It appears that aces and receive errors are highly correlated. I believe some systems don’t even distinguish between the two. I don’t mind that aces + receive errors count twice as much as service errors here. If you serve even half as many aces + receive errors as service errors, that’s still a very good ratio because side out percentages on good passes are pretty high. In fact, I could argue that the risk-reward factor of aces vs service errors should actually rank players like Clancy and Gottardi higher than what you have.
I do think that points scored while serving may be weighted too heavily. Bad receives already accounts for good serves getting opponents out of system, plus as you mention points scored is affected by many other factors other than serving. But I’m not sure I’d remove it altogether either.
Good stuff again. The men’s top ten rankings look fairly reasonable. The most surprising omissions for me on the women’s side are Clancy and Gottardi. A couple of questions:
1. What is the difference between receive errors and bad receives? The description isn’t clear.
2. In determining the overall ranking, you weighted each of the five categories equally, for simplicity I assume. Obviously you could do more analysis to determine the optimal weighting, but I’m curious which categories you think should count more in determining the best servers?
Clancy had an avg. rank of 28: 4th in aces, 3rd in generating receive errors, 31st in generating bad receives, 62nd in serve errors (22.5% serves were errors), and 41st in serve points won.
Gottardi had an avg. rank of 26: 2nd in aces, 4th in generating receive errors, 42nd in generating bad receives, 61st in serve errors (~18% serves were errors), and 19th in serve points won.
Both were just outside of the top 10, likely because of the serve errors. Clancy served about 34% of jump serves, though.
1. Receive errors mean the serve was not an ace but resulted in an immediate point. Bad receives mean that the receive was not a good pass but the point did not end. About 55% of bad receives result in a sideout.
2. Agree it's mostly for simplicity and also to use as much of the available data as possible compared to other sources. I could imagine grouping aces, receive errors, and bad receives together as 1 category. However, intuitively, I prefer weighting serve errors low at 20% (1/5th) vs. if it's 1 of only 2-3 categories. Serve points won also shows some preference to players who have good defense and transition offense (e.g. Ahman)
It appears that aces and receive errors are highly correlated. I believe some systems don’t even distinguish between the two. I don’t mind that aces + receive errors count twice as much as service errors here. If you serve even half as many aces + receive errors as service errors, that’s still a very good ratio because side out percentages on good passes are pretty high. In fact, I could argue that the risk-reward factor of aces vs service errors should actually rank players like Clancy and Gottardi higher than what you have.
I do think that points scored while serving may be weighted too heavily. Bad receives already accounts for good serves getting opponents out of system, plus as you mention points scored is affected by many other factors other than serving. But I’m not sure I’d remove it altogether either.
Hope you’ll continue the good work!
Good stuff again. The men’s top ten rankings look fairly reasonable. The most surprising omissions for me on the women’s side are Clancy and Gottardi. A couple of questions:
1. What is the difference between receive errors and bad receives? The description isn’t clear.
2. In determining the overall ranking, you weighted each of the five categories equally, for simplicity I assume. Obviously you could do more analysis to determine the optimal weighting, but I’m curious which categories you think should count more in determining the best servers?
Thanks! Appreciate the comment.
Clancy had an avg. rank of 28: 4th in aces, 3rd in generating receive errors, 31st in generating bad receives, 62nd in serve errors (22.5% serves were errors), and 41st in serve points won.
Gottardi had an avg. rank of 26: 2nd in aces, 4th in generating receive errors, 42nd in generating bad receives, 61st in serve errors (~18% serves were errors), and 19th in serve points won.
Both were just outside of the top 10, likely because of the serve errors. Clancy served about 34% of jump serves, though.
1. Receive errors mean the serve was not an ace but resulted in an immediate point. Bad receives mean that the receive was not a good pass but the point did not end. About 55% of bad receives result in a sideout.
2. Agree it's mostly for simplicity and also to use as much of the available data as possible compared to other sources. I could imagine grouping aces, receive errors, and bad receives together as 1 category. However, intuitively, I prefer weighting serve errors low at 20% (1/5th) vs. if it's 1 of only 2-3 categories. Serve points won also shows some preference to players who have good defense and transition offense (e.g. Ahman)