Three weeks ago Gary traded successfully for the week. Two weeks ago, he gave back all the gains plus some more, so I took it off my live account until I had time to do some trade analysis.
I have looked at the trades from the losing week; my comments and pics of the charts are in the attached pdf.
The thing that became screamingly obvious is that most of the losers were taken under conditions that a manual trader would avoid, and which mostly become obvious only by zooming out and looking at the wider period of time than just a few charts:
- trades taken with the market too far from the 240 trend-detection MA failed through 'snap-back'. The next release will have a distance-from-MA chart display, and an input to disable trading when the market is x pips from the MA.
- We do not need zooming out to see the distance from market to MA, but what follows does need zooming:
- GBPAUD and GU were taken both far too far from the MA, and in the teeth of a howling gale of upcoming resistance. The GU was taken not far from a 6 year high - and we want a big retrace or a lot of faith in recent economic news from the UK to go long on this pair.
- there was an AUDCAD sell taken after a series of higher highs and lows, and a CADJPY buy taken after a much clearer set of lower highs and lows.
Early days yet, but here are my initial conclusions:
- trading Gary cannot be left entirely to the bot. As well as not trading too far from the trend MA, we must zoom out to see the bigger picture and:
- disallow long trading when the market is close to significant resistance, or short when close to significang support.
- disallow long trading at the top of a range, or short at the bottom.
- disallow long trades after a series of lower hilo's, and short trades after a series of higher hilo's.
