View RSS Feed

mysterX

Scrub vs Qualifier: The Eternal Battle

Rate this Entry
I have been participating in various prediction contests for years, and one of the things I have learned is to be wary of picking against qualifiers. I can't count the number of times I made a prediction thinking "Player X has this one easy, he/she is playing a qualifier," then watching player X crash and burn against that very same qualifier. I started wondering about the overall success of qualifiers in tournaments, with two questions in mind:

1. How well do qualifiers do against other types of players?

2. Do qualifiers have an edge in the first round, having already played a number of matches and gotten used to the playing conditions?

To answer these questions, I put together a database of grand slam results going back to 1988. I picked 1988 for two reasons. First, it is the first year where all grand slams had a complete 128-player draw (Australian Open had 64 and 96-player draws in prior years). Second, for my analysis, I needed to look at the rankings of the players involved in the matches. I used the playing results from the ATP and WTA websites to get ranking data for the grand slam matches, and their sites don't keep the ranking data before 1987 or so.

The database I built has over 25000 match results for every singles match played at any grand slam, men or women. For each match, in addition to the score of the match, I recorded the ranking for each player at the time of the match, as well as any other characteristics (was the player seeded, was the player a qualifier or a wild card, did the player have a protected ranking). The next thing I did was to categorize the players. I created six player types:


  • Seed - Player was a seed at the time of the match. Prior to the early 2000's, there were 16 seeds per grand slam. Since then the number increased to 32/
  • Unseeded - Player was not seeded, but earned direct entry with a ranking at or above the "scrub cutoff".
  • Scrub - Player earned direct entry, but with a ranking below the scrub cutoff.
  • Qualifier - Player qualified for the tournament
  • Wild Card - Player was awarded a direct entry into the tournament, despite not having a ranking which would earn direct entry.
  • Protected Rank - Player was recovering from injury, and was awarded a direct ranking based on their ranking before the injury.


For the scrub cutoff, I chose a ranking of 75. Direct entry players with a ranking of 75 or higher who were not seeded are put into the Unseeded category, while direct entry players with a ranking of 76 or lower are put into the scrub category. I also did some experiments with scrub cutoffs of 50 and 32,

With each player falling into one of the six categories, we can now compare head-to-head results for matches which feature players in two of the categories. Earlier, I mentioned that I was interested in the results of qualifiers. Now we can see the results of matches featuring qualifiers vs players in the other categories, including different scrub cutoffs.

Player Type Win % All Rounds Win % 1st Round Win % Later Rounds
Seeded 16.07% 17.74% 14.7%
Unseeded/75 35.16% 36.11% 33.11%
Unseeded/50 30.25% 31.51% 27.78%
Unseeded/32 24.82% 26.7% 21.57%
Scrub/75 56.13% 56.22% 55.81%
Scrub/50 50.43% 50.91% 48.99%
Scrub/32 46.2% 47.07% 43.82%
Wild Card 54.02% 58.1% 37.78%
Protected Rank 40% 38.46% 50%

From this data, you can see that qualifiers struggle against seeded players, win a little more than 1/3 of the time against unseeded players, and have a winning percentage vs scrubs and wild cards. With the scrub cutoff set to 50, it is basically a coin flip between qualifiers and scrubs. With the scrub cutoff set to 32, the qualifiers still manage to win over 45% of the time. Qualifiers have a losing record against players with protected rankings, but that is based on only 15 matches.

As to whether qualifiers get a first match benefit, the winning percentages are slightly higher in the first round than for later rounds. I was expecting the qualifier percentages to go down after the first round, thinking that as other players had won a match in the tournament, that the rankings differences between them and the qualifiers would be more of a deciding factor. Instead, it appears that a qualifier who has played well enough to win a first round match will continue to play nearly as well in later rounds. What was striking to me was how well qualifiers did against scrubs. I expected qualifiers to have a good winning percentage, but I am surprised that it is over 50%.

A couple other things to show before wrapping up. The chart below shows the winning percentages of head-to-head matches across all types, using a scrub cutoff of 75.

Name:  HeadToHead_75.png
Views: 62
Size:  21.8 KB


And finally, here is a link to a spreadsheet with the summary data I used for my analysis.

https://docs.google.com/spreadsheet/...=1&output=html

I hope you find this information useful.

Submit "Scrub vs Qualifier: The Eternal Battle" to Facebook Submit "Scrub vs Qualifier: The Eternal Battle" to Digg Submit "Scrub vs Qualifier: The Eternal Battle" to del.icio.us Submit "Scrub vs Qualifier: The Eternal Battle" to Google

Categories
Tennis , ATP , WTA

Comments

  1. ptmcmahon's Avatar
    Very useful...amazing you were able to compile from such a large database.

    As for our prediction contests...dunno if this "helps" or makes it even trickier
  2. mysterX's Avatar
    I'm planning to use a coin to help with all my future predictions. Heads and I will pick the higher-ranked player, tails I go with the lower-ranked player.

    I am interested in trying to figure out at what point do ranking differences become significant. For example, if the 34th ranked player plays the 134th ranked player, I would expect the 34th ranked player to win most of those matches. But when the 34th ranked player plays the 35th ranked player, is there any edge to the higher ranked player? I have to do some more analysis to answer that question.
  3. Ribbons's Avatar
    I love actual analysis. This was a fun read!
  4. mysterX's Avatar
    Thanks Ribbons.

    I've been meaning to do some more analysis, now that I have some data to work with. Other things (mainly, life) seem to keep getting in the way though.