Does anyonw use this tool for trading.
Reason asking the predicted odds seam to be way off what is actuall trading?
Example Kvitova was 3.2 before first game today If she holds tennis trader predicts price is 2.95. the price actually moved to 2.68
Eaxmple kvitova was 3.05 at start of 5th game and tennis trader predicted if she got broke would go to 5.5 ...price only moved to 4.5...
Any thoughts on this... as i had some trading ideas but predicted price was off etc
Betangel tennis trader
Tennis trader is as accurate as you can get without extensive user intervention inplay on updating masses of statistics. During the US Open final it performed well. Where you see variances it will be down to the following: -
(1) The chances of winning vary from 1-100%, odds increments move in 351 ticks, so the two don't match and this may make the odds look out of whack. If somebody has a 3% chance of winning their odds will be 33, but if they have a 1% chance of winning it will be 100. Big odds difference, small % chance. In the example above you can see a 3-4% variance.
(2) Variations in the starting price to how the match unfolds. If a player is 2.00 at 0-0, then you could expect if the match is tied at some point in the future then the same odds should apply. That is the only way Tennis trader can price it, because it can't see the match. If there is a variation at that point you need to decide if the underlying match has shifted or whether the assumptions at the start of the match are still valid.
You can see logically that if players start at certain odds then they should return to those odds if the match is level and we have to use that assumption in Tennis trader and let users over rule it.
Don't forget to adjust the match type as well, that often catches people out.
There are limitations to how you model things, but it's better to have something than nothing at all. If we adding in all the variable you could use I doubt many would attempt to use it. So there has to be some give and take in complexity vs. usability. It's based on well established academic models, but you have to put some perspective on the limitations that and its usage imposes.
It was a key part of the process that made my Tennis trading pay, so that's why we put it into Bet Angel.
(1) The chances of winning vary from 1-100%, odds increments move in 351 ticks, so the two don't match and this may make the odds look out of whack. If somebody has a 3% chance of winning their odds will be 33, but if they have a 1% chance of winning it will be 100. Big odds difference, small % chance. In the example above you can see a 3-4% variance.
(2) Variations in the starting price to how the match unfolds. If a player is 2.00 at 0-0, then you could expect if the match is tied at some point in the future then the same odds should apply. That is the only way Tennis trader can price it, because it can't see the match. If there is a variation at that point you need to decide if the underlying match has shifted or whether the assumptions at the start of the match are still valid.
You can see logically that if players start at certain odds then they should return to those odds if the match is level and we have to use that assumption in Tennis trader and let users over rule it.
Don't forget to adjust the match type as well, that often catches people out.
There are limitations to how you model things, but it's better to have something than nothing at all. If we adding in all the variable you could use I doubt many would attempt to use it. So there has to be some give and take in complexity vs. usability. It's based on well established academic models, but you have to put some perspective on the limitations that and its usage imposes.
It was a key part of the process that made my Tennis trading pay, so that's why we put it into Bet Angel.
I use the tennis predictor and find it fairly useful. You have to take its predictions with a pinch of salt though, and use your own judgement to a large extent.
I find it works much better for men's matches than women's and am a little surprised there isn't a different setting for men's and women's matches as the break of serve is far less significant for women than for men, generally speaking.
I wouldn't have thought it would be too difficult to introduce that alongside the settings for different lengths of matches and final set tie-breaks.
I find it works much better for men's matches than women's and am a little surprised there isn't a different setting for men's and women's matches as the break of serve is far less significant for women than for men, generally speaking.
I wouldn't have thought it would be too difficult to introduce that alongside the settings for different lengths of matches and final set tie-breaks.
Hi,
Two remarks about tennis trader:
1) Academic models say that the Markov Chains involved in calculating all the % in a tennis match start with the probability each player has to win a point. From there you calculate the % of winning a game, a set a match.
Tennis trader uses Betfair odds as an input for the model. But, as everybody knows two matches can have the exact same PG odds but the players diferent % to win a point. If this is true inside the Mens or Womens game, this is even more true when comparing Mens to Womens game.
To give an example, a pick men's match on a Hard Court have around 70% of each player winning a point. And a women's pick match on the same hard court should be around 58%. This should be taken into account for the model.
Diferent surfuces are also not taken into account but should. First, we all know that are players that simply don't know how to play certain surfaces (see Pete Sampras in clay!!!!). And diferent types of surfuce produce a huge diference in serve - we can check that using as a measure the number of Aces per match on Clay match vs a match in Grass.
2) Even tho all the points I have made above and that are very important - by the way, if BA wants to incorporate this into the next version of the Tennis Trader I am very happy to help - it would be really good to have the possibility to connect the predicted prices to Excel in order to together with the markets connection with Excel desing new strategies for trading.
D
Two remarks about tennis trader:
1) Academic models say that the Markov Chains involved in calculating all the % in a tennis match start with the probability each player has to win a point. From there you calculate the % of winning a game, a set a match.
Tennis trader uses Betfair odds as an input for the model. But, as everybody knows two matches can have the exact same PG odds but the players diferent % to win a point. If this is true inside the Mens or Womens game, this is even more true when comparing Mens to Womens game.
To give an example, a pick men's match on a Hard Court have around 70% of each player winning a point. And a women's pick match on the same hard court should be around 58%. This should be taken into account for the model.
Diferent surfuces are also not taken into account but should. First, we all know that are players that simply don't know how to play certain surfaces (see Pete Sampras in clay!!!!). And diferent types of surfuce produce a huge diference in serve - we can check that using as a measure the number of Aces per match on Clay match vs a match in Grass.
2) Even tho all the points I have made above and that are very important - by the way, if BA wants to incorporate this into the next version of the Tennis Trader I am very happy to help - it would be really good to have the possibility to connect the predicted prices to Excel in order to together with the markets connection with Excel desing new strategies for trading.
D
Hi Euler,Euler wrote:That would be easy to do. The only reason it feeds from Betfair odds is for simplicity. All it does is hunt for a common value based upon data from around 20,000 matches.
What would be easy to do ? And, whatever it is, when can it be expected to be available ?
TA
D
-
NileVentures
- Posts: 79
- Joined: Mon Oct 22, 2012 9:13 pm
- Location: West London
Does anyone know if it possible to extract / link the predictions from Tennis Trader to excel for use as input to automated trades?
Is it possible to download the output of tennis trader into Excel? Or manipulate it like a function inside Excel given current score or a score scenario?Euler wrote:Just sent you an email.
Euler, this would be really helpful for us who use Bet Angel Pro. Is it possible?
Thanks,
PK
