Automated Forex System - Work In Progress

Discussion in 'Forex' started by FNHinVA, Jun 16, 2008.

  1. FNHinVA


    Stone Newbie With A Few Questions...

    First, I have never traded (other than marbles and baseball cards when I was a kid). So please be gentle.

    Some background: I have experience in financial modeling and quant methods. I also have 30+ years of programming experience (off and on). I have read a bunch of books and follow a few websites but that is the extent of my "training" in currency trading.

    I am working on a system for day trading EURUSD. I have built an Excel-based trading model and have been working with 3+ years of data. The system uses an "always in / reversing" approach (my terminology - I just made it up). I am optimizing on 18 months of data and forward testing on another 18 months of out-of-sample data. If it continues to look promising - as it does currently - I intend to automate it and paper trade it next. I don't know what to assume for transaction costs and slippage. This is, of course, rather critical to a system of this nature.

    My questions:

    First, I want to make sure my understanding of how the "spread" translates into a trading cost is correct. (Commission and slippage aside.)

    If, for example, a currency pair is quoted at 90/95 and I buy, I get it at 95. If it remains unchanged for some period of time and I sell, I get 90. I have just "paid" five pips for the trade. According to tech support at my data supplier, all Forex data are mid-point relative to bid/ask. So this transaction would show in my data as both a buy and a sell at 92.5 and I would add five pips as the transaction cost. Correct?

    I have seen some recent data that shows EURUSD with an average spread of .9 pips. But what is a realistic assumption for longer term simulated trading - assuming a NDD or pseudo-NDD with variable spreads and a commission?

    Also, I am seeing commissions of around $2.20 to $2.50 per contract. Is that round turn or on both ends?

    What is a realistic assumption for slippage per round turn?

    Related to the above is choice of broker; I am looking for a good combination of low trading costs, good execution AND a suitable platform for automation. But then, who isn't?

    At present, I am looking at IB and Tradestation. I may wind up using the TS platform and the IB brokerage. (Importantly, both have "live" paper trading capability.) The associated cost of using the TS platform without their brokerage is not an issue.

    Any comments on the suitability of those two? Others?

    Many thanks for any guidance on these issues.

  2. Realist


    Don't take this response the wrong way but all I can recommend is that you do your own due diligence. If you are coming to a message board to find these answers then you probably should stay away from automated trading on the FOREX. Especially if you have never traded real money on FOREX before. Your account will be eaten alive if you don't completely understand each and every dynamic of the currency market. Paper trading and live trading are two completely different domains. I hope you understand this. Good luck in your endeavor :)
  3. FNHinVA


    Don't take this response the wrong way but I am not the least bit concerned about having a paper-trading account eaten alive. That will be part of my due diligence.

    I am not going to put any money at risk before I learn what I need to know. Then it will be small and incremental. I understand risk control even if I don't know much else.

    But thank you for your concern.