Slippage in your automated trading platforms and other questions.

Discussion in 'Automated Trading' started by GRT00, Nov 30, 2008.

  1. GRT00


    Hello to all,

    I´been manually trading futures for some time (with X-Trader) and I normally can achieve like 70-80% of what my trading system stadistics really represent. I mean, I run my trading system on a charting program and I can view what is the amount of money that my system could achieve if I could perfectly fill the signals the system sends to me. But volatility and other factors makes that may be I can´t buy/sell at the ticks my system says, so I don´t have the exact profit/losses that my system could achieve.

    So, based in your experience with automated trading plaforms I would like to ask you 3 questions for those that use them for intraday trading (meaning that you have more than 10-20 trades everyday):

    1) What is normally the difference (in %) between what your trading systems says and what the automated trading platform can achieve?. Do you for example win or loss like 80% of what your system will do if it could perfectly fill all the signals at the exact ticks?. How many ticks do you normally lose with your trading platform?.

    2) Do you think that your automated trading platform is a better choice than if you were trading manually in terms of slippage or do you use it just for convenience and to forgot the stress of having to deal with the markets everyday (despite you lossing more money with the automated trading plaform than doing it yourself)?.

    3) Does your automated trading platform gives too many mistakes, instability and other problems that makes you to have to be all the day looking at the screen checking it out that there are no problems (like the automated trading platform is holding an open position when in fact it has had to be a closed position, making you losing a lot of money) or you trust 100% on its stability and effectiveness andyou can go all the day out without worrying about it?.

    It could also be good if you say what is the automated trading platform you are using.

    Thank you!

  2. rosy2


    1) we put in limit orders only. so its a matter of where in the order queue we are

    2) not sure what you mean by slippage in this sense. all our orders are limits.

    3) there's people that watch it all the time. no one lets there system run unmanned.
  3. 1. You have to be testing your models or systems considering those slippage on Stop/Market orders or non/partial fills for Limit orders. Of course, there can never be a testing engine which can simulate "EXACTLY" what the market may have done, but you can put in measure to put them into consideration.

    2. You're basically saying that you haven't coded your order entry logic. You have the signal that serves as a basis of going long/short... and now, if a performance difference between manual and your "test result" is large enough, it's obvious that you're missing a large component of your "system".

    3. ALL MY SYSTEMS act as they should, of course not in terms of monetary performance because that's controlled by the market, not the developer. There's alot of things that can't be controlled by a computer. I never leave my intraday systems unattended. There's no easy money... let go of all the shit you have in your head about an ATM machine. Playing golf or hitting the beach while you have the "system" make money for you.

    To put everything into perspective....This happens with all discretionary traders who moves to systematic trading. You need to map, plan, design out your trading rules and actions in detail.

    The system and the computer can only do what you tell it to do. If it's acting weird, then you're not telling the computer / system to do it correctly. It's not a matter of "trusting a system or computer"...


    Stop blaming other things and start looking at yourself....

  4. RedRat



    1) you test your system based on historical data and usually you simultaneously adjust internal system parameters. That leads to curvefitting. There should be low number of internal variables and large number of historical deals to trust into the system. And don't forget to test the system based on out-of-sample data.

    Then you take a look on the current market, is it the same as the historical data you tested system with? I am sure current market is totally different :(. My own system lost during current market conditions, partly from increased volatility and leveraging.

    I think 80% of historical performance is good result.

    2) I can't trade manually and fully automated system is the only way for me. Ideally I want my system to run without intervenience. I do not have time to look at it. This year I took 2 weeks vacation and did not connect to the server, the system was running, everything was fine.

    3) That depends upon multiple factors.
    a) trading platform and API
    b) internet connection
    c) brokerage outages
    d) mistakes in your own code, so on

    I can tell that with InteractiveBrokers I met with very serious errors in brokerage API, while updating new version of TWS.

    Now I am with Velocity Futures over FIX protocol and it is very stable.

  5. GRT00


    Thank you RedRat!. What is your current automated trading platform?. Did you try any other automated trading platform before using your current one?. Did you have to solve many problems until you could do what you said about being out for days without having to take a look at it?.

    In which Exchange do you trade and how many trades do you usually do in one day?.

  6. Craig66


    You are an utterly charmless man aren't you? People have to start somewhere, and that usually involves asking a few misguided questions, I would suggest getting over yourself.
  7. zxTrader



    It certainly is possible to run an auto system without having to baby sit. However that depends on its design... Not easy but not impossible...

    The more intraday trades you do the more effect slippages will have on your results.. Which may cause a variation from what your system achieves...

    You may want to look at reducing the number of trades generated to try to match the performance generated by your auto system. of course that does not equal a profitable system. But it will give you the confidence that once you have the stable strategy in place your system will do the rest. Happy trading

  8. RedRat



    As I wrote, I am using Velocity Futures as a brokerage and my API is the raw FIX protocol. I am using QuickFix library so I do not parse FIX messages myself. Velocity allows you to trade only futures contracts.

    As a programmer I have an experience with multiple platforms like StrategyRunner, InteractiveBrokers, Genesis Laser, Assent and finally FIX.

    It depends upon your requirements. If you need to trade stocks then you should go with InteractiveBrokers or Genesis. Most of platforms are stable, for me FIX is the best. There can be nuances about implementation, I do not remember all of them.

    Basically you need:
    -enter stop/limit/market order
    -monitor your existing positions on the market
    -may be monitor your existing orders on the market

    Those days when I had a vacation I traded Russell2000 e-mini futures, ticker = ER2, CME. And I made 100-200 rounds daily.