What cost should I assume for my backtesting?

Discussion in 'Forex' started by catmango, May 24, 2004.

  1. I'm trying to develop a system for trading the forex and am wondering what number should I plug into my backtesting for total costs (slippage and transaction fees) per round trip. I was thinking 6 pips per RT. Can anyone else provide some insight into what they assume for backtesting?

    btw, 6 pips sucks because it wipes out all of my average estimated profit. :D
  2. rezo_s


    6 pips? why?

    There are retail FX brokers that give guaranteed fills on orders and instant fills on market entries. sperad ios 4-5, so 4-5 is what I would say. There is oanda - I think it offers as little as 1-2 pis spread.
    I didnt hear about any bad fills with them, so it is also an option.

    If you are not talking about mini account, then there is also propfx.com (FXCM) that also offer you 1-3 pips, but they require:

    · $500 monthly fee to access the proprietary software
    · Minimum balance of $25,000
    · Trading minimum of 200 tickets per month
    · Monthly fee waived for traders opening 400 tickets or more per month.

    But if you are only backtesting, then it should be a mini account. I think oanda would be good. Refco/fxcm also ok, but spread is wider. . .

    good luck!
  3. I was suggesting 6 pips for a RT trade, meaning 3 pips per trade. Thanks for the advice on a mini account. Makes sense to me.

    Anyone else have any insight? Does 3 pips per trade sound about right, or should it be closer to 1-2 pips?
  4. Xenia


    Do they still have those requirements ?
  5. Well, there is more than enough variability among spreads, by currency, even within a single MM. To help make your backtesting as realistic as possible, I would set the cost per trade separately, by currency, at whatever MM you plan to trade at. For instance, here are the normal spreads on the majors at Oanda:

    EUR/USD 1.5
    AUD/USD 2
    USD/JPY 2.7
    USD/CHF 3
    GBP/USD 4
    USD/CAD 4

    These apply 99%+ of the time from 8 pm EST Sunday night to ~3 pm EST Friday afternoon, but will widen to 5-7-9 pips and more for a few minutes before and after significant news releases, and to 10 pips on weekends. However, you'd be wise to use the actual spreads in effect at your MM of choice in your backtesting.

    Slippage will not be an issue at all at any MM, unless your system actively trades during news releases. And if it does, then automated backtesting is likely to be unreliable by a factor considerably greater than the normal spread.

    Finally, why are you multiplying by 2? That is incorrect. If the spread on a given pair is, say, 3 pips, that is your cost per RT (not 6).
  6. There are three things that you may have not considered
    more valid on Euro FX (Euro/USD):

    1-Test you ideas on Tick data, instead of 1 minute
    Data, and then even 2 pips slippage on each side
    might be more than enough to account for.
    It also depends what time of day and which currency
    pair you are trading, because spread and slippage varies.

    2-If you are doing lot of trades intraday, you are not going
    to get slippage on every trade, it's more valid to account for
    slippage on every trade if you are doing few trades a day.
    Usually fill slippage happens more on Exiting losing
    trade (not entry losing trade) and entering winning trades
    (not exit winning trade) all because of liquidity reasons.

    3-When we talk about slippage, there are to reasons, either
    stop violation by brokers in spot FX, or illiquid in Globex
    , therefore accounts with PropFx seems to be the best for
    scalping (neither two above problems might exist, if you
    can fund your account with over $25k.
  7. Based on system trading, slippage (not pip spread)
    has to be added to both winning and lossing trades
    and round turn (both entry and exit).