Trading model

V

vats888

Hi all,

The last few years I have developed a trading strategy involving trading index futures. The backtested results were very close to 20% annual net profits assuming 0 leverage.
The last 1.5 years I have tested with real account and the results replicated almost exactly my back tested results and gave me 17% annual profit.

I have been in contact with a friend who works in a hedge fund and asked me for a pnl data series.

I was reluctant to give anything away to begin with but eventually I gave away the total pnl based on the trading of a few index futures as a whole. i.e. the pnl of the basket of futures

He came back with the following statistics

Annualized Return = 23.44%
Annualized Vol = 21.69%
Sharpe Ratio = 1.08
Worst Drawdown = -18.23%
Longest Drawdown = 9 months
Best Month = +27.88%
Worst Month = -16.46%

Next his fund said they were interested in the strategy but want me to give them the pnl per product with entry and exit signals.

This is the point am at and not sure what to do.

Would it be possible for them to deduce the strategy parameters using the entry-exit signals on 10 instruments they want this to be tested from 2000 till today?

My aim is to use the strategy but without giving away my model which is a difficult if not impossible task to accomplish.

My personal funds are limited and this is why am considering to making use of the strategy with someone elses funds for a big payout ratio to the investor and after a few years keep running it for my self.

Any ideas how to go about this are very welcome, but my main concern is if it is possible to deduce a model just by its results and signals it produces.

Thank you for your time reading (and hopefully replying)
 
Quote from vats888:

Would it be possible for them to deduce the strategy parameters using the entry-exit signals on 10 instruments they want this to be tested from 2000 till today?
More...

Does it work on 10 instruments? I doubt it but the answer is yes, they can develop an estimator that replicates the performance exactly. I don't know why they would ever want to do that but they can do it if they wanted to.

Can you enlighten us with some more details about the system performance? Have you traded ES with it? I'm developing a system for ES right at this moment with some special requirements set by a client. Do you have something like a backtest report to show to us? I will tell you whether they are truly interested or playing around with you possibly.
 
Thanks for coming back on this. Please see below my answers to what you are asking

“I doubt it but the answer is yes, they can develop an estimator that replicates the performance exactly.”
So in essence you are saying that they can reproduce the model? Is there a way that you might think I can avoid this?


“Does it work on 10 instruments?”
I do not know which instruments they will want tested on and certainly do not know if this would give similar results.


“I don't know why they would ever want to do that but they can do it if they wanted to.”
Why do you think they would not want to do so?


“Have you traded ES with it?”
The ones I have tested include ES with similar historic record but no I haven’t traded ES futures in a real envrironment.


“Do you have something like a backtest report to show to us?”
Everything was done through excel so would an equity curve of the SP trading suit you ?


“Can you enlighten us with some more details about the system performance?”
What other details of the system performance would you require other than the stats I provided? Am not a maths expert myself and this is the reason I have sent the results to be analyzed by a quant.
 
My recommendation is to offer to trade a small account for them. I would not disclose much more about the strategy than you have posted here. There is not much you can say about your strategy without risking competition. If they are not willing to let you trade for them, then you probably should look elsewhere or manage your own money with it. With a 20% return, your account will grow pretty fast.
 
For what it's worth places like Striker Futures post black box system results of real money account entries and exits. Not sure if you can deduce their systems from those results...perhaps if you don't give them time stamps for the trades. I would still be cautious because it takes alot of hard work to get a model that actually works and for how long into the future it will continue to work is unknown.
 
So I dug this up... I think it should relate.

Quote from logic_man:

Yeah, I would definitely say "no" to divulging the strategy to anyone. I'd trade it on their behalf and if they could reverse engineer it, well, I guess that I couldn't complain because I take that risk by even letting anyone know the entries and exits. Even so, the ES is big enough for two traders trading the same method.

As far as "fitting their business model", I could describe it in general terms like the typical holding period, trade frequency, etc., and that should, for anyone negotiating in good faith, be sufficient information for them to decide if they want to work with me or not.

The main reason I'm interested in working for someone else is that I can access scale that I simply can't get on my own. My estimate is that this strategy could trade hundreds, if not thousands, of ES contracts (whatever the amount is that one can trade without impacting the market's future price action) and I simply don't have that kind of capital laying around. If it also works in other markets, well, you can see that getting even a slice of that kind of action as an employee of a company capable of that kind of diversification would be huge.
More...

Personally, I'd suggest that (if you are intent on giving them information) you find a lawyer to cover your ass, just in case they try anything underhanded.
 
captain-obvious.jpg
 
Quote from vats888:

Hi all,

Would it be possible for them to deduce the strategy parameters using the entry-exit signals on 10 instruments they want this to be tested from 2000 till today?

My aim is to use the strategy but without giving away my model which is a difficult if not impossible task to accomplish.

My personal funds are limited and this is why am considering to making use of the strategy with someone elses funds for a big payout ratio to the investor and after a few years keep running it for my self.

Any ideas how to go about this are very welcome, but my main concern is if it is possible to deduce a model just by its results and signals it produces.

Thank you for your time reading (and hopefully replying)
More...
No. It is a total bullshit urban legend that anybody can deduce a trading strategy just from looking at entries and exits -- unless the strat is something really simple-minded like channel breakouts.. Anything they can mock up from just entry/exit info will quickly diverge from the real thing once they put it into action.

This myth reminds me of all the so-called code cracks for proprietary technical indicators that live on certain websites. Almost none of that stuff is anything like the original blackboxes.
 
You must log in or register to reply here.
Back
Top