how to protect trading strategies for a colocated server

Discussion in 'Automated Trading' started by trend2009, Dec 9, 2009.

  1. Jerry030

    Jerry030

    OK, thanks.

    My two actual Questions/Points are simply put:

    1) In your estimation the maximum value of a 100% mechanical trading system, regardless of implementation, expressed any way you like with two key components:

    A) Monetary dimension: a number of any kind you like expressing increase in value, capital, ROI, ROR or any other metric you care to use. If using a non-standard metric please explain the formula for calculation. This value would be something you believe impossible to realistically exceed and if demonstrated live would cause you to ask "how in the heck is that possible?"

    B) Chronological Dimension: Per hours, day, week, month, year, decade or anything you like.

    2) Do you mean that almost all strategies which as you say are "variations on several well known themes" are that because are ni others that could be called exceptionally profitable or because those that are exceptionally profitable are never revealed publically?

    I believe they exist but of course there would be no logical reason to ever reveal them. In general what gets revealed is where the profit to be made from the revelation as in managing money, selling software, books, and lectures is significant in relation the actual rate of return. For example if I think I can make X return per year trading my strategy but my editor or business associates feel it could generate 3x return if marketed correctly I will reveal it all over the place, endlessly. (assuming my goal is to make money)

    In contrast these outstandingly superior systems that are never revealed, lets call them "Dark Systems", like dark matter they are invisible to the naked eye/public, would not likely be found among the big boys. Since implementation in a large company requires a multitude of folks: strategists, analysts, programmers, testers and network admin folks, keeping it secret would be almost impossible.

    Even if you are well paid and have signed a ton of confidentiality agreements, programmers love to play god and flaunt the rules for the fun of it. Add the potential to make a great deal of money where nobody can prove you took the means to do so and you have a sure fire flaw in the development of an exceptional system in a large organization. Any such system in that environment is not going to remain proprietary or unhacked, human nature being what it is.



     
    #21     Dec 9, 2009
  2. you are correct my strategy is not scalable, otherwise, I will be rich than buffet :). I am fully aware of the risk and thus I always control the trading activity based on the market volatility, and only trade large stocks. On the other hand, I also try to maintain market neutral. even if I get 100 signals, if they are all long, I will not take all of them to maintain neutral. that could be another reason I achieved low drawdown.

     
    #22     Dec 9, 2009
  3. sjfan

    sjfan

    Your first question is misguided: it's somewhat non-nonsensical to ask what's the maximum return of a strategy. It should be what's the max return for some amount of risk. In that, I don't personally mechanical systems lie on a higher or lower risk/reward frontier than any other strategy.

    On your second point: by several well known themes I mean that there are only several different ways to make "legitimate" money in finance (this should be attributed to someone, but I forget who exactly said this to me first): take directional risk, taking relative value risk (long/short), flow trading (market making, dealers, etc), and true arbitrage (a lot of arb strategies are actually relative value strategies).

    Everything else is either a misunderstanding of risk (ie, your risk-less arb trade is really a relative value trade) or some specific implementation of one of the above.

    So, in short, I don't believe there are secret highly profitable never before seen strategies. Finance doesn't work like that.

     
    #23     Dec 9, 2009
  4. sjfan

    sjfan

    "reason I achieved low drawdown. " Realized risk is not the same as true risk.

    Anyway, I'm not trying to shit on your strategy. I'm glad it's making you cash. I've just been around this business long enough to cringe whenever I hear someone saying that they got a low risk high return strategy (and I've heard that enough times). Inevitably, true risk and reward catches up to them in a humbling way.


     
    #24     Dec 9, 2009
  5. you need to trust your guys. i don't know anyone who can come up with the trade and manage the IT/Security. People are paranoid about this all the way to the top - if you have a strategy that you are making $10mm++ (as stated by OP) you will be smart enough to hire a programmer that you trust and pay him/her enough that they won't run to the next shop.
     
    #25     Dec 9, 2009
  6. Jerry030

    Jerry030

    How much is enough? If you knew this disk contained code that has the capacity to generate $10mm in wealth and it's impossible to know that it has been copied by yourself and put to use generating money, how much woudl buy your honesty? The honesty of all programmers?
     
    #26     Dec 9, 2009
  7. if you have code that profitable you either write yourself or you pay them enough. if you have to ask how much is enough obviously you don't have code that can produce $10mm++/year.

    http://www.reuters.com/article/idUSTRE5B114220091202

    get over it - you aren't smart enough to reinvent the wheel and if you are your programmers aren't worth enough to bail on you.
     
    #27     Dec 10, 2009
  8. Jerry030

    Jerry030

    Quote from Jerry030:

    How much is enough? If you knew this disk contained code that has the capacity to generate $10mm in wealth and it's impossible to know that it has been copied by yourself and put to use generating money, how much woudl buy your honesty? The honesty of all programmers?


    Sorry if I wasn't clear in my intent.

    The question was really rhetorical. Having been in IT for over 30 years as everything from a programmer to an IT director I already knew the answer. Guess I was trying to be courteous and let you see the flaw in your statement about having to trust your programmer. If you think someone is going to pass up walking away with the code and design of a $10mm system for the cost of their labor say $50 to $100k, you don't actually know many professional programmers.

    I'd trust no one EVER with anything of truly exceptional proven performance characteristics. As you said you can write it yourself or more practically for those lacking the skill and the time to write high level software you can design and write it as a binary or trinary application: split it into 2 ore 3 parts that run independently and separately with the actual results only created when the part are joined. The join takes place physically and logically separate from the apps that create them, which are also separate from each other.
     
    #28     Dec 10, 2009
  9. Jerry,

    Since you are veteran programmer, possibly you have idea how to load an encrypted code into memory and decrypted it there and execute without writing temporary file on hard disk first. Is that possible and how difficult to do it?

    I guess this could be one of the best ways to protect the code. As one of the poster mentioned previously, you can use the first application to load the second application and encrypt the second one and run it in memory. the password for the decryption could be dynamic sent to the first application each morning.
     
    #29     Dec 10, 2009
  10. CGNobody

    CGNobody

    password could simply be hand entered but beware key loggers.

    As I said, bitlocker or some such software should provide the same level of security unless there is a flaw in their design.

    If you really need the technology for encryption you can PM me, I can refer you to a company that does it for .Net software. Not sure how hard it would be for something else then .Net and once more the folder encryption software out there already handles that probably well.
     
    #30     Dec 10, 2009