Hi there,
I was wondering how Tennis Trader comes up with it's initial statistics prior to calibration (% win game, % win set, % win match)
Is this inferred through the starting odds for the game or do you use a data set to determine these numbers?
Thanks,
Dan
accuracy and statistics
Tennis throws of lots of stats that are useful for modelling. First serve %, second server %, points on serve etc. etc.
When a market is formed on Betfair people look at these stats and put up a price. What Tennis Trader does is look at the price and decompile the odds back to their core underlying metrics. You can then project these forward using an algorithm contained in the product. The algorithm isn't radically different from that which is used by most bookmakers and spread betting companies.
Where the difficulty lays is in-play. If you project forward at the off Tennis Trader assumes that these metrics remaining the same throughout the match which is obviously unrealistic. As the match progresses there will be subtle changes according to how the players are playing how tired they are etc. Also, there is speculative money in the market that can push a price beyond what could be consider reasonable by any model, inplay delays etc. If you visit the back office of in-play bookies or similar organisation they tend to tweak models up or down based on how they feel the match is going. However if you calibrate in-play Tennis trader will attempt to decompile the odds according to what it sees and what you have entered. From there you can start working forward again. It is likely that prices can drift inplay but you show try to look at them from a % viewpoint and not ticks. If you calibrate at low odds there are a number of ways the match can reach those odds so it is less effective calibrating at low odds.
My approach to all markets is that if I can't model a market I won't trade it. Modelling allows you to pinpoint key areas in a market that you can trade with confidence and work out your approximate up and downside. That is whole concept behind these modelling tools.
We know it's impossible to nail the odds exactly, but we can get close and our philosophy is that it is whole lot better than no information whatsoever.
When a market is formed on Betfair people look at these stats and put up a price. What Tennis Trader does is look at the price and decompile the odds back to their core underlying metrics. You can then project these forward using an algorithm contained in the product. The algorithm isn't radically different from that which is used by most bookmakers and spread betting companies.
Where the difficulty lays is in-play. If you project forward at the off Tennis Trader assumes that these metrics remaining the same throughout the match which is obviously unrealistic. As the match progresses there will be subtle changes according to how the players are playing how tired they are etc. Also, there is speculative money in the market that can push a price beyond what could be consider reasonable by any model, inplay delays etc. If you visit the back office of in-play bookies or similar organisation they tend to tweak models up or down based on how they feel the match is going. However if you calibrate in-play Tennis trader will attempt to decompile the odds according to what it sees and what you have entered. From there you can start working forward again. It is likely that prices can drift inplay but you show try to look at them from a % viewpoint and not ticks. If you calibrate at low odds there are a number of ways the match can reach those odds so it is less effective calibrating at low odds.
My approach to all markets is that if I can't model a market I won't trade it. Modelling allows you to pinpoint key areas in a market that you can trade with confidence and work out your approximate up and downside. That is whole concept behind these modelling tools.
We know it's impossible to nail the odds exactly, but we can get close and our philosophy is that it is whole lot better than no information whatsoever.
I had a go with the new Tennis tool during Wimbledon and found it to be pretty useful. I only did a couple of matches and made a bit on both. I have a broad strategy for the tennis and I'm sure the pros will tell you it's great for highlighting moments in the match where a good trade is likely with little downside.