Jump to content
** March Poker League Result : =1st Bridscott, =1st Like2Fish, 3rd avongirl **
** Cheltenham Tipster Competition Result : 1st Old codger, 2nd sirspread, 3rd Bathtime For Rupert **

Algorithm Tennis Predictor


Recommended Posts

Right guys, I'm back again. I have a bit more time on my hand, I'm always doing some thinking and some learning so here is my next go at finding something that works. As always it's tennis, and I want it to be set and forget as I don't have time to sit and trade matches all day and night. 

This new way is using an algorithm to predict the match odds for tennis matches. Currently, some are fairly close, and some are a long way off. This is just the testing and trying from day one so adjustments will no doubt be made but let's see how it goes from the start.

The following are taken into account and weighted differently:

W/L for last year

W/L for last 2 years

W/L all time before that

Same three as above but just for the surface of the current match.

H2H games

H2H matches

Results from playing in current matches country in the past

Matches against the same nationality as opponent

Current form (10 matches)

Tiredness (matches in last 20 days)

Any recent retirements

Previous at this tournament. 

(Edit: Ranking as well)

Because it's throwing up odds that are obviously not bang on current Betfair odds, there should be lots of selections. I will try to do as many matches as I can but early in the week sometimes might be hard with lots of matches to go through. 

Currently paper trading, will use fictional £10 stakes for the purposes of the spreadsheet, and also just using the "what if" on Betfair so they will have 5% commission taken off as well. 

Let's hope it's interesting and something to continue working on. 

Wish me luck

 

 

 

Edited by Smiles Tennis
Adding ranking to criteria used
Link to comment
Share on other sites

Two more selections. I have also included a screenshot of the algorithm at work for the Hassan selection. Came out at around 50% but he is 2.52 so worth a pick. Not the brightest start for the first day of the trial but if it doesn't go really well then really bad can also work as we can trial backing their opponents instead!

 

 

hassan.png

Screenshot 2023-04-06 100119.png

Link to comment
Share on other sites

Almost 4 points paper trading loss from the first days matches. These are selections up to tea time matches later. If I am around and the liquidity has increased a bit there might be some more for this evening. 

One thing I have noticed is it's hard to get the predicted odds close to the Betfair odds when the fave is very short most of the time. Tsitsipas has been a bit close and therefore been a selection, but even having a little play with the parameters you have to make big changes to the influence of some factors to get a large difference in the predicted odds. This has been the case for many of the matches with the fave around 1.2. 

Saying that I have just checked the Alcaraz v Evans match where Alcaraz is currently 1.07 (I don't usually check matches this short) and the algorithm gave Alcaraz 100%, so it can get a big enough difference if the data is there. More to look at and think about as we get more results in. 

tennis.png

evans.png

Link to comment
Share on other sites

A run of losses means these early paper trading results swings back to the negative. Be nice if it keeps heading one way instead of swinging back and forth but time will tell if it's worth following properly one way or another. 

Around 50 bets so far, another 50 or so and we might get a little idea of which way it might go. 200 would be better of course, and a month would give a much better picture. 

Another 3 for this evening to keep an eye on. 

 

tennis.png

Link to comment
Share on other sites

5 hours ago, Smiles Tennis said:

Huge run of losses building so good data to look at. 

Only following from a distance (checking in but not looking at the individual selections). Gut feel is too many picks in too short a time which suggests you either need to review the accuracy of the model in terms of identifying "true" odds or that the required edge for placing a bet is too small. It basically feels like something's off if you're identifying this many bets though that could be countered if it ended up showing a huge profit as the sample gets bigger!

Link to comment
Share on other sites

54 minutes ago, harry_rag said:

Only following from a distance (checking in but not looking at the individual selections). Gut feel is too many picks in too short a time which suggests you either need to review the accuracy of the model in terms of identifying "true" odds or that the required edge for placing a bet is too small. It basically feels like something's off if you're identifying this many bets though that could be countered if it ended up showing a huge profit as the sample gets bigger!

I think it's way off with regards to actually finding true odds. The markets won't be that wrong and I wont be that right with just throwing (mostly) together an algorithm to pull stats and use that for price. It's looking more likely now that the picks are losing so often that it will be worth looking at backing the opponents instead.

Link to comment
Share on other sites

Speaking from experience, you'll need thousands of bets before you know if you're on the right track. I know you've had a few goes at this sort of thing so my advice would be to persevere rather than tweaking all the time. When I say persevere I don't mean forever as that could lead to large losses, but at least until you reach a stop loss point.

Link to comment
Share on other sites

Also, the suggestion that if betting on player 2 all the time leads to losses then betting on player 1 will make gains is flawed. There's juice on both sides so what you'll find in the long run is betting on player 1 all the time will lead to a loss as well. If player 1 is normally the favourite then that loss should be achieved more slowly as you'll have more winners than losers, but it'll be a loss just the same and most likely a lesser loss than when picking underdogs.

Link to comment
Share on other sites

1 hour ago, Torque said:

Speaking from experience, you'll need thousands of bets before you know if you're on the right track. I know you've had a few goes at this sort of thing so my advice would be to persevere rather than tweaking all the time. When I say persevere I don't mean forever as that could lead to large losses, but at least until you reach a stop loss point.

Cheers, all just paper trading at the moment and seeing what happens. I understand what you mean in your other comment about the fave also ending up as losses, just more slowly. I think if the losses are huge and consistent with this method then paper trading the other method must be worth a try. No money down for a while yet, I will persevere with these picks just to see how it goes first. 

Link to comment
Share on other sites

Few more for this evening. I have started tracking the results of backing the opponent of these selections as well, separately. I won't post them at the same time in here but if this keeps heading the way it's heading then I will be swapping the selections for the opponents, with no adjustments to the algorithm. 

It's still just paper trading, collecting bets and seeing what we can find out and maybe use.

 

 

tennis.png

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...