Taking proprietary strategies from discretionary to systematic......

Discussion in 'Risk Management' started by Arcap, Sep 22, 2009.

  1. Arcap


    Anybody out there have any useful input (read ridicule) on how to slowly gravitate from a discretionary trader with systematic guidelines into a more systematic nature...without having that feeling that you have just pulled your pants down to the world revealing 15 years of work in blueprint form to a developer??? I am not saying that I have the goose that lays the golden egg or anything, but it is an edge that was accrued over 15+ years of sweat, tears, learning new ways to string cuss words together etc. I have recently been "missing" far more trades than usual, beyond statistical liklihood, so I am already a bit paranoid about it. Fortunately, I have room to give, but I do not like the trend. I guess what I am asking is has anybody been in a similar position and made that jump? Is it worth it? I have never been in the environment where I was around quality developers so I don't know what I am missing. Quick broad strokes..... trade frequency is approximately three per week with holding periods of a few hours to a couple of days, mostly stock index futures. Thanks in advance-
  2. I'm not sure I understand your problem. Do you miss trades because you do not trust your method or do you get more losers than you expected lately?
  3. i think he's trying to say he doesn't want to reveal the system to the coder?

    I agree...they'll claim they'll sign non-disclosure, etc....but it's B*S*. If it is good they'll sell it off to third parties or use for them-selves one way or another.

    You'll have to search high and low for a programmer who knows nothing about trading. See him/her face to face...believe no-one on the net.
  4. oh yes it is worth it......some people claim you cannot code systems....That's rubbish.

    It's worth puruing as it will give you more free time. Your job is to manage the system etc..not watch each little bar go up and down all day long.
  5. There is certainly a trade-off with implications. On one hand the coder must be familiar with trading to code a system well. On the other hand, I think anyone with trading experience will recognize a novel idea and use it, at least.

    The solution: Either learn how to code yourself or stay with manual execution and work on your discipline. There is no middle way to protect your idea.
  6. NDAs aren't worth the paper they are written on.

    If you were a software developer and someone provided with with some ideas for implementation that are profitable, what would you do?

    I think that answers your question.
    You're better off learning how to code - try learning some Python, it is not a particularly difficult language to pick up.
  7. maler


    Divide your process in several independent components.
    The data gathering and raw processing you can outsource (outsource the processing of different data sets to different people).
    At the center of these raw ingredients will sit a (usually small) program that more or less defines the logic of your buy and sell decisions (alpha generation process). That you should never outsource.
    Once you have your alphas you generate your trades based on some risk management scheme(examples range from simple fractional betting sizing to optimization portfolio construction). The shorter you holding period the less sophisticated you can be.
    This you can outsource or use third party providers of risk models.
    Once you know what, when and how much to buy/sell, you feed
    that information into an execution engine. You can get outside help for building an execution engine or programming an existing one.

    So my advice is that you can outsource the development of noncritical pieces to several different people. Do not underestimate the ability of people to reverse engineer what you are doing. Only the paranoid survive in this business.
  8. minmike


    Using index, TT has a couple of products that might help.

    If you can translate your ideas into if then statements on a piece of paper, then just translate the paper onto TT Autotrader. Really easy for someone with working knowledge of excel. Easiest that I have found.
  9. Arcap


    Sorry, that was not very clear. I have been paranoid due to a level of price competition in front of me that was previously not as intense. I went on one stretch in July where I missed 4-5 trades over a two week period by a total of .75 on the ES. Two traded at my price without a fill (I know, seems criminal to me too and it was investigated thoroughly), one missed by a quarter point and one missed by a half. And these are trades with a 6-8 handle stop and 10-20 handle type profit, not hyperscalping. Sure, two of those were outside of normal trading hours, but they were during early European hours where activity picks up. It has cooled off a little bit, but I am fairly sure it is similar logic. I run about 5 million, roughly half dedicated to this particular strategy so liquidity is not an issue, even overnight for the most part. I use little to no leverage on a notional basis. One shift I have made is trading some of the foriegn stock index markets of lesser liquidity and therefore lower institutional competition. The numbers there are great, but they do not scale. Generating alpha with 5-10 million is nowhere near the battle of going into the 50-100 million shark pool. That is another debate I am having.....how much is enough asset wise? Smaller asset base and great numbers or larger asset base and lower numbers. The math is easy, but not all in life is quant. Sorry to stray off topic, that is probably a discussion for another thread.
  10. Arcap


    That was my line of thinking...albeit without the level of detail. Very interesting. Thanks for the advice, this is very helpful.
    #10     Sep 24, 2009