Slam and BP trends at Slams 1991-2013

This is based on data I have collected so far from ATP Match Facts. This R128 for all four slams from 1991-2013 (1990 is unreliable as is the time period before that).

Ace trend at different slams...



W on Grass, by far, has the highest Ace count, while Clay on RG is the lowest.

BP trend at different slams...



BPs faced is a measure of total Bps in a single match and is the sum of break points created by either player.

RG shows a strange aberration between 1999-2002, but seems to become similar to the other slams.

I will try and publish the raw data on Google Docs for public consumption.

I can either leave this at R128, and proceed to R64 and further at slams, or use specific tournaments, if suggested and desired.

Cheers.

Laverfan this basically confirms what I have been saying and the writer of the last piece has been saying. In fact, it goes further to confirm what I have been saying about the need for slowed down conditions with the ever increasing power, size, and technology these players play with. It basically shows that there is widespread differentiation to this day with the slower courts RG and AO, always being behind the USO and the leader in Ace counts wimbeldon. EVeryonone can see that there is a direct correlation. The faster the surface the more aces, no stop, non-fail at all the slams in order of Fastest to Slowest.

Additionally, it is interesting that ace counts have gone up in the era of slow conditions across the board in recent years. But the crucial thing is here that continually with few exceptions across the board you get a similar if not bigger differentiation between aces counts today at the slams then you do in the past. This is good research but it does not support the conclusion you are drawing. Also as in regards to bps faced the writer of the homogenization article talked about actual won breaks not break points faced.

Nothing changes the fact that, regardless of the cause, there is less variety on offer for the viewer across surfaces, the skill set required to win e.g. Wimby & FO is far more similar than it was, and that the physical aspect of the game if far more important than it was, and thus the ratio of required physical/non-physical attributes has changed.
That's all that matters, as far as I'm concerned.

socal1976 wrote:Laverfan this basically confirms what I have been saying and the writer of the last piece has been saying.

My analysis shows that Ace counts are NOT a measure of Surface speed. W was slowed down (apparently in 2001-2), but there is NO difference, in fact the Ace counts is increasing, despite the apparent slowdown. USO was apparently slowed down in 2002-3, but the Ace count next year jumped from 854 to 1011.  

socal1976 wrote:In fact, it goes further to confirm what I have been saying about the need for slowed down conditions with the ever increasing power, size, and technology these players play with.

Why should conditions be "slowed" down? Why this desire to kill variety? So there can be more 6+ hour matches on ALL surfaces?

Limit the power, technology. There has also been an argument that the court size needs to be modified to balance power, size and string tech.

socal1976 wrote:It basically shows that there is widespread differentiation to this day with the slower courts  RG and AO, always being behind the USO and the leader in Ace counts wimbeldon. EVeryonone can see that there is a direct correlation. The faster the surface the more aces, no stop, non-fail at all the slams in order of Fastest to Slowest.

RG is slow, that is Clay, but why slow down "faster" surfaces and dumb them down? Stuttgart switched from Clay to Grass.

Grass is inherently fast, and slowing it down just causes Grass players more problems. Leave it alone. If a player cannot play on grass (a la Muster), so be it.

socal1976 wrote:Also as in regards to bps faced the writer of the homogenization article talked about actual won breaks not break points faced.

You win a break point, only if you create it first, right?

In W 2010 MahIsner produced 206 aces.

laverfan wrote:

socal1976 wrote:Laverfan this basically confirms what I have been saying and the writer of the last piece has been saying.

My analysis shows that Ace counts are NOT a measure of Surface speed. W was slowed down (apparently in 2001-2), but there is NO difference, in fact the Ace counts is increasing, despite the apparent slowdown. USO was apparently slowed down in 2002-3, but the Ace count next year jumped from 854 to 1011.  

socal1976 wrote:In fact, it goes further to confirm what I have been saying about the need for slowed down conditions with the ever increasing power, size, and technology these players play with.

Why should conditions be "slowed" down? Why this desire to kill variety? So there can be more 6+ hour matches on ALL surfaces?

Limit the power, technology. There has also been an argument that the court size needs to be modified to balance power, size and string tech.

socal1976 wrote:It basically shows that there is widespread differentiation to this day with the slower courts  RG and AO, always being behind the USO and the leader in Ace counts wimbeldon. EVeryonone can see that there is a direct correlation. The faster the surface the more aces, no stop, non-fail at all the slams in order of Fastest to Slowest.

RG is slow, that is Clay, but why slow down "faster" surfaces and dumb them down? Stuttgart switched from Clay to Grass.

Grass is inherently fast, and slowing it down just causes Grass players more problems. Leave it alone. If a player cannot play on grass (a la Muster), so be it.

I haven't yet looked at the figures but it sounds like a really fine piece of research.

However, I have to take some issue with this post. If ace counts continue to rise that suggests that the slowing down of the courts is not that significant or, even more significantly, suggests that some element of slowing down was completely necessary to stop serve dominating. From Socal's post it seems that aces continue to be far more prevalent on Hard and Grass - surely evidence of a lack of homogenisation of courts. Hence the reason for the change in play style would seem to be likely to be unrelated to the courts.

Born Slippy wrote:

laverfan wrote:

socal1976 wrote:Laverfan this basically confirms what I have been saying and the writer of the last piece has been saying.

My analysis shows that Ace counts are NOT a measure of Surface speed. W was slowed down (apparently in 2001-2), but there is NO difference, in fact the Ace counts is increasing, despite the apparent slowdown. USO was apparently slowed down in 2002-3, but the Ace count next year jumped from 854 to 1011.  

socal1976 wrote:In fact, it goes further to confirm what I have been saying about the need for slowed down conditions with the ever increasing power, size, and technology these players play with.

Why should conditions be "slowed" down? Why this desire to kill variety? So there can be more 6+ hour matches on ALL surfaces?

Limit the power, technology. There has also been an argument that the court size needs to be modified to balance power, size and string tech.

socal1976 wrote:It basically shows that there is widespread differentiation to this day with the slower courts  RG and AO, always being behind the USO and the leader in Ace counts wimbeldon. EVeryonone can see that there is a direct correlation. The faster the surface the more aces, no stop, non-fail at all the slams in order of Fastest to Slowest.

RG is slow, that is Clay, but why slow down "faster" surfaces and dumb them down? Stuttgart switched from Clay to Grass.

Grass is inherently fast, and slowing it down just causes Grass players more problems. Leave it alone. If a player cannot play on grass (a la Muster), so be it.

I haven't yet looked at the figures but it sounds like a really fine piece of research.

However, I have to take some issue with this post. If ace counts continue to rise that suggests that the slowing down of the courts is not that significant or, even more significantly, suggests that some element of slowing down was completely necessary to stop serve dominating. From Socal's post it seems that aces continue to be far more prevalent on Hard and Grass - surely evidence of a lack of homogenisation of courts. Hence the reason for the change in play style would seem to be likely to be unrelated to the courts.

Exactly BS, these results confirm exactly what we would have expected. From the 90s til now without exception the slower slams produced less aces. The faster court slams USO and wimby produce more aces. Therefore, there is a direct correlation between surface speed and aces what Laverfan claims does not exist. In actuality these numbers support the idea that the slam surfaces are varied, and that ace differences are a good estimate of surface speed. Also what these numbers suggest is that despite the supposedly vast slowdown of conditions that ace numbers are on the up at the slams. What is the answer to this question, well simple that the changes in conditions are overstated. And two that other factors are involved in the baseline dominance of the modern game. IE the strings, racquets, and stronger athletes.

Interesting that I have been one of the only people to see something else that this research points to. In order to preserve the rallies at the power level of the modern tour we may very much need the slower conditions. Despite the slowdown that we know happened in the early 2000s at the USO and Wimbeldon the ace numbers continue to march upwards. If we changed conditions now, and put down technology restrictions that favored attack and power players the game would be much more fast paced than even what we saw in the 90s, and I remember how much people were complaining about how power and big serve tennis was ruining the game. So we are left in a conundrum, because of the sheer power of the modern male tennis player we may need slower conditions and the new technology. If we pushed these group of athletes into a game where big serving is easier and defensive tennis even harder we could rapidly see the Isnerization of the game.

Born Slippy wrote:However, I have to take some issue with this post. If ace counts continue to rise that suggests that the slowing down of the courts is not that significant or, even more significantly, suggests that some element of slowing down was completely necessary to stop serve dominating.

Serve dominated in the 80-90s, but that did not stop fans from watching Tennis. I am uncertain why such domination is considered a cancer to be exterminated? It dominated on "fast" surfaces, but did it dominate on Clay?

Why should the primary stroke of the server NOT dominate? The tech is there to return 130+mph serves now.

Born Slippy wrote:From Socal's post it seems that aces continue to be far more prevalent on Hard and Grass - surely evidence of a lack of homogenisation of courts. Hence the reason for the change in play style would seem to be likely to be unrelated to the courts.

In 1999, AO (892), RG (700), USO (947) and W(1140) so AO is 24%, RG is 19%, USO is 25%, and W is 32% of the total.

In 2010, AO (879), RG (955), USO (1151) and W(1587) so AO is 19%, RG is 20%, USO is 25% and W is 34% of the total.

Does this imply that AO got slower? Perhaps a MahIsner match (206 aces) is the aberration in W 2010.

We need to consider "homogenization" outside the context of aces and bps. The "style" of play is also "homogeneous", which has no statistical measure unless the length of a rally is such a measure.

socal1976 wrote:In order to preserve the rallies at the power level of the modern tour we may very much need the slower conditions. Despite the slowdown that we know happened in the early 2000s at the USO and Wimbeldon the ace numbers continue to march upwards. If we changed conditions now, and put down technology restrictions that favored attack and power players the game would be much more fast paced than even what we saw in the 90s, and I remember how much people were complaining about how power and big serve tennis was ruining the game.

If S&V died a natural death, why should "rallying" need to be preserved from "dying"?

socal1976 wrote:So we are left in a conundrum, because of the sheer power of the modern male tennis player we may need slower conditions and the new technology. If we pushed these group of athletes into a game where big serving is easier and defensive tennis even harder we could rapidly see the Isnerization of the game.

There are alternatives, like changing the dimensions of the court.

Serve still dominates today Laver. Karlovic is at 94% service games won this year. However, the snooze-feasts at Wimbledon were putting people off in the mid to late 90s as serve was too dominant. Rallying needs protecting from simple serve tennis.

The only way to get back serve volley tennis (not just serve tennis) would be to revert back to wooden racquets and I'm not sure even that would work. I don't see how changing the court dimensions would help - making the court bigger makes it easier to pass a net player and making the court smaller makes it easier for a baseliner to cover the court.

Born Slippy wrote:Serve still dominates today Laver. Karlovic is at 94% service games won this year. However, the snooze-feasts at Wimbledon were putting people off in the mid to late 90s as serve was too dominant. Rallying needs protecting from simple serve tennis.

Is Karlovic or Isner representative of the Tour? He has never been past the QF of a slam, IIRC. You now sound like SoCal. You have a choice to watch a specific style. There is an off button on your favourite electronic device. You or SoCal still have not mentioned one reason for "protecting" rallying, but have stated that it needs protecting. Again, if S&V suffered a demise, so should rallying, should it not?

If we do not want Isenrization, do you want Djokovicfication  or Nadalisation of Tennis? Individual preferences of styles should be allowed to thrive. Just like 606v2 allows variety of opinions.

Born Slippy wrote:The only way to get back serve volley tennis (not just serve tennis) would be to revert back to wooden racquets and I'm not sure even that would work. I don't see how changing the court dimensions would help - making the court bigger makes it easier to pass a net player and making the court smaller makes it easier for a baseliner to cover the court.

This should be a separate thread. If a spectator is allowed Twitter, Facebook, a Smart-always-connected-phone, why should a Tennis player be denied technology. Would you consider Boxing or T20 Cricket as a model for Tennis? Fixed number of serves (like fixed number of overs)? Basketball with a shot clock?

The same player can play either style as the surface demands. This is the variety that I am referring to. A "fast" and "slow" surface and styles that match the surface. Should this be preserved?

https://www.youtube.com/watch?v=YKLkOYiviN0

https://www.youtube.com/watch?v=elK42cXMM48

Variety vs "sameness" is what I consider the march towards "homogenization".

The "fast" surface has aces, the "slow" has rallies. Let the games continue with the variety.

Serving an ace is not a trivial exercise at the levels at which the professionals play.

laverfan wrote:

socal1976 wrote:In order to preserve the rallies at the power level of the modern tour we may very much need the slower conditions. Despite the slowdown that we know happened in the early 2000s at the USO and Wimbeldon the ace numbers continue to march upwards. If we changed conditions now, and put down technology restrictions that favored attack and power players the game would be much more fast paced than even what we saw in the 90s, and I remember how much people were complaining about how power and big serve tennis was ruining the game.

If S&V died a natural death, why should "rallying" need to be preserved from "dying"?

socal1976 wrote:So we are left in a conundrum, because of the sheer power of the modern male tennis player we may need slower conditions and the new technology. If we pushed these group of athletes into a game where big serving is easier and defensive tennis even harder we could rapidly see the Isnerization of the game.

There are alternatives, like changing the dimensions of the court.

Serve and volleying was losing momentum rapidly long before luxilon strings and slower courts. At the time of sampars already 90 percent of the tour played almost exclusively from the back. By the late 90s you had Henman, Rafter, and Sampras as the last of dying bread. I certainly am not up for radically altering he landscape of the game and risking what we have to bring back something that has been on the down slope since the advent of the graphite racquet. And there is a reason the tournaments that went for slower conditions did so, they realized fans are not excited by 2 shot rallies, and matches and broadcasters don't want events that get wrapped up in a hour and 15 minutes. They understand the finances of their business much better than we do. If two and three shot tennis was what the fans really wanted they would give it to them.

Genuinely not sure what you are trying to argue Laver. You can't seriously be arguing that any tournament should just be a serving competition? Having a big serve remains the most important shot in tennis but its possible to succeed without one - as it should be. However, having a big serve also remains far more important on grass than clay.

I have had a bit more chance to look at the research now. Very impressive piece of work LF. Just curious but how did you do this so quickly? The only way I can see to do it is to note down the match and then find it by searching by the relevant players and reviewing the matchfacts. I would have thought that process would take about 1 minute per match and you have reviewed 6,400 of them, which would be over 100 hours on that basis. Did you automate it in some way?

socal1976 wrote:Serve and volleying was losing momentum rapidly long before luxilon strings and slower courts. At the time of sampars already 90 percent of the tour played almost exclusively from the back. By the late 90s you had Henman, Rafter, and Sampras as the last of dying bread. I certainly am not up for radically altering he landscape of the game and risking what we have to bring back something that has been on the down slope since the advent of the graphite racquet. And there is a reason the tournaments that went for slower conditions did so, they realized fans are not excited by 2 shot rallies, and matches and broadcasters don't want events that get wrapped up in a hour and 15 minutes. They understand the finances of their business much better than we do. If two and three shot tennis was what the fans really wanted they would give it to them.

Neither are fans excited by longest 3-set matched (aka Shanghai) or 5-set matches of ~6 hours where players cannot even stand for the award ceremony.

The point that is being lost here is variety.

If you show up at a French restaurant expecting PIMS and Strawberries & Cream, Are you not in the wrong place to begin with? Did you pick the wrong restaurant?

Let there be fast surfaces with two shot rallies, and if you want to see 30+ shot war of attrition, go to the French Restaurant, or take a Spanish siesta.

Neither of you (BS/SoCal) have made a single argument for preserving rallying.

If it dies, let it die. NBA put in a shot clock, Cricket evolved from Tests to T20s, but Tennis scoring is still the same as 1887 (except the Van Ness system, favored by some slams).

There has been enough analysis, that HC and Clay are roughly similar in terms of tournament distribution. Grass is being killed slowly without a MS (or not).

At one point there were 3 slams on Grass and one on Clay, but now we have 2 on HC, 1 on Grass and 1 on Clay. And we want to make the 2 on HC look like the 1 on Clay. Why?

Born Slippy wrote:I have had a bit more chance to look at the research now. Very impressive piece of work LF. Just curious but how did you do this so quickly? The only way I can see to do it is to note down the match and then find it by searching by the relevant players and reviewing the matchfacts. I would have thought that process would take about 1 minute per match and you have reviewed 6,400 of them, which would be over 100 hours on that basis. Did you automate it in some way?

The method used is as follows. Like a web-browser, there is a Unix command-line (CLI) utility called curl ( aka cURL). I use it to point to the ATP web-site and extract draws (and adding now scores and games per set, total points, etc.). Once extracted, I use awk/sed/perl to extract ATP PlayerIDs (Nadal = N409, Federer = F324, etc.). Using such keys, I can go back to curl and extract match facts by year/slam (each slam has a unique tournament id).

The URLs  are of the following form...

# Wimbledon ---  url="http://www.atpworldtour.com/Share/Event-Draws.aspx?e=540&y=${yr}"; echo ${url}; curl "${url}"
# US Open --- url = "http://www.atpworldtour.com/Share/Event-Draws.aspx?e=560&y=${yr}"; echo ${url}; curl "${url}"
# AO --- url = "http://www.atpworldtour.com/Share/Event-Draws.aspx?e=580&y=${yr}"; echo ${url}; curl "${url}"
# RG --- url = "http://www.atpworldtour.com/Share/Event-Draws.aspx?e=520&y=${yr}"; echo ${url}; curl "${url}"

Having built the draws, this way, I can extract per match stats using a shell/curl script using a URL of the format

"http://www.atpworldtour.com/Share/Match-Facts-Pop-Up.aspx?t=580&y=2012&r=5&p=F401"

Using awk/sed/perl I can create pipe-delimited text files that end up in MS Excel, if desired, or Calc (OpenOffice).

This is driven by parameter files like

RG 520
W 540
USO 560
AO 580
Paris 352

and there is a list of years that I can put in a parameter file and run it across years that I want to collect the data from. I am using 1991-2013 as my year list currently. 1969-1973 is fraught with errors. I may need minor tweaks for R64...Finals to my scripts. I plan to publish these on Google docs in a couple of weeks, so posters can modify these freely.

I have been asking the ATP folks to let me have access to their back-end databases, but I get pushed back every time I try. The only way I know is to go work for Greg Sharko. I hope no one from ATPTennis reads this and shuts down my ability to cURL (which is not very likely, because that would mean monkeying with web pages as well).

May not be around much of the rest of the day, but will follow this on my mobile device.

Thanks Laver. Very interesting to see how you have done it.

(Updated with all four slams).

Based on this, I do not see any significant differences between RG and W (if 2010 MahIsner is discarded)

74 Aces in 2009 RG in this match - http://www.atpworldtour.com/Share/Match-Facts-Pop-Up.aspx?t=0520&y=2009&r=1&p=H432
84 Aces in 2005 W in this match - http://www.atpworltour.com/Share/Match-Facts-Pop-Up.aspx?t=0540&y=2005&r=1&p=B570

Taking one match in each year isn't an ideal method to assess the position. However, I would say those stats still strongly support an argument that (a) its easier to serve aces at Wimbledon and (b) it is as easy to serve aces at Wimbledon now as it was in the late 1990s.

Born Slippy wrote:Taking one match in each year isn't an ideal method to assess the position. However, I would say those stats still strongly support an argument that (a) its easier to serve aces at Wimbledon and (b) it is as easy to serve aces at Wimbledon now as it was in the late 1990s.

Neither is "matched" pairs across years to look at Aces/BPs vs surface speeds. .

The MaxAces shows that aces can be served consistently across surfaces. One surface is "faster" than "others", but many other factors get ignored in the Aces/BP/Surface Speed equation.

Laver - I don't get your arguments. You keep producing evidence which supports the paired matches article but claiming it does not. No-one has said you can't serve aces across all surfaces. The point is that in general more aces will be served on quicker surfaces.

Born Slippy wrote:Laver - I don't get your arguments. You keep producing evidence which supports the paired matches article but claiming it does not. No-one has said you can't serve aces across all surfaces. The point is that in general more aces will be served on quicker surfaces.

The argument in the OP is that "Homogenization" is not evident, because Aces/BPs should converge, if the surfaces were similar. The negative corollary of this statement is that if they do not converge, then there is no "homogenization".

Aces/Bps are completely unrelated and irrelevant to the "homogenization" argument. "Homogenization" is evident when the playing styles do not offer any diversity. There is very little attack left in the game as it is now. Patient rallies and opponent errors are now prevalent, whether it is due to surfaces or technology or advances in medical sciences can be debated.

If Aces can be served equally well on all surfaces and Bps created equally, then there is no correlation between them and the so called "homogenization" being derived from aces/bps. To use a Quantum Mechanics analogy, the "system" of Tennis is unperturbed by aces/bps, and hence aces/bps are a non-sequitur in this argument.

For 20+ years, if aces/bps are roughly similar, how can it be used to argue homogenization has or has not occurred. Matched "pairing" is fraught with ignoring the "human" aspect of a player. The assumption that all players play at the same level, day in day out is also a fallacy.

socal1976 wrote:
Here is a simple correlation that your own research proves.

1. The faster the court the more aces (assuming all else is equal)
2. The slower the court the less aces (assuming all else is equal)

The highlighted part is a crutch, and is mathematically impossible to prove. How is this "(assuming all else is equal)" a valid assumption?

Djokovic in 2008 @ MC served 2 bagel sets to two opponents.

http://www.atpworldtour.com/Tennis/Players/Top-Players/Novak-Djokovic.aspx?t=pa&y=2008&m=s&e=410#

Djokovic in 2014 has served 2 bagel sets to two opponents - Montanes and Carreno-Busta.

Can the following statement be derived from the above two observations -

Is Djokovic playing at the same level at 2014 MC as 2008 MC?

Laver, the data you have provided shows that more aces are served on quicker surfaces (compare the clay to grass). It also shows that the number of aces at each of the slams has remained relatively consistent. I am therefore at a loss to understand how you reach your conclusion.

Born Slippy wrote:Laver, the data you have provided shows that more aces are served on quicker surfaces (compare the clay to grass).

That is expected.

Born Slippy wrote:It also shows that the number of aces at each of the slams has remained relatively consistent. I am therefore at a loss to understand how you reach your conclusion.

I am equally at a loss at the hypothesis that Aces and Bps (having remained constant - as you state) can be used to derive "homogenization" or lack thereof.

Can a constant be used to define a variation?

I have purposefully hidden the column "E" in this image which is the surface of the tournament. It went through several changes.

Can it be determined,

1. What surface category (HC/Grass/Clay) this is? (The likelihood of a correct guess is 33.33% ...)? Are there any indications that the surface was manipulated to "slow" it down?

2. As the number of MaxAces is constant across 1991-2013, what conclusions can be drawn by the drop around 2000 in number of Aces?

Laver - its a fairly simple proposition. If aces have remained constant then it suggests that the surfaces have not slowed to a significant degree.

I don't know what tournament that is but there seem to have been fairly significant changes in 1994 and 2000?

Born Slippy wrote:Laver - its a fairly simple proposition. If aces have remained constant then it suggests that the surfaces have not slowed to a significant degree.

Are Aces a good measure of Surface speed?

Does the person across the net have any effect on a serve being an ace? (Matching pairs probably create a semblance of making the names of players a constant, but why is everything else not a factor and why are such factors  being ignored). A very well-known Professional player has said that all matches are different with score starting at 0-0 and I agree whole-heartedly.

For example, look at the  following two pieces of statistics (on the same surface - Clay and consecutive meetings between the same two players)...

http://www.atpworldtour.com/Share/Match-Facts-Pop-Up.aspx?t=416&y=2009&r=4&p=SA49

http://www.atpworldtour.com/Share/Match-Facts-Pop-Up.aspx?t=520&y=2009&r=4&p=SA49

Do the environmental conditions (apart from the Surface itself) have any impact on this? (BTW, do you play Golf?).

Born Slippy wrote:I don't know what tournament that is but there seem to have been fairly significant changes in 1994 and 2000?

I would like you to venture on what could have changed and how such changes impact your previous conjecture.

The OP is a very simplified model, but I should be allowed to question the validity of such model, should I not be?

Born Slippy wrote:Laver - its a fairly simple proposition. If aces have remained constant then it suggests that the surfaces have not slowed to a significant degree.

Unless other factors would have led to an increase in aces had the surfaces not slowed down.

Quite so - a point I have already made.

Laver, why don't you explain what you consider are the reasons for the changes in that particular tournament? The obvious answer is that the surface changed but I'm guessing that isn't the answer?

JuliusHMarx wrote:

Born Slippy wrote:Laver - its a fairly simple proposition. If aces have remained constant then it suggests that the surfaces have not slowed to a significant degree.

Unless other factors would have led to an increase in aces had the surfaces not slowed down.

Pretty accurate. Strings, Graphite racquets, better techniques, better fitness....

Born Slippy wrote:Laver, why don't you explain what you consider are the reasons for the changes in that particular tournament? The obvious answer is that the surface changed but I'm guessing that isn't the answer?

The tournament is Rotterdam...

A player named Ivanisevic shows up at Rotterdam, but his "aceing" capacity is not peak during all his years he played at Rotterdam. So the Ace count varies as the player goes through a natural life-cycle.

This is the Ace count for Ivanisevic in all his R1 matches at Rotterdam...



Same graph with more details. The surface tweaks start in 2001+ when Ivanisevic's ace output is dropping on the same surface. This is what I consider the "human" aspect of a player and matched pairs in the OP have not accounted for such variations at all.

I cannot categorically state that the surface tweaks are related to Ivanisevic (or similar players in Rotterdam). In 2009, for example, Karlovic shows up with 25 aces (but he loses to Llodra - 5 aces - in R1). Soderling serves 26 aces against Serra in 2010 and wins.

Also, 2001, onwards, the "surface" slowing can be seen. Greenset is usually a fast or medium-fast surface, Decoturf is medium-fast to medium, but can be medium-slow based on what is added to surface. I cannot be certain which version of Proflex was used in 2013, but it can vary between fast and medium-slow. It is unlikely to be a fast surface, IMHO.

TDs have the authority to choose the players and surfaces on which they play as long as they are ATP/ITF certified. As SoCal says, financial drivers go hand in hand with such decisions.

I am assuming that you can't find out the surfaces any earlier than 2001? I would be very interested to know what the surface was in 1999 and 2000. My only reservation about this data is that it seems to be a small data set per year - less than 16 matches? Figuring out the number of aces in the median match would be useful I think.

Ok, Wiki suggests a change from carpet to hard in 2000 which would explain the drop at that point.

It is 16 matches over a 23 year span at a single tournament, that is 368 matches. ATP draws that I looked at did not have surface before 2001, but WiKi is more than likely correct. Carpet was officially removed from ATP surfaces in 2009~. Shanghai TMC was played on Carpet for some years.

So, back to the tournaments themselves then, as it appears to be down to them. I got in touch with the ABN AMRO World Tennis Championships and asked about their decision to slow the courts down.

Dimitri Bonthuis from the ABN AMRO Championships explained the reasons thus: "We have changed from [surface speed rating] fast hardcourt 38 to fast hardcourt 25, making the court slower than before. This was done because last year the combination between ball and court simply proved too fast."

Too fast for whom, one wonders? The fans? Well, yes, according to Bonthuis, who continued: "Longer matches and longer rallies, simply meant more tennis, especially during the weekend, when only two matches are scheduled in a session."

And what about the players? What did they think of the new courts? "Almost all players found that the speed of the court was what it should be," Bonthuis said. "Some said it was a bit slow, but not necessary to make any changes."

And there we have the crux of the matter. "Almost all" the players were happy with it. I'm sure we all know the playing style of the ones who weren't.

http://betting.betfair.com/tennis/general/why-the-obsession-with-slower-courts-on-the-atp-tour-250212.html

I ignored the rest of the article, but the discussion with Bonthuis is rather enlightening.

Yes, I have seen that quote, made during the 2012 event I think. I wonder if the 2011 figures are a bit of an out-lier. There were very few three setters in the first round that year. It may be more appropriate to compare the 09-10 ace tallies to those of 12-13.

There are 15 3-setters in 2009-10 vs 14 in 2012-13, out of 32 R1 matches, so the ratio is the same.

Aces/BPs totals are 249/189 (2009), 215/200 (2010), 178/195 (2012) and 183/184 (2013), which looks like a "slowing" down to me.

A "long" time ago, so-called "Clay" specialists refused to play on W Grass, because it was considered "too fast" (recall Bothuis' remark). What if the TD wanted to get a "great" Clay player to show up and play and did not want to such a player to be thrown out in R1, or perhaps the reverse is true at Queens.

Recall Borg never did play a Grass warm-up.

Yes, I agree it does seem to show a slow-down. My point was that 2011 doesn't seem to fit and I wonder if that is because there were only 2 3-setters that year.

Of course, I have been saying all along is that aces vary based on court speed. Are we agreed that the fact there has been no similar decrease at Wimbledon seems to suggest the courts aren't noticeably slower?

From

1. W 2005, if we discard Braciali (33 Aces) v Karlovic (51 Aces) and take the next one which is Dent (23 Aces) v Norman (23 Aces) the MaxAces of 84 Aces drops to 46 Aces.

2. W 2010, if discard MahIsner (216 Aces) and take the next one, which is Ilhan (31 Aces) v Daniel (19 Aces) the MaxAces of 216 Aces drops to 50 Aces.

So let us see what happens when we add a trend line to this using a polynomial (we can use something else as well, but it may be a bit simpler to show it this way).



Nothing out of the ordinary for me as a hack on the court, but to someone like Henman ( http://www.atpworldtour.com/Tennis/Players/He/T/Tim-Henman.aspx?t=pa&y=0&m=s&e=540# ), this is what it looks like...



Henman's 13 years at W look like this (from an Aces/BPs perspective)....



The "human" aspect and the opponent across the net are parameters which need to be added to the equation definitely.

We both agree that faster surfaces should allow more aces, but so-called "minor" tweaks to surfaces, can cause issues at pro levels? Do you think that is a fair analysis?

BTW, there is an image that I saw, which shows that (this is MC 2014 on Clay), that the opponent has roughly 0.7 to 0.8 seconds to return the serve. I need to find that image.

Also, there is a follow-up to the OP... http://heavytopspin.com/2013/11/19/the-speed-of-every-2013-surface/