Let's discuss stats for real (in clear format)

[AnOctorokForDinner]

Seeing that Waspsting's new stat threads are gathering deserved attention and a fair bit of meta stat discussion as well, I'd like to invite you guys to this specific thread: @Waspsting @abmk @krosero @Moose Malloy @WCT and anyone else who is interested.

Plain topics (ordered in descending order of difficulty, I guess):

1. What makes an error forced/unforced? What is the general difference between forced and unforced errors?

2. What makes a net point? When does a point count as a net point?

3. What makes a service winner? Where should the return go that the point can be counted as the server's winner rather than the returner's forced error?

Please feel free to have a go at any question here, thanks~

 

[AnOctorokForDinner]

My take:

1. The determining factor is whether the player who made the error had enough time since the incoming shot was hit to get into position to hit a safer shot. If they had sufficient time, and did try a safer shot, but missed, that's an UE, and if they could've hit it safer but chose to go for a riskier shot and missed, that's also an UE. The whole concept of forced errors is about making opponents short on time, I say.

Obviously, there's no mechanical algorithm to determine how much time is enough, but we know it's based on the players' positioning and the speed/path of the ball, and use our watching experience and observation to judge it. The sound of the ball colliding with the racquet helps measure by ear the time elapsed between the error-inducing shot and the error itself; visual input helps work out whether the player could have assumed a safer position by the time the ball got to them.

2. When a player crosses the service line to hit a shot and stays in the net area (not just happening to cross the line by a little and immediately running back), it's a net point for them, whether they approached of their own accord or were forced to go to net by a very short stroke, be that a drop shot, a short FH/BH or a shallow net cord.

When a player misses an approach shot and stops their forward movement just short of the service line, but would have evidently crossed it had their shot gone in, it still counts as an approach shot and therefore a failed net point in TA charting that I use. I think it makes sense for net stats to reflect not only volleying, but also approaches, both errors and winners (i.e. points when the player hit an approach shot, but didn't have to attempt a volley afterwards).

Finally, a special distinction for serving and volleying (since s&v serve is the only common case of an approach shot - in this case, the serve itself - being hit at the baseline; groundstroke approaches are almost exclusively hit from inside the court): when the server's forward movement is thwarted just short of the service line because of the sheer pace on the return forcing them to play a low volley / half-volley behind it, that's also a s&v point even though the volley / half-volley is played behind the service line, since the server would have clearly crossed the line on their volley if the return wasn't so big.

(At the same time, when the server stops halfway to the service line to see what the return is like and then acts accordingly, going to the net if it's a weak return and falling back if it's a deep one, that's definitely not a s&v point. Early/mid-prime Sampras would often do that, from what I've watched of him. I've seen Edberg punish him for that by making returns with average depth right behind the service line to Sampras's feet, forcing him to half-volley and then come to net, and approaching on a half-volley loses you points more often than not.)

3. This is the simplest one: I agree it makes good sense to count service winners on extreme return errors, so all that was left was to decide the criteria for 'extremity'. Eventually, I've decided on the following: if the ball lands wide or long by more than twice the width of the doubles alley (extreme wide/long error), or bounces before the service line (including behind the returner, duh), thus not even making it halfway to the net (extreme net error), I put that down as a service winner, otherwise it's a forced return error.


Think I outlined 1. and covered 2. and 3. as I see them. The UE/FE distinction is very complex and every charter relies on their own understanding/experience/court sense, but perhaps more detail can be worked out.

 

[krosero]

About ten years ago I got in touch with Leo Levin and asked him if balls tipped off the edge of the racquet were counted as winners in official stats. His reply was:

balls that are barely touched by the opponent are usually counted as
winners - if the opponent directs the ball back toward the net and the ball
reaches the net on the fly, the stroke is counted as a forced error not a
winner. Balls that are tipped - directed towards the side fences or stands
- are considered to be winners.

The goal of this was to appropriately award the resulting shot - i.e. was
the player able to make any kind of reasonable attempt to put the ball back
in play - if not, the result was a winning shot from the opponent.​

When Moose and I charted our matches, we counted only clean winners, for the most part; we wanted a uniform standard. On serve we did also use judgment calls, to count service winners. I sometimes used Levin's standard described above, to try to track the differences between official network stats and my own winner counts. Usually the official counts were slightly higher than mine, and by using Levin's standard I was usually able to account for those extra winners -- those shots that were just barely tipped off the racquet, and not just talking about serves, but all shots.

I think Levin's is a very common standard, but this was a while ago and I'm sure others will have at least slightly different takes on it.

One of many threads where this question was discussed: https://tt.tennis-warehouse.com/ind...007-w-qf-gasquet-roddick.204471/#post-2454001

 

[AnOctorokForDinner]

About ten years ago I got in touch with Leo Levin and asked him if balls tipped off the edge of the racquet were counted as winners in official stats. His reply was:

balls that are barely touched by the opponent are usually counted as
winners - if the opponent directs the ball back toward the net and the ball
reaches the net on the fly, the stroke is counted as a forced error not a
winner. Balls that are tipped - directed towards the side fences or stands
- are considered to be winners.

The goal of this was to appropriately award the resulting shot - i.e. was
the player able to make any kind of reasonable attempt to put the ball back
in play - if not, the result was a winning shot from the opponent.​

When Moose and I charted our matches, we counted only clean winners, for the most part; we wanted a uniform standard. On serve we did also use judgment calls, to count service winners. I sometimes used Levin's standard described above, to try to track the differences between official network stats and my own winner counts. Usually the official counts were slightly higher than mine, and by using Levin's standard I was usually able to account for those extra winners -- those shots that were just barely tipped off the racquet, and not just talking about serves, but all shots.

I think Levin's is a very common standard, but this was a while ago and I'm sure others will have at least slightly different takes on it.

One of many threads where this question was discussed: https://tt.tennis-warehouse.com/ind...007-w-qf-gasquet-roddick.204471/#post-2454001

I remember reading about that in one of your threads, yes. I agree it makes sense to count such framed shots that the player has no control over (in a way, I guess you could view them as 'forced shanks') as winners for the opponent, but not doing that also helps by removing one area of subjective judgment. TA charting technique that I use doesn't allow for counting non-clean winners except on the service, anyway. Given the tiny percentage of shots like that (except on the serve), however, it makes little difference whether one counts non-clean rally winners or not.

 

[Waspsting]

I like - and agree - with virtually everything you've said. In some areas, we think about the play differently rather than disagree

The only thing I outright do differently is I don't count failed attempts to take the net as a net point.

Its a tricky one and I can see it being interpreted both ways - net point or not a net point.... seeing it as such, I chose to err on the side of technical correctness: Hasn't crossed the service line, therefore, not a net point

I can also liken it to a serve-volley attempt on what turns out to be a fault... obviously, that's not a s/v point

But what a radical difference our way of doing stats can make! Look at the running Borg-Connors 1980 Masters thread https://tt.tennis-warehouse.com/ind...rg-vs-connors-masters-semi-final-1980.620354/

The way I've counted net points, baseline to baseline points are Connors 57, Borg 55 and net points on approach (i.e. sans serve volley) are Connors 21/42 @ 50%, Borg 12/20 @ 56%

Connors made 16 approach errors. The way you count it, the numbers would change to baseline to baseline, Connors 57, Borg 39 and net points to Connors 36%, Borg 56%

not sure if Borg made any approach errors, but the point is how drastic the shift can be. They've gone from even-stevens from the baseline to Connors thrashing Borg and at net, Connors has gone from below par to abysmal

1. What makes an error forced/unforced? What is the general difference between forced and unforced errors?

I look at a few things. Lets call it abc methodology and lets start with groundstrokes) -

- a) power of shot
- b) placement of shot (i.e. extent player had to move to reach it)
- c) depth of shot
-----
- d) error maker's shot choice

Taking a 'regulation shot' to be one to a ball of medium pace, average depth, doesn't require more than a couple steps to cover... how far from that situation is the shot on which the error has been made?

One factor alone being up from the norm is enough if its extreme (i.e. a very powerfully struck ball, a very deep ball or a ball placed right in the corner), but that's fairly rare.

Generally, you need at least two of a), b) and c) to be significantly up from the norm to make for a forced error

Any two factors at an extreme are a forced and three, of course... See any "greatest shots of all time" compilation.... those are shots to balls with abc all cranked up to extreme or near extreme

Regarding factor d)... a guy who errors on something due to shot selection might get an unforced from me. Say he's over near the side of the court, can comfortably put the ball in play crosscourt but goes down the line instead and nets it.... I might give that an unforced. More commonly, low percentage aggressive shots to balls that are up from the abc norm but not too much might get unforceds from me.

Virtually every runaround shot error is unforced... if you have time to runaround the shot, than you have time to play a normal shot, so if you miss... that's unforced

For volleys, I try to adapt the same basic structure as for groundstrokes.... but it's a bit like trying to put a round pole in a square peg

I see volleys as difficult, easy or makeable.... that last category covers a lot of ground and not as clear to judge forced or unforced as for groundstrokes (or maybe that's because we see so little volleying relative to groundstrokes these days?)

On groundstrokes, I can usually judge as I'm watching. One re-watch sometimes, 2 maximum

Volleys by contrast, I usually need a second viewing. .. as many as 5-6 even

Here I judge a bit algorithmicly, so to speak... and weigh up abc, if 2/3 of them were significantly up from regulation, I'll give it a forced

2. What makes a net point? When does a point count as a net point?

I agree with everything you said and as noted, we interpret failed approach attempts differently (both are legit though)

Adding on a bit.... the main thing to me about net points (non-forced ones, anyway) is that someone is looking to volley. People hit groundstrokes at net frequently but no one goes to net looking for it

The essence of a net point, and so the guiding principle in judging one to me is "player is looking for a volley"... This is my determining criteria when a volleys been made from behind the service line or a groundstroke has been made in front of it

3. What makes a service winner? Where should the return go that the point can be counted as the server's winner rather than the returner's forced error?

Ace - clean winner
I also give not-clean aces (same way we give not-clean winners) if the balls just snicks the returners racquet.... particularly if its just the frame and not strings

Service winners... no real fixed rules. Bouncing before the service line is a guideline. And I only give them for a top serve, not if the returner has made a hash of his shot

I diverge from common guidelines as far as 1st and 2nd serves go.... and use far more judgement along abc lines

You've got guys returning 80% first serves sometimes, so automatically calling a 1st serve return error forced seems a bit silly to me

And then you have some guys whose first serves are obviously not very forceful and guys whose standard 2nd serve is more forceful than them.... so I take it case by case, just being mindful of whether it's a first or second serve, and using that as a "tiebreaker" when it could go either way

balls that are barely touched by the opponent are usually counted as
winners - if the opponent directs the ball back toward the net and the ball
reaches the net on the fly, the stroke is counted as a forced error not a
winner. Balls that are tipped - directed towards the side fences or stands
- are considered to be winners.

The goal of this was to appropriately award the resulting shot - i.e. was
the player able to make any kind of reasonable attempt to put the ball back
in play - if not, the result was a winning shot from the opponent.

I do it like you - Winner = clean winner

Rare non-clean winners I give are along the principles for non-clean aces.... ball grazed the racquet frame. If strings are involved, I mark it a forced error

I have noted the problem with this though. Taking stats, I want to record in numbers the most significant shot

In case of what I call "flagrant forced errors" (where the guy did well to get a racquet on the ball and didn't have a cat in hells chance of actually getting it back).... that would be the shot that drew the error, not the error itself

The way we do it, that gets missed. Unlike Levin's way

What to do? We have to draw a line somewhere

Ace corresponds to winner....
Non-clean ace corresponds to non-clean winner....
What does service winner correspond to?? We'd need a new term

Informally, I call them flagrant forced errors but that misses the essence of the defining shot (the shot that drew the error)

 

[Waspsting]

Some confounding factors

1) Movement in judging forced/unforced

There was a Federer - Djokovic match where I gave Fed forceds because he was stretched or on the move when he made his shots... but Djoko was getting unforceds because he'd covered the ball and was stably balanced when he hits the shot

I notice Borg running to the ball quickly enough that he's set when he actually makes the shot.... so if he misses, it looks unforced

Is it? I tend to mark it so but that misses giving just due to the players movement

So the broader question. ...is the shot judged at the time it's made? Or at the time of the preceding shot?

I go with the former, but it's not perfect

2) Anticipation - McEnroe reads Connors serve perfectly. He moves to the return correctly during the ball toss

So if Mac misses a well placed serve that he'd read and gotten into comfortable position to return, is that forced? Or unforced?

Similar to 1) , I tend to go with unforced but that's not respecting Mac's anticipation (or Connors' serve for that matter)....It's holding McEnroe to his own standard, not a general one

Problematic

3) Style

Some players make things look easy, some don't... you need to be mindful of this when deciding forced/unforced

For example, most McEnroe volleys appear to be unforced because he looks so comfy playing them

Connors and Borg by contrast make all but put away volleys look like hard work

Be mindful of stuff like that and try to separate it from your judgement

@The Green Mile might find this discussion interesting

 

[AnOctorokForDinner]

The only thing I outright do differently is I don't count failed attempts to take the net as a net point.

...

I can also liken it to a serve-volley attempt on what turns out to be a fault... obviously, that's not a s/v point

I rather treat the fault points differently from other points, since that's the only kind of point where the opponent didn't have to do anything (an ace also means no play from the opponent, but they tried, and it partially depends on the returner's play as usually, except for the biggest and baddest serves, there's a small chance of at least getting a racquet on any serve provided the returner guesses correctly and reacts quickly, whereas with a service fault, the opponent could turn their back to the court and that wouldn't change anything), so giving a service fault or a double fault any additional categories seems unnecessary. That said, I understand counting failed s&v on double faults, that makes sense as well - but never on 1st serve faults, otherwise you could possibly get two s&v failures in one point (1st serve fault, 2nd serve fault), which ruins the count.

But what a radical difference our way of doing stats can make! Look at the running Borg-Connors 1980 Masters thread https://tt.tennis-warehouse.com/ind...rg-vs-connors-masters-semi-final-1980.620354/

The way I've counted net points, baseline to baseline points are Connors 57, Borg 55 and net points on approach (i.e. sans serve volley) are Connors 21/42 @ 50%, Borg 12/20 @ 56%

Connors made 16 approach errors. The way you count it, the numbers would change to baseline to baseline, Connors 57, Borg 39 and net points to Connors 36%, Borg 56%

not sure if Borg made any approach errors, but the point is how drastic the shift can be. They've gone from even-stevens from the baseline to Connors thrashing Borg and at net, Connors has gone from below par to abysmal

I think missed approaches should go into the net point column, since that was the intention. However, I think we should divide net points based on whether the incoming player hit or attempted to hit a shot at net (volley or sometimes a groundstroke) or not, since only the former reflects the success of his net play (net game = approaches + net play). So, when the approach (serve, groundstroke, volley / drive volley behind the service line) is a winner, an error-forcer, or an unforced error, that goes into 'net points won/lost purely on the approach', and if the opponent makes a passing shot attempt that the net player intercepts or tries to intercept with a volley, then it goes into 'net points that involved net play'. So you get four categories: s&v on unreturned serves (and DFs if you want to include them), s&v on returned serves, net points with unreturned approaches and approach UEs, net points with returned approaches (legit pass attempts).

Indeed, if Connors got the better of Borg in that match when neither player attempted to go to net, maybe he would have actually had greater success if he approached less and stayed back more.

I look at a few things. Lets call it abc methodology and lets start with groundstrokes) -

- a) power of shot
- b) placement of shot (i.e. extent player had to move to reach it)
- c) depth of shot
-----
- d) error maker's shot choice

<...>

Regarding factor d)... a guy who errors on something due to shot selection might get an unforced from me. Say he's over near the side of the court, can comfortably put the ball in play crosscourt but goes down the line instead and nets it.... I might give that an unforced. More commonly, low percentage aggressive shots to balls that are up from the abc norm but not too much might get unforceds from me.

Virtually every runaround shot error is unforced... if you have time to runaround the shot, than you have time to play a normal shot, so if you miss... that's unforced

In the end, it's all about time to get into position. Runaround shots require more time to get into position for than 'normal' shots, so of course nearly all of those errors are unforced unless the player is wrongfooted (e.g. player A gets a floater to rip through, player B, who is in the middle, must guess the direction and chooses to run to the BH corner, player A does hit to BH but softly, while player B expected a bigger, anglier shot and ran too fast, so the ball ends up behind them, forcing player B has to stop and turn 180, at which point it's already flying past so they have to swing with no preparation - that's when a technically runaround shot is a forced error).

Errors hit on the run (at more than half-speed) are always forced, that's easy since they are clearly out of position. The gray area is when the player was in just about a decent position to hit the ball, but not quite - not running fast, but not exactly stationary - how much out of position do they need to be for the error to count as forced? That's when the extra factors such as shot selection and opponent positioning come into play, and subjective judgment reigns supreme.

For volleys, I try to adapt the same basic structure as for groundstrokes.... but it's a bit like trying to put a round pole in a square peg

I see volleys as difficult, easy or makeable.... that last category covers a lot of ground and not as clear to judge forced or unforced as for groundstrokes (or maybe that's because we see so little volleying relative to groundstrokes these days?)

On groundstrokes, I can usually judge as I'm watching. One re-watch sometimes, 2 maximum

Volleys by contrast, I usually need a second viewing. .. as many as 5-6 even

Here I judge a bit algorithmicly, so to speak... and weigh up abc, if 2/3 of them were significantly up from regulation, I'll give it a forced

Volleys and 2nd serve returns are sure more difficult to judge than regular groundstrokes, often take several re-watches as you said. Again, I think about the time first and foremost, then the positioning - errors off heavily angled or awkwardly low returns/passes would be forced.

Ace - clean winner
I also give not-clean aces (same way we give not-clean winners) if the balls just snicks the returners racquet.... particularly if its just the frame and not strings

Service winners... no real fixed rules. Bouncing before the service line is a guideline. And I only give them for a top serve, not if the returner has made a hash of his shot

I don't think anyone else uses non-clean aces. That seems like an obfuscating distinction, to be frank. A clean winner on serve is an ace, a non-clean winner on serve is a service winner - that's what I follow along with everyone, it seems.

I diverge from common guidelines as far as 1st and 2nd serves go.... and use far more judgement along abc lines

You've got guys returning 80% first serves sometimes, so automatically calling a 1st serve return error forced seems a bit silly to me

And then you have some guys whose first serves are obviously not very forceful and guys whose standard 2nd serve is more forceful than them.... so I take it case by case, just being mindful of whether it's a first or second serve, and using that as a "tiebreaker" when it could go either way

Some 1st serve return errors are unforced, but that is really rare except for those with particularly poor serves, or if the server is injured and has to serve weak as a result. I know, though, that in earlier eras the likes of Connors and Borg would often spin their serves in not only on clay, but HC as well, so the number of 1st serve return UEs must have been higher with them.

<...>

In case of what I call "flagrant forced errors" (where the guy did well to get a racquet on the ball and didn't have a cat in hells chance of actually getting it back).... that would be the shot that drew the error, not the error itself

The way we do it, that gets missed. Unlike Levin's way

<...>

Informally, I call them flagrant forced errors but that misses the essence of the defining shot (the shot that drew the error)

As I chart for tennisabstract, the charting technique doesn't allow for non-clean winners except on the serve (service winners); however, you can still mark them by not filling in shot direction, like this:
'1st serve wide; forehand return crosscourt (shallow); backhand volley, forced error.'
This means that the volleyer couldn't even control the ball enough to direct it, much less make the volley. In this specific point, as I remember, the volleyer caught the ball with the tip of the racquet (edge of the stringbed right under the frame), and it flew away from court. I chart with the idea that most of such directionless errors could be considered non-clean winners; other charters may have a different understanding, naturally.

 

[AnOctorokForDinner]

Some confounding factors

1) Movement in judging forced/unforced

There was a Federer - Djokovic match where I gave Fed forceds because he was stretched or on the move when he made his shots... but Djoko was getting unforceds because he'd covered the ball and was stably balanced when he hits the shot

<...>

2) Anticipation - McEnroe reads Connors serve perfectly. He moves to the return correctly during the ball toss

So if Mac misses a well placed serve that he'd read and gotten into comfortable position to return, is that forced? Or unforced?

<...>
Problematic

As it is written, what's a forced error for Isner can well be an unforced error for Djokovic. Indeed, if Djokovic got into position where most players wouldn't, then it's an unforced error for him, as the preceding shot wasn't good enough to put him out of balance even if it would be good enough for an average player. Stats don't reflect everything, gotta make notes as you do to describe match trends and specifics.

3) Style

Some players make things look easy, some don't... you need to be mindful of this when deciding forced/unforced

For example, most McEnroe volleys appear to be unforced because he looks so comfy playing them

Connors and Borg by contrast make all but put away volleys look like hard work

Be mindful of stuff like that and try to separate it from your judgement

@The Green Mile might find this discussion interesting

I pay more attention to players' movement to counter that. McEnroe or Edberg may make volleying look easy, but the actual movement is still very complex and gives you the right idea.

 

[KG1965]

Some confounding factors

1) Movement in judging forced/unforced

There was a Federer - Djokovic match where I gave Fed forceds because he was stretched or on the move when he made his shots... but Djoko was getting unforceds because he'd covered the ball and was stably balanced when he hits the shot

I notice Borg running to the ball quickly enough that he's set when he actually makes the shot.... so if he misses, it looks unforced

Is it? I tend to mark it so but that misses giving just due to the players movement

So the broader question. ...is the shot judged at the time it's made? Or at the time of the preceding shot?

I go with the former, but it's not perfect

2) Anticipation - McEnroe reads Connors serve perfectly. He moves to the return correctly during the ball toss

So if Mac misses a well placed serve that he'd read and gotten into comfortable position to return, is that forced? Or unforced?

Similar to 1) , I tend to go with unforced but that's not respecting Mac's anticipation (or Connors' serve for that matter)....It's holding McEnroe to his own standard, not a general one

Problematic

3) Style

Some players make things look easy, some don't... you need to be mindful of this when deciding forced/unforced

For example, most McEnroe volleys appear to be unforced because he looks so comfy playing them

Connors and Borg by contrast make all but put away volleys look like hard work

Be mindful of stuff like that and try to separate it from your judgement

I think it is a problem that is difficult to solve because it is very relative, subjective.

An ace is an ace. Easy. Also on the ace could open a discussion ... because a very strong and well placed serve may not be ace and a less strong serve could seize unprepared a player not very snappy and good ...

To analyze an error forced by an unforced one enters the field ... with mines.