Understanding tennis players ranking

A tennis player’s world ranking is a controversial subject. Calculated on a 52-week rolling basis, ranks tend to fluctuate significantly during the Tennis season. 

A long-term injury - think Juan Martin Del Potro, Nicolas Almagro and Tommy Haas recently - will affect a player’s rank severely and may leave them unable to enter directly into tournaments without qualifying or receiving wild cards.  Due to this, some observers feel that a 2-year (or 104 week) period would be fairer, and more representative of a player's actual ability.

Furthermore, world rankings tend to reflect the current predominant surface being played in the majority of tournaments, at that time of the season. For example, clay court experts are likely to be ranked higher, after the main clay season finishes, than they would prior to it. This means that rankings are often skewed and are not a true reflection of the all-surface abilities of players.

On this basis, the market may interpret player rankings incorrectly, with a player who has risen up the rankings recently due to several strong tournaments likely to be over-rated. So what does this mean for tennis bettors?

Are lower ranked favourites undervalued in the ATP?

An interesting angle would be to look at players who are favourites in the market, but are the lower ranked player. These players are perceived by the market to be the better player, but considered worse by the rankings by the ATP/WTA Tours. Logically, these players are likely to be under-rated despite being favourite, due to being the lower ranked player. 

The following table assesses a sample of players on the ATP Tour throughout their career (main draw and qualifying matches) when they started the match as favourite, but as the lower ranked player. 

A cross-section of players ranked from 1 to 100 were sampled, with two players around the midpoint of each ten ranking places included.

All prices used are Pinnacle’ closing prices, and only matches where at least one set was completed were included:

Player	Rank	Matches	Wins	Win %	P/L	ROI %
Nishikori	5	23	12	52.17	-476	-20.70
Murray	6	13	8	61.54	-28	-2.15
Anderson	15	27	18	66.67	354	13.11
Bautista-Agut	16	30	19	63.33	252	8.40
Benneteau	25	32	21	65.63	109	3.41
Mayer L	26	45	33	73.33	945	21.00
Querrey	35	52	30	57.69	-181	-3.48
Mannarino	36	22	13	59.09	-70	-3.18
Sock	45	33	17	51.52	-333	-10.09
Seppi	46	12	9	75.00	326	27.17
Sousa	55	11	7	63.64	149	13.55
Young	56	39	27	69.23	757	19.41
Bellucci	65	41	27	65.85	83	2.02
Tomic	66	44	29	65.91	110	2.50
Jaziri	75	12	8	66.67	190	15.83
Haas	76	35	25	71.43	676	19.31
Berankis	85	46	30	65.22	277	6.02
Dodig	86	21	14	66.67	246	11.71
Kamke	95	23	14	60.87	-197	-8.57
Gojowcyzk	96	26	12	46.15	-504	-19.38
Overall		587	373	63.54	2685	4.57

As can be seen from the table, a significant sample of data was generated with 587 matches sampled. 373 wins came from this sample (a win percentage of 63.54%) and when a £100 hypothetical stake was applied to all matches sampled, a profit of £2,685 was generated (return on investment of 4.57%).

This ROI is very strong for a blind backed sample of a significant size, and can be treated as a huge edge in the market.

Andreas Seppi, Leonardo Mayer and Tommy Haas all recorded win percentages over 70% from this scenario, with Haas particularly interesting, as he recovered his ranking following a long-term injury in 2012. The same issue will affect Haas in the future, as he has been absent from Tour through injury since May 2014. Donald Young, with a win percentage of 69.23% and ROI of 19.41%, was also historically very strong.

Just seven players of the 20 sampled recorded negative ROI figures, with Kei Nishikori, Peter Gojowczyk and Jack Sock recording the worst figures of over -10% ROI. Interestingly both top ten ATP players sampled (Andy Murray was the other player) had negative figures, illustrating that both currently have poor records against elite level opponents.

Are lower ranked favourites undervalued in the WTA?

Player	Rank	Matches	Wins	Win %	P/L	ROI %
Ivanovic	5	29	20	68.97	346	11.93
Radwanska A	6	8	6	75.00	229	28.63
Jankovic	15	12	7	58.33	18	1.50
Safarova	16	26	18	69.23	403	15.50
Pavlyuchenkova	25	16	10	62.50	118	7.38
Svitolina	26	12	8	66.67	122	10.17
Keys	35	29	22	75.86	637	21.97
Garcia	36	13	8	61.54	10	0.77
Vinci	45	15	10	66.67	297	19.80
Koukalova	46	10	7	70.00	169	16.90
Beck	55	21	11	52.38	-337	-16.05
Jovanovski	56	11	6	54.55	-152	-13.82
Vesnina	65	24	18	75.00	766	31.92
Shvedova	66	41	26	63.41	234	5.71
Erakovic	75	30	18	60.00	-77	-2.57
Rogers	76	14	6	42.86	-381	-27.21
Vekic	85	12	9	75.00	308	25.67
Parmentier	86	22	11	50.00	-275	-12.50
Konjuh	95	30	23	76.67	545	18.17
Gibbs	96	20	14	70.00	385	19.25
Overall		395	258	65.32	3365	8.52

Interestingly, the WTA figures were even stronger, with a return on investment of 8.52% from 395 matches, as well as a slightly higher 65.32% win rate. These statistics are truly stellar numbers for a blind-backed scenario.

Of the players with a large sample, Madison Keys, Elena Vesnina and Ana Konjuh had win percentages of 75% or over, and these look like players whose rank is likely to be, or has been below their level of ability.

Quite incredibly, only four WTA players of the 20 sampled recorded negative figures - Annika Beck, Bojana Jovanovski, Shelby Rogers and Pauline Parmentier. 

Overall, combining the ATP and WTA figures, profits of £6050 were generated from a stake of £98,200, giving a return of investment of 6.16%.  From such a huge sample, this profit level is likely to be statistically significant, and it would appear that the market has underestimated this scenario in the extreme.

Bettors would be very strongly advised to consider this in their future tennis analysis.