Ray Robot 2 Makes Money Trading Breakouts with MetaTrader 4

[Jim]

Following some festive Halloween fun over on the AlgoTrader development group a new "robot" was conceived. After a remarkably brief gestation period the MetaTrader version of the son of Ray the Random Robot has now been born!

Ray Robot II has learned something from his father's experiences, and uses slightly more intelligent entries. RR2 waits for the end of a bar, then places a pair of pending orders at or just beyond the high and low of that bar. These "bracket orders" can either be stops, in which case the son of Ray hopes for a breakout, or they can be limit orders, in which case he is hoping for some "mean reversion" to take place.

Ray Robot the second has also now seen his father's equity curve, and decided to ditch Martingale money management in favour of the far more sensible methods used by our "London Breakout Robot". He still uses his father's original exit method though, which consists of a fixed profit target and stop loss.

Here's the source code for our latest creation, together with RR2's very first backtest results:

RayRobot2_v100.zip

This is what a few month's worth of breakout trades from daily bars of EUR/USD on an Alpari UK demo account looks like:

RR2-EU-D1-H2-2011.jpg

Not bad pocket money for such a young whippersnapper?

P.S. Here's the latest version:

RayRobot2_v101.zip

[Laurence]

Hi Jim,
Thanks for that, will have a play

Laurence

[Laurence]

http://screencast.com/t/czJgujD2
1hr Chart since 1st July

[Laurence]

http://screencast.com/t/aaF2wWWlEQy
DAX!!! 1hr, TP = 20, SL = 10

oops - ignore that - dp error, that's using TP = 2, SL = 1!

[Laurence]

http://screencast.com/t/Ab8P0BRgm

DAX Optimised at 800 TP (80 pips), 200 SL (20 pips) using GKFX to run tests

[Laurence]

http://screencast.com/t/i0PCjlbM

And just for fun.... Final optimisation on Dax 1hr suggests that a stop loss of 1 pip & a target of 90 pips will bring wealth beyond your wildest dreams...
Could Ray be the Holy Grail disguised as a donkey?

[Laurence]

http://screencast.com/t/x5Y3yf6vYPce

5 mins, 7pip TP, 1 pip SL,- seems to be the optimum set-up on GKFX back tester.
Will run on demo from Mon with MM type 1, 5% risk and see what happens.
I suspect GKFX would have a sense of humour failure running an ea with these settings on a live account as it probably falls into their definition of scalping...
Certainly an interesting experiment, Dow & FTSE also move in a similar manner and have 1 pip spreads (Dow is 3 outside US open) on GKFX so should give similar results

[Jim]

Hi Laurence,

Blimey, you've been working hard! Thanks for all those screenshots, and the "interesting" tables of backtest results too. As you know, Ray is a robot of very little brain. His son is a bit brighter, and makes some attempt to call a pip a pip on FX. His efforts in that direction haven't got as far as indices yet I'm afraid.

Ray undoubtedly looks just like a donkey, though the countenance of Ray the 2nd has yet to be revealed! I fear neither of them look anything like "The Holy Grail" however. When I have a spare moment or two I'll port RR2 to Dukascopy and backtest him on tick data. I have a sneaking suspicion that those impressive looking results with very tight stops are an artefact of MetaTrader 4's less than accurate backtester, and won't be repeatable in "real life". Nonetheless I very much look forward to seeing your "forward test" results!

I'll be doing some similar tests starting on Monday also. Following the recent announcement from Alpari I figured comparing their spread bet platform with GKFX's would be very enlightening, so I think I'll let the son of Ray start playing with modest amounts of real money at both brokers. Those experiments will be limited to FX of course, since that's all Alpari have to offer spreadbetting wise at present.

Cheers,

Jim

[Laurence]

Agreed, a 1 pip (all right, point as we're talking indices) is an unusual stop loss value!
I was assuming that FTSE / Dow would give similar looking results but not at all, they nose dive like the proverbial flying piano on the short tf's I tried so will be interested to see how demo works out on DAX.
I will alter the magic numbers and run on 5, 15, 1hr & daily tf's concurrently and see how they perform

[Laurence]

1st observations;
GKFX demo is only 4 dp's so there is an immediate variation to the backtests performed on the live 5dp platform
'Trade context is busy' prevents 2 or more concurrently running versions of Ray placing simultaneous orders so I am just running one chart on 5 mins.
That seems to be placing orders ok - will ignore now for the rest of the day & see what it does...