WTA Model analysis – Injuries

Previously I showed about how the updated WTA model goes with all sorts of different probabilities, odds and overlays; all with very good results.

Now I will look at all the NOBET scenarios and how the WTA model goes betting for or against a player with injuries and the like. For want of reminder, the NOBET’s are as follows:

NOBET1 – Player has recorded 2 retirements in last 2 months
NOBET2 – Player has not won a match (or played) in last 2 months
NOBET3 – Player has not won a match since last retirement
NOBET4 – One of the two players have played less than 10 matches on the particular surface

Shown below are the results for the above NOBET’s. A value of 1 means that model bet on a player that has the above attributes, whilst a value of -1 means that the model bet against that player. A score of zero meant that we bet on a player where neither them nor the opposition has a NOBET attribute.




-11478759.2% $48,041.23 $4,144.90 8.6%
0197185743.5% $341,181.29 $23,897.87 7.0%
1672537.3% $9,742.43 -$1,124.31 -11.5%
218596944.3% $398,964.95 $26,918.45 6.7%


-11034442.7% $18,179.95 $3,627.83 20.0%
0204991244.5% $373,744.77 $24,232.88 6.5%
1331339.4% $7,040.23 -$942.26-13.4%
218596944.3% $398,964.95 $26,918.45 6.7%


-11588151.2% $43,729.74 $4,516.20 10.3%
0198186043.4% $343,344.15 $22,176.97 6.5%
1462860.9% $11,891.07 $225.271.9%
218596944.3% $398,964.95 $26,918.45 6.7%

Results from above are from 2009, and unfortunetly have a small sample size, although there is something that we can discuss about it.

It seems clear that betting on players who fit the NOBET’s 1 to 3 category still have a negative profit, whist betting against them is, in general, quite successful. The model does take into consideration a player who has NOBET1 to 3, and adjusts their probability, however it seems that it doesn’t do it enough.

Hence there are two options left for me, to manually change the coefficients so that the model accounts for this more so, or have NOBET flags that appear again as before.

There are reasons why, despite the new wta model accounting for the NOBET’s that betting on them still appear unsuccessful. This is because a player might be seriously injured or just marginally injured. The model accounts for all such players as the same level of injury. Hence we might well find ourselves betting more on players who are more injured than the average, and less on players who are less injured than normal.

Also, if we were to bet on a NOBET player, it would be because that player is in some sort of form, but has recently been injured or not played for a while. Perhaps the odds account for this inform play a little too much for a player in this situation.

Either way, this is a small sample size, and it could pay to look back before 2009 to check the results. Also, betting against NOBET’s 1 to 3 seems to have been successful. In fact from 275 bets, betting against a NOBET1-3 has resulted in a 10.9% ROI. It might also be interesting to see if we were to blindly bet against NOBET’s1-3 if it is a successful betting technique or not. More on that later.

As far as new players goes, the model seems to adjust for them quite well. NOBET4 seems to be profitable betting for or against these players. A discussion about how to deal with these scenarios could well be worth it

To sign up to Sportpunter’s WTA model or view the prices, click here.

Related Posts with Thumbnails
This entry was posted in Model, Sport Models. Bookmark the permalink.

3 Responses to WTA Model analysis – Injuries

  1. mark says:

    I’ve this is a difficult area to filter/model.

    For instnce, not all injuries are alike. Quite often players retire from heat or gastro which have very different effects to an ankle or wrist.

    I have pretty good records going back to 2005 and one area that has helped is eliminating bets on players who retired in their last match (due to physical injury).

    The current WTA retirement count is around 102 so far this seaon (which is normal) so there’s a few matches in a 6 season sample … but never really enough to reach a definitive conclusion.

  2. sportjake says:

    Yes, would be interesting to see if decreasing prob even more on NOBETS would decrease loss (and increase profit on other player).
    In meantime would it be possible for WTA Update Page to include a new column indicating if player page is suggesting as bet is a NOBET1 to NOBET4?

  3. admin says:

    yes that was my plan sportjake, there is one column free which I shall use to notify if any of the above issues occur. It will also help people find players whos names change int he database and it needs updating.

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.