After a couple of seasons trading tennis on Betfair I've started leaning less on gut feel and more on data. Over the winter I put together a rough tennis prediction algorithm that pulls in surface, recent form and head-to-head, then spits out a win probability I can compare against the market price before the off.
It's far from perfect (clay throws it off badly, and it has no idea what to do when someone retires mid-match), but it's been a handy sanity check against the prices Bet Angel shows me on the ladder.
Curious how many of you run something similar? Do you build your own models in Excel or Python, or do you mostly read the market and trade off momentum in-running? Keen to hear which inputs you've found actually move the needle for pre-match tennis.
Anyone else building their own model for pre-match tennis?
Combination of the following with own personal adjustments dependent on level (ie 250+ or challenger), court speed factored & lots of data.
In a nut shell all down to predicted service hold.
Peter has shared videos on calculating predicted holds
https://martiningram.github.io/papers/b ... _based.pdf
https://journals.sagepub.com/doi/pdf/10.3233/JSA-200345
In a nut shell all down to predicted service hold.
Peter has shared videos on calculating predicted holds
https://martiningram.github.io/papers/b ... _based.pdf
https://journals.sagepub.com/doi/pdf/10.3233/JSA-200345