Robot Targeting GBP/USD

Discussion in 'Strategy Development' started by rttrader11, Nov 8, 2011.

  1. In May this year I placed a “Robot FX Trading Strategy” on Ebay only to find that you cannot sell to the International market unless you have previously sold items on ebay.
    The advert sat on ebay for 1 month but I had no enquires from within the UK and so let the advert expire.
    I was still surprised that no one here in the UK enquired about the Robot and a WebMate suggested I join some forum and ask its members what their assessment of the Robot was as this would help me gauge demand for such things. The Robot is not up for sale anywhere on the web these days but I am interested to know how many people are interested in such things and if people are interested in long term strategies or only interested in fast (get rich quick) strategies. This will help me determine how to move forward with this.

    My webmate is a day trader, where the Robots approach is long term and he thinks there is no interest in Long term strategies as people want things today that are Fast like their food he says. He also thinks that few even know how to read these back test reports and that Tradestation is not that popular in the UK and is mainly used by traders in the states and Canada (is that true?).

    So I have attached the full back test for you all to view and those with experience of Tradestation reports will need to know the following:
    £30 commissions factor in.
    No Profit Target
    For those with no experience of reading back tests I offer the quick guide below:

    Any Automated Trading Strategy can be Back Tested but if the Test is not run over the Entire Data Set (Market History) then it is not worth as much. You could easy make a Robot that looked Brilliant over the last 2 or 3 years but then if somebody started using it for real, they could get a very quick and nasty shock (wipeout). So the minimum you should look for is the FULL Complete Performance Report. (This was at the time of making it in April this year)
    Then before you look at the report you should look at a chart (rt1.jpg) of the Market on which it was run so as to get an idea of the Market conditions the Robot had to deal with. If the Market went straight UP or Straight Down over the testing period then it would not be hard to Trade that Market well, but what happens when the Market Turns or goes Sideways in the future.

    First you should look at the settings page to see what the Testing Period was "Start Date/End Date". Note that TradeStation allows for back testing from October 2002, Other platforms will be different.
    Then look for "Commissions", on this report you will see it is $15 (Thats $15 in and $15 out = $30 per Trade {The Spread})
    If NO Commissions are factored in then be careful especially if the Robot makes a lot of trades. Remember, Lots of Trades = Lots of Risk.
    "MaxBars Back" means how many Bars of DATA the Robot needs to make its first calculation. Here you see it set to 50 which is the default setting and tells you that 50 Bars of
    the Highest Data Stream (Day) are needed before the Robot will START looking for a Trade. Note, normally many data streams are used, this Robot uses 5 Data Streams {3min/15/30/60/Day}.
    Robots run at the end of the BAR on a BAR Chart so look to confirm "Look Inside Bar Back Testing" is Disabled other wise the Back Test may be Flawed as the Back Test will assume it got in and out at the price it wanted too but in reality the Robot would not execute the trade until the bar closes and 180 seconds (3mins) can be a very long time in the FX markets.

    "Initial Capital" is set to $100,000 which is the default setting for TradeStation and in the Trade list you will see this in the "Shares/Ctrts" column 100000 but the Robot is only Trading 1 Contract and this can be seen by the profit and loss shown for each Trade ( 1 pip = $10) so (100000 = 1 Contract).
    Trades 1 contract GBP/USD Spot (1 Contract = Price X 1000 {$1600}, using TS 100/1 margin)
    1 pip =$10
    I would trade this Strategy with $10,000 Compounding at 10 so when the Robot had made $10k profit (Trade 52) I would start trading 2 Contracts.

    Now you are ready to read the report. Go to the TRADE LIST page ( The Most Important page) Start at Trade 1 and work your way through the trades, this way you will get a feel for the Robot and will be able to see all the details of every trade.

    1 Buy 11/07/02 10:33 Buy $1.56540 $0.00 100000 $1,270.00 0.83% $3,110.00 94.53% $15.00000 $0.00000
    Sell 11/15/02 04:33 SellShort $1.57840 $1,300.00 $1,270.00 ($180.00) 44.98% 39.51% $15.00000 $0.00000

    So Trade 1 was a BUY @ $1.56540 and the Trade was $180 Underwater at its worst point and $3110 UP at its best point and in the end it took $1300, minus Commissions $30 = $1270 profit and the trade lasted a week.

    Note that this Robot has a Stop Loss setting of $400 and a BreakEven setting of $400 so Losing Trades cost $400 + $30 and a Break Even trades actually lose $30 (Commission).

    On the Performance Summery page there are 3 key things to look for. First the Profit Factor {top right}
    Next look for Largest Losing Trade.
    Then Max. Drawdown (Trade Close to Trade Close) this tells you the worst losing run of Trades.

    I hope that helps newcomers to Tradestation performance reports and remember to let me know what you think of the Robot. It was emailed to some Uk funds that had lost nearly all their Investors money but they showed no interest, I Have my conclusions on why but what’s yours ?

  2. zip attached
      File size:
      573.2 KB
  3. Fairly low profitability PA. Only winning trades are short. Fairly low sample for backtesting due to longer hold times (not scalping robot). Looks like the last few months have started underperforming. Perhaps the market isn't fitting the robot any more. Jumps in equity curve very unreliable, hints at greater risk of drawdown. The yearly results indicate that it fits only certain market conditions, and probability of net 0% is quite high.

    Read up more on Expert Advisors. There are forums out there (some private, pay per view) that are littered with them. They have backtesting results which go up to 1000%+ per year. In reality though, they don't work so well. Why? Because of backtesting methodolgy above all else. The backtests often take into account only middle prices, no spikes, no actual tick data, no spread reduction, no bad data, no slippage, etc. The real world is quite different. Sadly for you, I doubt you can sell this robot to anyone, but it's a good start and you may get somewhere some day.

    Have a look at Collective2, they have a bunch of "robots" there that generate signals. Some are quite cheap and have been running for a while, but not much intrest. The "get rich quick" ones die the fastest. What you need is a proven track record, a system you have used for over a year, and you have an audit trail in a real money account to prove it was profitable. Then, you could possibly sell it.