Ive been testing some automation since the beginning of december, on in-play horse racing markets. It identifies early support in the in-play market and backs horses, based on the backing liquidity forcing the price down to a % of bfsp. It has left me puzzled and i'm hoping for some thoughts
Running on all UK an USA markets, it has runs of a few weeks of fantastic profit and then similar periods of losses. Watching the races on the winning periods, every horse is competitive and challenging come the finish, like it selects the "right" horses. On the losing runs they are often uncompetitive and never look like winning. The results suggest it's deeper than just natural variance and volitility and the evidence of watching the races really makes it feels like its either "on" or "off". Like it's "there" or it isnt.
I've been trying to work out what unerlying factors could be leading to this. I considered it could be automation more advanced than mine, identifying the likely horses and mine was just piggybacking off this, in periods it doesn't run, mine struggles. Stable money waiting to see if a horse had jumped off and settled ok before backing them, trackside traders, but i'm at a loss. Strangely, it is across the board, so it'll be profitable on NH, AW and USA racing at the same time, and then not profitable at the same time.
Overall it is profitable, but somehow recognising these "on" and "off" periods would be a huge step forward.
Any Ideas?
Many Thanks