Nice article, as always. This can be solved with AI. An unforced error, should be based on a prediction model producing a probability of the average ATP player making the shot. If the probability exceeds a threshold (say 80%), then it's an unforced error. One could then adjust the threshold however one likes.
Serve + 1 seems a legitimate stat to capture effectiveness. It would be interesting if AI could track shot selection and come up with, essentially, a tennis IQ factor.
Great write up. And assessing an unforced error as one where the player attempts at a topspin shot is brilliant. Tennis is so complicated because it's not just one player that you're focusing on but what his opponent is doing as well. I'm interested in reading how your method plays out!
You haven't explicit said it, but I guess you are not ever counting a single point as two metrics? e.g. a unforced error and a winner no pressure?
I agree that what your stats would provide would be a lot more useful info, but I think there's still going to be subjectivity on a lot of points. I guess taking a stance of errors over winners makes all those 50:50 easier to come down on the error side.
I've never really understood why aces is such a quoted statistic, I guess there is a element of returner quality in URS but so many points are essentially errors forced by the serve, which is surely gives a fuller picture of how well someone served.
No, it is only ever counted in one of those metrics: "What is hard to do is draw the line between calling it an error under pressure for one player, rather than a winner for the other player (as I am counting winners for a player even if the opponent touches the ball). I draw the line (approximately) when the player attempts a topspin shot—like with Auger Aliassime below on his backhand—but is being rushed or pulled wide."
I think anytime you take a stance and try to create various categories of errors, subjectivity has to enter into it, but I still think doing it this way gets a better sense of which side and in which situations players were missing certain shots. We may see that a particular player misses a lot of forehands, but only when under pressure, and that tells us something that otherwise we wouldn't get.
i guess aces is a good proxy for overall serving effectiveness. The best servers hit a lot of aces. It breaks down when certain players—like Nadal—use their first serve more as a placement ball to set up their forehand on the next shot, so he isn't going for aces, but specifically targeting an area (such as Fed's backhand) and in that instance it helps to have a full picture (serve locations, aces, hold %, URS, etc.)
I typically agree with most of your analysis, and really enjoy your perspective.
You are WAY off base here and just plain incorrect. You cannot change the definition of a winner. If they touch the ball, it isn’t a winner. A winner, no pressure? I have assumed you played tennis at a high level, but a strong player would know that there are no “No pressure winners.”
How could a forehand put away be counted as a backhand winner? This makes no sense and no serious player or fan would ever consider using your metrics as match statistics.
As I wrote in the article, I propose to use the term "no pressure" to simply be opposite what I consider "under pressure winner". Perhaps I could say "low pressure"? It's really just semantics.
And the point of doing it this way is not to try and approximate what the numbers from TennisTV or Infosys will be (they aren't even close to each other anyway with UE's), or to make my stats the gold standard of tennis metrics, as I wrote in the last paragraph ("As much of the game is ultimately founded on topspin baseline consistency, I am trying to capture how well players are hitting their topspin shots from the baseline") I'm actually trying to capture something *different* from the typical analysis that makes no concession for position on the court, ease of shot, etc.
If you get a very easy ball on top of the net after hitting a great backhand down the line that did a lot of damage, I think there is an argument there that tallying that as a backhand winner conveys more accurate information of what really mattered in that point.
If the people reading my stats understand what gets tracked in certain boxes, does it matter whether I count Popyrin's backhand as a winner?
If the definition of a winner is so clear, how is it possible that tennistv and infosys get different numbers for winners as well?
My understanding of all this is that these stats are not to be heavily relied on (especially the unforced errors). So my rationale is, if I at least give people a template of how I'm tracking matches, they can make better sense of the numbers I would use.
Let me know what you think, and I appreciate honest feedback!
I am a fan of this idea, but would also like to see the most basic stats – winners and errors. What if both of these stats had an assisted by x shot metric? Volley winner assisted by DTL backhand or forehand error assisted by whatever shot the opponent hit well.
Nice article, as always. This can be solved with AI. An unforced error, should be based on a prediction model producing a probability of the average ATP player making the shot. If the probability exceeds a threshold (say 80%), then it's an unforced error. One could then adjust the threshold however one likes.
Would love that! I've also liked tennis insights who use data to come up with shot quality indexes.
https://www.atptour.com/en/news/insights-shot-quality
Serve + 1 seems a legitimate stat to capture effectiveness. It would be interesting if AI could track shot selection and come up with, essentially, a tennis IQ factor.
There are some stats that get serve +1 metrics I think.
Great write up. And assessing an unforced error as one where the player attempts at a topspin shot is brilliant. Tennis is so complicated because it's not just one player that you're focusing on but what his opponent is doing as well. I'm interested in reading how your method plays out!
You haven't explicit said it, but I guess you are not ever counting a single point as two metrics? e.g. a unforced error and a winner no pressure?
I agree that what your stats would provide would be a lot more useful info, but I think there's still going to be subjectivity on a lot of points. I guess taking a stance of errors over winners makes all those 50:50 easier to come down on the error side.
I've never really understood why aces is such a quoted statistic, I guess there is a element of returner quality in URS but so many points are essentially errors forced by the serve, which is surely gives a fuller picture of how well someone served.
No, it is only ever counted in one of those metrics: "What is hard to do is draw the line between calling it an error under pressure for one player, rather than a winner for the other player (as I am counting winners for a player even if the opponent touches the ball). I draw the line (approximately) when the player attempts a topspin shot—like with Auger Aliassime below on his backhand—but is being rushed or pulled wide."
I think anytime you take a stance and try to create various categories of errors, subjectivity has to enter into it, but I still think doing it this way gets a better sense of which side and in which situations players were missing certain shots. We may see that a particular player misses a lot of forehands, but only when under pressure, and that tells us something that otherwise we wouldn't get.
i guess aces is a good proxy for overall serving effectiveness. The best servers hit a lot of aces. It breaks down when certain players—like Nadal—use their first serve more as a placement ball to set up their forehand on the next shot, so he isn't going for aces, but specifically targeting an area (such as Fed's backhand) and in that instance it helps to have a full picture (serve locations, aces, hold %, URS, etc.)
I typically agree with most of your analysis, and really enjoy your perspective.
You are WAY off base here and just plain incorrect. You cannot change the definition of a winner. If they touch the ball, it isn’t a winner. A winner, no pressure? I have assumed you played tennis at a high level, but a strong player would know that there are no “No pressure winners.”
How could a forehand put away be counted as a backhand winner? This makes no sense and no serious player or fan would ever consider using your metrics as match statistics.
Hi Alex, firstly, thanks for your input!
As I wrote in the article, I propose to use the term "no pressure" to simply be opposite what I consider "under pressure winner". Perhaps I could say "low pressure"? It's really just semantics.
And the point of doing it this way is not to try and approximate what the numbers from TennisTV or Infosys will be (they aren't even close to each other anyway with UE's), or to make my stats the gold standard of tennis metrics, as I wrote in the last paragraph ("As much of the game is ultimately founded on topspin baseline consistency, I am trying to capture how well players are hitting their topspin shots from the baseline") I'm actually trying to capture something *different* from the typical analysis that makes no concession for position on the court, ease of shot, etc.
If you get a very easy ball on top of the net after hitting a great backhand down the line that did a lot of damage, I think there is an argument there that tallying that as a backhand winner conveys more accurate information of what really mattered in that point.
If the people reading my stats understand what gets tracked in certain boxes, does it matter whether I count Popyrin's backhand as a winner?
If the definition of a winner is so clear, how is it possible that tennistv and infosys get different numbers for winners as well?
My understanding of all this is that these stats are not to be heavily relied on (especially the unforced errors). So my rationale is, if I at least give people a template of how I'm tracking matches, they can make better sense of the numbers I would use.
Let me know what you think, and I appreciate honest feedback!
Agree. You want to know which shots are doing the actual damage in a match and not the shot that just happened to hit the put away ball.
I am a fan of this idea, but would also like to see the most basic stats – winners and errors. What if both of these stats had an assisted by x shot metric? Volley winner assisted by DTL backhand or forehand error assisted by whatever shot the opponent hit well.