Let’s say we have got a horse racing strategy executed automatically by betfair bot strategy. The only information we have got are messages displayed by trading bot in Output view.
What if we want to analyze this strategy learning what is the best profit loss target, what entry point is the best one, or how our nominated selection could compare to the actual winner of the race.
In my short video I present such solution. To my machine learning horse racing strategy I added code to save all traded prices and volume for nominated selection and for the actual winner selection, when nominated selection is not the winner.
Bfexplorer supports so called “Data Providers”, so your code offering data and of course displaying them in some way. In my case I used charts displayed in web browser.
So what I could learn afterwards from all these data?
In the one race where my machine learning (ML) strategy triggered trading on the horse Loveherandleaveher, the market was open 5 minutes before official race start time and the opening price was 5.6.
The trading bot opened its position backing at 5.8, the target profit is 4 ticks, so bot placed lay hedge bet at 5.4, and the bet was matched at 16:54:48, so 12 seconds before start time.
During the time the trading bot was active on the market, the maximal trading price was 6.4. Well ML model nominated the second favourite, the favourite (Everlanes) actually won the race, but again from our data analyzing we can see that before the race start time, the favourite price went up so was not tradable, but our ML selection was.
The in-play data says that our ML selection went minimal to 2.8 at 16:58:55 and that time the race favourite was traded at 5.60.