to sell the code, he has two options. One time payment, or continual payment. There is no way that it could ever be more profitable for him to sell it for a one time payment. Whomever he sells it to will have to agree to pay him in perpetuity which is what he should already be getting at your company.
You also seemed to miss the other point. They must be allowed to participate in the program for free. IOW, they are making a decent salary, sharing in the revenue from OPM, and allowed to put their own money to work in the program and realize geometric returns without paying fees. If you've selected a good employee in the first place, that will be enough to keep them loyal.
Most (at least 95%) of traders on ET lose money. Why would you ever want to reverse-engineer a losing strat? Of the ones who make money, their system is either much more complex than 1-chart, or if they are trading a single chart they are most likely at least 50% discretionary, which is even harder to reverse-engineer.
Some good contributions on this thread so far. Thanks everyone. Keep it coming. Just don't make it into a thread - "whether reverse-engineering is possible or not".
I discovered 2 more threads where this topic has been discussed before:
1) How can I protect my own trading system's source code (C#)
http://www.elitetrader.com/vb/showthread.php?threadid=234507
2) How to prevent brokers from knowing your strategy?
http://www.elitetrader.com/vb/showthread.php?s=&threadid=220599&perpage=20&pagenumber=1
This thread started out well but quickly turned into a slugfest. So, ignore final 2 pages.
I discovered 2 more threads where this topic has been discussed before:
1) How can I protect my own trading system's source code (C#)
http://www.elitetrader.com/vb/showthread.php?threadid=234507
2) How to prevent brokers from knowing your strategy?
http://www.elitetrader.com/vb/showthread.php?s=&threadid=220599&perpage=20&pagenumber=1
This thread started out well but quickly turned into a slugfest. So, ignore final 2 pages.
Sometimes there's a fair amount of information that can be inferred, depending on the nature of the strategy.
If say, some algo is trading single outrights, it's going to be hard to reverse engineer, as you don't see the model, or the hidden states, you just see the output.
But, if it's multi-legged spreading, the reverse engineering job is a bit simpler. More of the moving parts are displayed as the strategy gets hit, and offsets. Simultaneous hedge legs can give it away.
2c.
If say, some algo is trading single outrights, it's going to be hard to reverse engineer, as you don't see the model, or the hidden states, you just see the output.
But, if it's multi-legged spreading, the reverse engineering job is a bit simpler. More of the moving parts are displayed as the strategy gets hit, and offsets. Simultaneous hedge legs can give it away.
2c.
Honestly, I think the biggest threat is outright theft - either via employees, or outside hackers. The broker angle, again, despite what JCL says about the impossibility of it all, I think is valid. If a trader is turning 100-200% a year on a seven figure account, sure as shit, they'll sniff around. Why wouldn't they? In this day and age, where ethics are nonexistent, and banks own regulators, who cares? It's worth a shot...