SUBSCRIBE TO THE TENNISRATINGS YOUTUBE CHANNEL FOR THE LATEST TENNIS TRADING & TENNIS BETTING VIDEOS!
Tomas Berdych has excellent break deficit recovery stats...
In the second of a small series of articles, I look at how we can use break-back percentages in-play and the maths and statistics behind them.
The first article deconstructed the Rolling Projected Hold stats in the match in Miami between Agnieszka Radwanska and Dominika Cibulkova, in which the stats illustrated that despite being underdog at the start of the match, and having a low projected hold, Cibulkova actually had a higher projected hold in-play because of the game state. In that match the break-back percentages indicated that it wasn't viable enough to oppose either player a break up, hence the need to write a further article about a match where it was.
The reason that I picked this match - played in Miami on the 27th March 2014 - is purely because I discussed how it was viable to lay Dolgopolov a break up on Twitter, but it was hard to explain my rationale fully in 140 characters...
The pre-match market is an interesting starting point.
Berdych opened at 1.53 which was a ridiculous opening line. Dolgopolov had achieved excellent results (although against poorer players than Berdych on hard court) and this skewed the opening lines. As the higher ranked player enjoying a 10.4% projected hold edge, as well as having a better break point 'clutch' score, Berdych should have been 1.23 according to my model. The market saw that the opening line was erroneous, and backed Berdych into 1.38. However, for me he still started the match at too big a price.
To clarify the projected holds, Berdych's was 78.0% - almost exactly the ATP mean of 78.3%. Dolgopolov's was low, at 67.6%.
However, it's the break-back percentages that is the main focus of this article and at the time of the match, my stats on Dolgopolov was that he lost a break lead 33.3% in the last 12 months, with Berdych better at a lower 20.7%.
Berdych also had better break deficit recovery stats, getting a break deficit back on serve an excellent 47.6% in the last year. For a player that's probably better known for his serving than returning, that is highly impressive and close to elite level. Dolgopolov recovered a break deficit 39.7% - decent and above the top 100 average of 34.6%, but not nearly at the level of Berdych.
This lead to combined scores of 80.9 (33+47.6) on Dolgopolov when a break up, and 60.4 (20.7+39.7) on Berdych.
As Tier Two Spreadsheet subscribers and regular match previews will be aware, I recommend that the trade of laying a player a break up can only be viable when the combined score is over 75, and my earlier break-back percentages article explains the maths behind that approach.
Therefore it is viable to oppose Dolgopolov in this situation, with a combined score of 80.9.
In this match, Dolgopolov broke Berdych in the opening game of the first set, to lead *1-0. On Twitter we managed to establish that with market prices after the break, prices after a potential break-back and the 'set price' should Berdych not break-back (e.g. Dolgopolov wins the first set) all taken into account, Berdych would need to recover the break deficit 50% of the time for the trade to break even in the long term.
Now the average percentage chance of Berdych breaking back is 40.45% (80.9 combined score/2) but this is at the AVERAGE POINT of the set. The average point of the set is probably a little contentious, but it's probably fair to assume it's around the *3-2 or *2-4 stage of the set.
Clearly at *1-0, there is a lot less time decay than at *3-2 or *2-4 so Berdych has a lot more time at *1-0 to Dolgopolov than at the average point of the set. To use a football context, if Manchester City play Norwich and concede a goal in the first ten minutes of the match, they have 80 more minutes to fight back, but if they concede the opening goal on the stroke of half-time, they have just 45 minutes to fight back. Obviously conceding early, giving Manchester City more time to recover the goal deficit is beneficial, and the same can be applied in Tennis as well.
We can see from the break-back article that players with a combined score of between 75 and 90, when broken in the opening game of the set, recovered the deficit 54.93% of the time:-
This actually rose to 60.87% after game 2 of the first set (e.g. for *0-2). With sample sizes not huge, it's probably fair to assume that a realistic figure is somewhere between the two, so perhaps around the 57.9% mark would be a fair percentage for the likelihood of a break-back when the combined score is between 75 and 90. However, for analysis I want to stick to the initial 'break after game 1' percentages.
The midpoint of the combined score bracket is 82.5, marginally more than the 80.9 on Dolgopolov. Using midpoint analysis, the following percentage is generated for the likelihood of Berdych when a break down at *1-0 to Dolgopolov.
Midpoint of the 60-75 bracket = 67.5
Midpoint of the 75-90 bracket = 82.5
Difference = 15.0
Break-back percentage of the 60-75 bracket = 40.98%
Break-back percentage of the 75-90 bracket = 54.93%
Difference = 13.95%
Break-back percentage difference/midpoint difference = 13.95/15.0 = 0.93.
Therefore we can assume that the break-back percentage increases by 0.93% every time the combined score rises by 1 between 67.5 and 82.5. On that basis we can generate the following percentage for Berdych to break-back:-
((80.9-67.5)*0.93)+40.98 = (13.4*.93)+40.98 = 53.44%
If our maths assuming that the break-back needs to come 50% of the time to break even was correct, based on the break-back percentages in the break-back article, we just generated a 3.44% edge over the market. This doesn't sound like much, but is a fairly realistic figure in a market which is relatively efficient. If every single one of our trades had this edge, we wouldn't be doing too badly...
Furthermore, this is without accounting for the fact that - as discussed above - it is logical that 57.9% would be a slightly more realistic figure for the 75-90 bracket as opposed to the 54.93% we used and if this was the case, it would increase our edge further. Also in our favour was the fact that my model found the starting prices to offer value on Berdych, and this would also increase our edge.
The result is largely irrelevant, because this trade now has a proven long-term edge. An individual match neither proves or disproves any analysis, but for the record, Berdych immediately broke back for 1-1, and took the first set 6-4. Dolgopolov also lead by a break in the second set after game five to lead *3-2, and Berdych broke back for 5-5 when Dolgopolov was serving for the set, and Berdych took the set - and the match - on a tiebreak.
A Selection of TennisRatings Products for 2017
Please visit the
TennisRatings Products links for a full overview of our fantastic Tennis Trading tools, and the TennisRatings Subscription Packages link to see our great value range of discounted subscription packages!
Please check out our testimonials page!
The TennisRatings Daily Trading Spreadsheets have never been more popular!
To find out more on how these can dramatically improve your Tennis Trading, check out the YouTube Video we made.
The Challenger Daily Spreadsheets cover all ATP Challenger Events and include projected hold percentages (for traders) and model prices (for bettors and traders).
Subscriptions are available for 3 months:-
The Lead Loss/Recovery Data Spreadsheets have taken the Tennis Trading World by storm - discussed in detail in October 2015 at the Matchbook Traders Conference these incredible spreadsheets highlight lead loss & deficit recovery in individual sets, as well as how often a player loses/gains the first break of the second set based on whether they won or lost the first set!
INCLUDES FREE REGULAR UPDATES - THIS IS A ONE-TIME PURCHASE!