When do players peak? (The Definitive Answer - with Charts!)

[El Dude]

Well, I did some research and made some charts. I took two approaches, with two charts each. The first two look at big titles, the second two by Elo Rating.

Chart 1: Open Era Big Titles By Age

This first one is pretty straightforward. It depicts every big title in the Open Era, by the age of the title winner. I also differentiated between various degrees of greatness with the different shades of green.

As you can see, age 24 is the "ultimate peak" age for big title wins, with 22-25 being a four-year peak, and a longer prime of about 20 to 31, with decline after. It is worth noting that 16 of 22 Slams won at age 32 or older were by the Big Three. The other six include four by Ken Rosewall, one by Andre Agassi, and one by Andres Gimeno.

Chart 2: Big Title "Points"
This chart depicts the same data, but assigns point values to give a clearer sense of relative strength of each age.

As you can see, once again age 24 is shown as the absolute peak, with age 21-29 being prime dominance. We also see the decline trend being a bit more steady after age 24, falling in "two-year steps": age 28-29, 30-31, and then plateauing at 32-35 (as the Big Three remained strong; most other players declined and retired during that time).

Chart 3: Peak Elo Ratings By Age
This chart depicts the age a player was at their Peak Elo Rating, including all players of 2100 Peak Elo or higher (which roughly equates with a top 10 Elo rating in a given year).

First, a note: I excluded younger players (born in 1995 or later) who may not yet have peaked, as well as those players who probably peaked before the Open Era. But I included them, just so you can see where they are so far.

This one offers a slightly different perspective, focused less on the production of players and more looking at when individual players reached their peak form. We can see a clear peak range between 22-27, especially age 25-27, with a sharp drop at 28-29, then again at 30, and plummeting at 31.

Meaning, this chart shows us that the vast majority of players reach their peak at age 21-30, especially 22-27.

Chart 4: Peak Elo Ratings By Age and Generation

This last chart is the same as Chart 3, but adds in generational colors (or five-year birth ranges).

The key data here is the number lines below. What it tells us is that absolute peaks starter out around 27, gradually went down and reached a low with the 1964-68 generation (Wilander, Edberg, Becker, etc), then jumped with Federer's generation (1979-83), rising slightly higher with the Nadal/Djokovic cohort. The proverbial "Lost Gen" (1989-93) has started to drift younger, but given the few players of that generation represented ((just 7).

Conclusions
What these charts show, is a few things:

• Over the course of the Open Era, most players peak in their mid-20s, with age 24 being the absolute peak according to big titles and age 25 according to Elo.
• In terms of big titles won, players reach their absolute peak at 24-25, then start to gradually and steadily decline, with a leveling off in the late 20s to early 30s at a lower "plateau" level, before declining further in their 30s.
• The vast majority of players don't reach their peak until age 22 or later, though rarely after age 27; thus 22-27 is the most common range for peaking (according to Elo).
• Peak age got younger from the beginning of the Open Era until around the turn of the century, then trended sharply upward with Federer's generation and later players (i.e. those born in the 1980s). Meaning, the generation that peaked in the early to mid 70s (Nastase, Smith, Okker, Newcombe, etc) peaked around age 27; this dropped to age 26 with the next gneeration, then was around age 24 for the next five generations--or those players peaking from the late 70s until the mid-00s, after which it jumped sharply back up to age 26-27 for the next two generations.
• It is too soon to tell whether that trend will continue; we should know more in a few years, once Next Gen (turning 25-29 this year) is through those peak years (that is, all mostly older than 27 or so).

Meaning, nothing remarkably new, though the whole question, I think, has been somewhat definitively clarified and answered.

 

[Jelenafan]

Well, I did some research and made some charts. I took two approaches, with two charts each. The first two look at big titles, the second two by Elo Rating.

Chart 1: Open Era Big Titles By Age

View attachment 7907


This first one is pretty straightforward. It depicts every big title in the Open Era, by the age of the title winner. I also differentiated between various degrees of greatness with the different shades of green.

As you can see, age 24 is the "ultimate peak" age for big title wins, with 22-25 being a four-year peak, and a longer prime of about 20 to 31, with decline after. It is worth noting that 16 of 22 Slams won at age 32 or older were by the Big Three. The other six include four by Ken Rosewall, one by Andre Agassi, and one by Andres Gimeno.

Chart 2: Big Title "Points"
This chart depicts the same data, but assigns point values to give a clearer sense of relative strength of each age.

View attachment 7908
As you can see, once again age 24 is shown as the absolute peak, with age 21-29 being prime dominance. We also see the decline trend being a bit more steady after age 24, falling in "two-year steps": age 28-29, 30-31, and then plateauing at 32-35 (as the Big Three remained strong; most other players declined and retired during that time).

Chart 3: Peak Elo Ratings By Age
This chart depicts the age a player was at their Peak Elo Rating, including all players of 2100 Peak Elo or higher (which roughly equates with a top 10 Elo rating in a given year).
View attachment 7909

First, a note: I excluded younger players (born in 1995 or later) who may not yet have peaked, as well as those players who probably peaked before the Open Era. But I included them, just so you can see where they are so far.

This one offers a slightly different perspective, focused less on the production of players and more looking at when individual players reached their peak form. We can see a clear peak range between 22-27, especially age 25-27, with a sharp drop at 28-29, then again at 30, and plummeting at 31.

Meaning, this chart shows us that the vast majority of players reach their peak at age 21-30, especially 22-27.

Chart 4: Peak Elo Ratings By Age and Generation

View attachment 7911

This last chart is the same as Chart 3, but adds in generational colors (or five-year birth ranges).

The key data here is the number lines below. What it tells us is that absolute peaks starter out around 27, gradually went down and reached a low with the 1964-68 generation (Wilander, Edberg, Becker, etc), then jumped with Federer's generation (1979-83), rising slightly higher with the Nadal/Djokovic cohort. The proverbial "Lost Gen" (1989-93) has started to drift younger, but given the few players of that generation represented ((just 7).

Conclusions
What these charts show, is a few things:

• Over the course of the Open Era, most players peak in their mid-20s, with age 24 being the absolute peak according to big titles and age 25 according to Elo.
• In terms of big titles won, players reach their absolute peak at 24-25, then start to gradually and steadily decline, with a leveling off in the late 20s to early 30s at a lower "plateau" level, before declining further in their 30s.
• The vast majority of players don't reach their peak until age 22 or later, though rarely after age 27; thus 22-27 is the most common range for peaking (according to Elo).
• Peak age got younger from the beginning of the Open Era until around the turn of the century, then trended sharply upward with Federer's generation and later players (i.e. those born in the 1980s). Meaning, the generation that peaked in the early to mid 70s (Nastase, Smith, Okker, Newcombe, etc) peaked around age 27; this dropped to age 26 with the next gneeration, then was around age 24 for the next five generations--or those players peaking from the late 70s until the mid-00s, after which it jumped sharply back up to age 26-27 for the next two generations.
• It is too soon to tell whether that trend will continue; we should know more in a few years, once Next Gen (turning 25-29 this year) is through those peak years (that is, all mostly older than 27 or so).

Meaning, nothing remarkably new, though the whole question, I think, has been somewhat definitively clarified and answered.

Some interesting data, Eldude, great research.

Going now to the current state of mens tennis, I think for me its not as important when a player “peaks” as much as when they finally break through to actually have tangible winning results. Most champs who have a peak have to get near the summit first…

Piggybacking on the NextGen you referred to and their age range, here are the ATP current top 20:
Now throwing out the older age outliers ( Djokovic, Nadal,Carreno) & the younger age outliers (Alcaraz, Rune & Sinner)

These are the specific ages the other top 20 players have turned or will turn this year:

Cameron Norrie 28
Karen Kachanov 27
Danill Medvedev 27
Borna Coric 27
Hubert Hurcacz 26
Alexander Zverev 26
Andrei Rublev 26
Taylor Fritz 26
Tommy Paul 26
Francis Tiafoe 25
Stefano Tsitsipas 25
Casper Ruud 25
Alex de Minaur 24
Felix AA 23

For added measure Ive added a once highly touted younger player

Dennis Shapavolov 24

**********
As you mentioned many are aging out, so
My guess is these guys hope the new standard of peaking is Stan Wawrinka ; )

 

[El Dude]

As you mentioned many are aging out, so
My guess is these guys hope the new standard of peaking is Stan Wawrinka ; )

Haha, yeah right. Maybe Stan opened the door, but I highly doubt that we're going to see players peak at age 28-31 in droves. Maybe a few - and there have always been some who peak later, but they are outliers. But certainly, when one guy does something, it opens the door for others. And when you couple the timing of the Big Three aging out, maybe Next Gen will win their most titles in their late 20s. But that doesn't mean "peak," necessarily - as far as form. That's why I took two angles: the big titles represent accomplishments, but peak Elo tells us when a player was at their very best. These don't always correlate, especially when we're talking players with fewer Slams (e.g. Goran Ivanisevic was in his peak in his early to mid 20s, but won his lone Slam at 29).

That said, where historically 22-27 is especially peak (according to Elo), maybe it is now 24-29 or so. But I think we'll need to see when and how Next Gen declines to get a better sense of that.

On the other hand, Alcaraz and Rune muddied that hypothesis a bit. They said, "Look, we're breaking through and winning big titles as teenagers - like great players often have done." Meaning, maybe those two are the first true (potential) greats we've seen since Djokovic/Murray/Del Potro.

There are other angles to this that I could look at, such as the gap between first title (or first big title) and peak Elo. That would be interesting in terms of speculating on when players who have won titles (or big titles) might be expected to reach their highest form.

 

[Jelenafan]

....but peak Elo tells us when a player was at their very best. These don't always correlate, especially when we're talking players with fewer Slams (e.g. Goran Ivanisevic was in his peak in his early to mid 20s, but won his lone Slam at 29)...

Alexander Zverev to me is an interesting case, he's turning 26 this year and has 1 Olympic Gold medal in singles, 2 YE ATP’s, 10 Masters finals & 5 Master titles, 19 singles titles, 1 Major final, 4 Major SF’s, and 4 Major QF’s and reached #2 in the ranking. He won first two Masters back in 2017, and in 2021 he again won 2 Master titles in addition to the YE ATP, Olympic singles, best overall results for all 4 Majors in a single year, and had a career high 6 titles and was at his highest YE ranking at #3. Of course the injury at the '22 FO may have derailed things a lot going forward. : (

So Arguably Zverev may have already peaked circa 2021 and age 24 . But to your point, if he ever wins that elusive Major it would be an accomplishment achieved as opposed to finally reaching his best form at, say, age 26, 27, 28 etc.

 

[Nadalfan2013]

Nadal hasn't peaked yet, the best is yet to come! French Open #15 upcoming with many bagels and then finally will be healthy at Wimbledon and reclaim his throne, followed by a 5th US Open! Thank you for reading, have a wonderful day.

 

[El Dude]

Nadal hasn't peaked yet, the best is yet to come! French Open #15 upcoming with many bagels and then finally will be healthy at Wimbledon and reclaim his throne, followed by a 5th US Open! Thank you for reading, have a wonderful day.

 

[El Dude]

Alexander Zverev to me is an interesting case, he's turning 26 this year and has 1 Olympic Gold medal in singles, 2 YE ATP’s, 10 Masters finals & 5 Master titles, 19 singles titles, 1 Major final, 4 Major SF’s, and 4 Major QF’s and reached #2 in the ranking. He won first two Masters back in 2017, and in 2021 he again won 2 Master titles in addition to the YE ATP, Olympic singles, best overall results for all 4 Majors in a single year, and had a career high 6 titles and was at his highest YE ranking at #3. Of course the injury at the '22 FO may have derailed things a lot going forward. : (

So Arguably Zverev may have already peaked circa 2021 and age 24 . But to your point, if he ever wins that elusive Major it would be an accomplishment achieved as opposed to finally reaching his best form at, say, age 26, 27, 28 etc.

Yes, agreed all around. For whatever reason, I'm reminded of the career arc of Goran Ivanisevic, who was arguably the best player of the 90s not to win a Slam during that decade - at least in terms of total career accomplishments (meaning, maybe Rios was better at his best, but it was literally only a year or two).

Goran's peak Elo was in 1992 at age 21, though his Elo prime extended through 1997 or so, when he turned 26. He started declining after, partially due to a shoulder injury, and bottomed out in early 2001, then improbably won Wimbledon as a wildcard - beating an impressive list of players, including Moya, Roddick, Rusedski, Safin, Henman, and Rafter. He finished the year #12, but then collapsed and retired the next year.

I'm not saying that "Sascha" will have a similar story, just that--like Goran--he was a very accomplished Slamless player for quite awhile and, quite frankly, Sascha is too good not to win a Slam. As you said, he turns 26 in a few weeks so presumably has at least a few good years left in him. His peak Elo of 2303 is #30 in the Open Era; only four players above him didn't win a Slam: Ferrer, Gottfried, Okker, and Gene Mayer, the first three of whom are always mentioned in conversations about the best Slamless players.

He's also one of only ten players to win the Tour Finals more than once (not counting the Alt Tour Finals like the WCT Finals and Grand Slam Cup). The "worst" players who won it 2+ times are Hewitt and Nastase, both of whom could be considered lesser greats. And he's one of 24 players to win 8 or more big titles, tied with Chang, Vilas, and Ashe.

 

[Nadalfan2013]

100% sure if he is healthy.

 

[El Dude]

Interesting Post. Felix played absolutely Brilliant at RG and AO 2022.. A couple of lucky bounces of the ball and FAA could have been a 2 time GS champion. I think something has went wrong at the GSs since which I feel is between Felix's ears . Maybe it's the primary coach that needs to go and Toni needs to be promoted to primary or switch their roles . Remember Felix had an absolute brilliant post US open campaign on the European hard courts.. Just my two cents worth. @El Dude your three cents worth

One thing that might be a factor--even explain--Felix's "lack of fire" (to quote Kieran) is that he has a chronic heart condition. I believe he had surgery to correct it some years back, but I do wonder if that plays a part.

I see a guy with a similar display of raw talent and range of skills as Sinner and Alcaraz, but lacking that "it" factor (contrast him to Holger Rune, who is more rough but definitely has the "it" factor). I'm not sure what it would take to get him to that next level, but as I wrote elsewhere, he's kind of missed his window for "the path to greatness," at least in terms of historical Elo ratings. Now just as Stan set new precedents for multi-Slam winners and "near greats," presumably someone comes along and doesn't fully blossom until their mid-20s and ends up becoming an all-time grat. But even Ivan Lendl, who didn't win his first Slam until age 24, made a huge jump before--and was an elite player by--his 21st birthday (then continued from there).

Or consider that, along the lines of my study in another thread, every all-time great (6+ Slam winners, plus Murray) had surpassed the 2200 Elo threshold before their 22nd birthday...Felix's peak is still at 2140 and he's 22, going to be 23 in August.

So again, who knows, maybe he sets new precedents like Stan did, and makes a big jump over the next year or two, but right now he's more on a path that sees him becoming somewhere between Gasquet/Dimitrov and Medvedev/Thiem. No shame in that, but not quite at the level of all-time greats.

 

[Kieran]

One thing that might be a factor--even explain--Felix's "lack of fire" (to quote Kieran) is that he has a chronic heart condition. I believe he had surgery to correct it some years back, but I do wonder if that plays a part.

I see a guy with a similar display of raw talent and range of skills as Sinner and Alcaraz, but lacking that "it" factor (contrast him to Holger Rune, who is more rough but definitely has the "it" factor). I'm not sure what it would take to get him to that next level, but as I wrote elsewhere, he's kind of missed his window for "the path to greatness," at least in terms of historical Elo ratings. Now just as Stan set new precedents for multi-Slam winners and "near greats," presumably someone comes along and doesn't fully blossom until their mid-20s and ends up becoming an all-time grat. But even Ivan Lendl, who didn't win his first Slam until age 24, made a huge jump before--and was an elite player by--his 21st birthday (then continued from there).

Or consider that, along the lines of my study in another thread, every all-time great (6+ Slam winners, plus Murray) had surpassed the 2200 Elo threshold before their 22nd birthday...Felix's peak is still at 2140 and he's 22, going to be 23 in August.

So again, who knows, maybe he sets new precedents like Stan did, and makes a big jump over the next year or two, but right now he's more on a path that sees him becoming somewhere between Gasquet/Dimitrov and Medvedev/Thiem. No shame in that, but not quite at the level of all-time greats.

I didn’t know about the heart condition. It’s under control now, I’m sure, but it’s probably in his mind somewhere.

He could definitely spurt. He reminds me often of Sampras in his demeanour, and Pete took a few seasons to assimilate his status as a USO champ. Jimmy Connors I remember being particularly scathing about Pete, but suddenly it all clicked again for Pete, and he surged ahead of everyone. It’s not impossible that Felix could do the same..

 

[El Dude]

I didn’t know about the heart condition. It’s under control now, I’m sure, but it’s probably in his mind somewhere.

Yeah, that's what I'm thinking. I've been blessed without major illness or injury in my life, but in the minor to moderate ones, I know there's a lingering "tentativeness."

He could definitely spurt. He reminds me often of Sampras in his demeanour, and Pete took a few seasons to assimilate his status as a USO champ. Jimmy Connors I remember being particularly scathing about Pete, but suddenly it all clicked again for Pete, and he surged ahead of everyone. It’s not impossible that Felix could do the same..

Yeah, I agree. But in terms of likely outcomes, unfortunately the odds veer more towards Gasquet/Dimitrov than Sampras. Of course there's a huge range in-between, and one of the fun things I look for and take interest in, is when players take different trajectories than before. Like Stan. I could see Auger-Aliassime being that sort of unique player.

 

[Kieran]

Yeah, that's what I'm thinking. I've been blessed without major illness or injury in my life, but in the minor to moderate ones, I know there's a lingering "tentativeness."

Plus there’s the obvious trigger warnings when uncle Toni yells, Vamos kid, show some de hart!

And Felix is like, seriously? You went there?

Yeah, I agree. But in terms of likely outcomes, unfortunately the odds veer more towards Gasquet/Dimitrov than Sampras. Of course there's a huge range in-between, and one of the fun things I look for and take interest in, is when players take different trajectories than before. Like Stan. I could see Auger-Aliassime being that sort of unique player.

Yeah, that sounds about right, but I’ll always miss a player with the cat-like predatory presence of Pete and Felix is the nearest I’ve seen…

 

[El Dude]

Yeah, that sounds about right, but I’ll always miss a player with the cat-like predatory presence of Pete and Felix is the nearest I’ve seen…

Interesting comparison. He is fun to watch. In addition to not quite having the heart of, say, Alcaraz, he also seems to be more error prone - but I haven't checked the stats.

Anyhow, I do think he'll end up with a better career than Gasquet or Dimitrov. I mean, he's still only just 22 and already looks well compared to them:

Gasquet: 16 titles, all 250s; Slams: 3 SFs, 2 QFs: peak Elo 2196, peak ATP rank 7.
Dimitrov: 8 titles, 1 Tour Final, 1 Masters, 1 500; Slams: 3 SFs, 3 QFs; peak Elo 2203, peak ATP rank 3.
Auger-Aliassime: 4 titles, 2 500s; Slams: 1 SF, 2 QFs; peak Elo 2140, peak ATP rank 6.

In other words, if he turns out like Gasquet or Dimitrov, it will mean he doesn't improve a huge amount beyond where he's at now (2140 Elo vs 2200ish). At least Grigor got those two big titles in 2017...poor Richard never even won an ATP 500. I would be very surprised if Felix doesn't win at least a few Masters, and still think he's got at least one Slam in him.

My current projection for Felix is as someone who wins a handful of big titles, maybe a Slam or two, and is a regular in the top 10, frequently reaching the top 5. Maybe a Yannick Noah level player? I think there's a sizable chance he ends up more like Tsonga/Berdych, with Gasquet/Dimitrov being the low end. And a small chance he blooms into a bonafide #1 and wins 2-3 Slams, but not a true great. But we shall see.

 

[Kieran]

Interesting comparison. He is fun to watch. In addition to not quite having the heart of, say, Alcaraz, he also seems to be more error prone - but I haven't checked the stats.

Anyhow, I do think he'll end up with a better career than Gasquet or Dimitrov. I mean, he's still only just 22 and already looks well compared to them:

Gasquet: 16 titles, all 250s; Slams: 3 SFs, 2 QFs: peak Elo 2196, peak ATP rank 7.
Dimitrov: 8 titles, 1 Tour Final, 1 Masters, 1 500; Slams: 3 SFs, 3 QFs; peak Elo 2203, peak ATP rank 3.
Auger-Aliassime: 4 titles, 2 500s; Slams: 1 SF, 2 QFs; peak Elo 2140, peak ATP rank 6.

In other words, if he turns out like Gasquet or Dimitrov, it will mean he doesn't improve a huge amount beyond where he's at now (2140 Elo vs 2200ish). At least Grigor got those two big titles in 2017...poor Richard never even won an ATP 500. I would be very surprised if Felix doesn't win at least a few Masters, and still think he's got at least one Slam in him.

My current projection for Felix is as someone who wins a handful of big titles, maybe a Slam or two, and is a regular in the top 10, frequently reaching the top 5. Maybe a Yannick Noah level player? I think there's a sizable chance he ends up more like Tsonga/Berdych, with Gasquet/Dimitrov being the low end. And a small chance he blooms into a bonafide #1 and wins 2-3 Slams, but not a true great. But we shall see.

Yeah I agree. By the way, I don’t know what ELO means. I know chess players are ranked according to an ELO, but can you give a brief description? Cheers!

 

[El Dude]

Yeah I agree. By the way, I don’t know what ELO means. I know chess players are ranked according to an ELO, but can you give a brief description? Cheers!

Sure. There are two systems that I know of, one used by Tennis Abstract, the other by Ultimate Tennis Statistics. I use the latter, mostly because there's more info and it is better organized. Their numbers are higher than TA, in case you ever check them both.

Basically, Elo rating is a number that represents the level of a player, relative to other players - both current and historic. Every player enters the tour at 1500 Elo and goes up from there based upon who they play and their results. A win gets you Elo points, a loss means you lose them.

For instance, if you look at Medvedev's results for the Miami Masters you can see his Elo rating at the beginning of each match, as well as his opponent's - and what they gained or lost. In the final, he started at 2272, gained +9 and ended with 2281, while Sinner started with 2184, lost -9 and ended with 2175. Over the course of the whole tournament, he gained just +20 points, mostly because his first four rounds were against low ranked players (one a walkover). Sinner actually gained +28 points because he faced harder opponents, despite losing -9 to Daniil in the final.

Or consider Holger Rune's meteoric rise in late 2022. At the Paris Masters, he gained +87 points because he faced strong opponents, most of them with higher Elo ratings. He got +22 for defeating Novak in the final.

How many points you gain or lose depends upon a few factors: the tournament level and the differential between a player and their opponent, and I believe the round. You also lose a small amount of points for not playing, walkovers, etc.

So for instance, let's revisit the dreadful 2013 Wimbledon (remember that?). Steve Darcis began the match vs Rafa at 1882 Elo, Rafa at 2528. Darcis won, gaining +30, Rafa lost -23.

On the other hand, in the 2008 final, Rafa was at 2426 and Roger at 2423 - basically the same, which led to a 16-point gain and loss, respectively. Rafa had only gained +1 Elo points from beating Rainer Schuettler the round before, because Schuettler's rating (1882) was so much lower, and gained just +30 Elo points during that tournament, because most of his opponents were rated much lower.

Or when Sergiy Stakhovsky (1821) beat Roger (2376) at the 2013 Wimbledon, Stakhovsky gained +35 and Roger lost -23 (not sure why Roger didn't lose more...just the formula, I guess).

So basically Elo rating is a moving number that is essentially a "finger to the wind" test as to the level of a player at any give time. As @Moxie rightfully pointed out, it is a bit tricky, because it is cumulative over time. Meaning, a player could be playing terribly for a few months and while their Elo will go down, it won't go down as quickly as their ATP ranking.

For instance, Andy Murray ended 2016 with a 2500 Elo because it was his best year and he was murdering everyone, including the top Elo player, Novak. He was still good in 2017 but struggled, not winning a single tournament, and fell to 2353 by the end of the year - a huge fall of 147 points, but still at elite levels, and an Elo ranking of #3. Andy didn't even play after Wimbledon in 2017, so wasn't exactly at an elite level at the end of the year. At the end of 2018, in which he only played 8 events, he still finished with a 2255 Elo, 5th highest, even though he clearly wasn't the 5th best player that year, or at that point in time.

Another limitation is that it takes into account all performance, so hurts someone like Sampras, whose relatively suckiness on clay dragged his overall Elo down every year. Consider:

Pete Sampras - Peak Elo
Overall: 2407 (12th all-time)
Hards: 2524 (3rd best)
Grass: 2501 (7th best)
Carpet: 2407 (9th best)
Clay: 2226 (73rd best)

Only Novak and Roger reached a higher peak Elo than Pete on hards. On the other hand, his peak clay Elo of 2226 is between Jimmy Arias and Joakim Nystrom - still good, but not great. His 2407 peak Elo is lower than guys like Guillermo Vilas (2431) and Andy Murray (2500) because neither of them were as poor on any given surface, or at least they weren't when they reached their peak.

What Elo is good for, though, is getting a general sense of a player's level, and comparing it over time. And more to the point, it tells us the level they're at, not necessarily whether or not they won the big final. This is why I find it very useful for projecting young players - it tells us their overall or average level of play at a given age, regardless of whether or not they have a tendency to blow it in the final.

A 2300 Elo is a 2300 Elo, whether in 1975 or 2015. So we can say that Andre Agassi at his best (2376 in 1995) was about as good relative to his peers as Stan Smith was at his best (2381 in 1973). We also learn things like the five highest peak Elo ratings ever were: Djokovic 2629 (early 2016), Borg 2622 (mid 1980), McEnroe 2583 (early 1985), Nadal 2552 (late 2013), Federer 2550 (early 2007).

As an aside, Rod Laver peaked at 2509 Elo in 1969, but this is without the date from the pro era. I'm guessing his peak Elo was somewhere in the late 60s - either in 1969 or before, and would have easily surpassed 2600, maybe been highest all-time.

Furthermore, rankings are relatively stable, so we have average Elo ratings by rank over the Open Era of:

Average Elo Rating by Year-end Elo Rank
#1: 2415
#2: 2348
#5: 2223
#10: 2125
#30: 2001
#100: 1831

Meaning, a 2000 Elo means a player is roughly the 30th best in the sport; 2200 means you're roughtly top 5, 2400 means you're as good as the typical best player. Etc.

Anyhow, sorry if that was longer than you wanted. There are some fine points that, I think, are necessarily to both A) understand the usefulness of Elo and B) understand its limitations. I don't see it as an absolute measurement, or even the best one - but it is very handy when comparing across generations and also considering things like player development.

 

[Kieran]

Sure. There are two systems that I know of, one used by Tennis Abstract, the other by Ultimate Tennis Statistics. I use the latter, mostly because there's more info and it is better organized. Their numbers are higher than TA, in case you ever check them both.

Basically, Elo rating is a number that represents the level of a player, relative to other players - both current and historic. Every player enters the tour at 1500 Elo and goes up from there based upon who they play and their results. A win gets you Elo points, a loss means you lose them.

For instance, if you look at Medvedev's results for the Miami Masters you can see his Elo rating at the beginning of each match, as well as his opponent's - and what they gained or lost. In the final, he started at 2272, gained +9 and ended with 2281, while Sinner started with 2184, lost -9 and ended with 2175. Over the course of the whole tournament, he gained just +20 points, mostly because his first four rounds were against low ranked players (one a walkover). Sinner actually gained +28 points because he faced harder opponents, despite losing -9 to Daniil in the final.

Or consider Holger Rune's meteoric rise in late 2022. At the Paris Masters, he gained +87 points because he faced strong opponents, most of them with higher Elo ratings. He got +22 for defeating Novak in the final.

How many points you gain or lose depends upon a few factors: the tournament level and the differential between a player and their opponent, and I believe the round. You also lose a small amount of points for not playing, walkovers, etc.

So for instance, let's revisit the dreadful 2013 Wimbledon (remember that?). Steve Darcis began the match vs Rafa at 1882 Elo, Rafa at 2528. Darcis won, gaining +30, Rafa lost -23.

On the other hand, in the 2008 final, Rafa was at 2426 and Roger at 2423 - basically the same, which led to a 16-point gain and loss, respectively. Rafa had only gained +1 Elo points from beating Rainer Schuettler the round before, because Schuettler's rating (1882) was so much lower, and gained just +30 Elo points during that tournament, because most of his opponents were rated much lower.

Or when Sergiy Stakhovsky (1821) beat Roger (2376) at the 2013 Wimbledon, Stakhovsky gained +35 and Roger lost -23 (not sure why Roger didn't lose more...just the formula, I guess).

So basically Elo rating is a moving number that is essentially a "finger to the wind" test as to the level of a player at any give time. As @Moxie rightfully pointed out, it is a bit tricky, because it is cumulative over time. Meaning, a player could be playing terribly for a few months and while their Elo will go down, it won't go down as quickly as their ATP ranking.

For instance, Andy Murray ended 2016 with a 2500 Elo because it was his best year and he was murdering everyone, including the top Elo player, Novak. He was still good in 2017 but struggled, not winning a single tournament, and fell to 2353 by the end of the year - a huge fall of 147 points, but still at elite levels, and an Elo ranking of #3. Andy didn't even play after Wimbledon in 2017, so wasn't exactly at an elite level at the end of the year. At the end of 2018, in which he only played 8 events, he still finished with a 2255 Elo, 5th highest, even though he clearly wasn't the 5th best player that year, or at that point in time.

Another limitation is that it takes into account all performance, so hurts someone like Sampras, whose relatively suckiness on clay dragged his overall Elo down every year. Consider:

Pete Sampras - Peak Elo
Overall: 2407 (12th all-time)
Hards: 2524 (3rd best)
Grass: 2501 (7th best)
Carpet: 2407 (9th best)
Clay: 2226 (73rd best)

Only Novak and Roger reached a higher peak Elo than Pete on hards. On the other hand, his peak clay Elo of 2226 is between Jimmy Arias and Joakim Nystrom - still good, but not great. His 2407 peak Elo is lower than guys like Guillermo Vilas (2431) and Andy Murray (2500) because neither of them were as poor on any given surface, or at least they weren't when they reached their peak.

What Elo is good for, though, is getting a general sense of a player's level, and comparing it over time. And more to the point, it tells us the level they're at, not necessarily whether or not they won the big final. This is why I find it very useful for projecting young players - it tells us their overall or average level of play at a given age, regardless of whether or not they have a tendency to blow it in the final.

A 2300 Elo is a 2300 Elo, whether in 1975 or 2015. So we can say that Andre Agassi at his best (2376 in 1995) was about as good relative to his peers as Stan Smith was at his best (2381 in 1973). We also learn things like the five highest peak Elo ratings ever were: Djokovic 2629 (early 2016), Borg 2622 (mid 1980), McEnroe 2583 (early 1985), Nadal 2552 (late 2013), Federer 2550 (early 2007).

As an aside, Rod Laver peaked at 2509 Elo in 1969, but this is without the date from the pro era. I'm guessing his peak Elo was somewhere in the late 60s - either in 1969 or before, and would have easily surpassed 2600, maybe been highest all-time.

Furthermore, rankings are relatively stable, so we have average Elo ratings by rank over the Open Era of:

Average Elo Rating by Year-end Elo Rank
#1: 2415
#2: 2348
#5: 2223
#10: 2125
#30: 2001
#100: 1831

Meaning, a 2000 Elo means a player is roughly the 30th best in the sport; 2200 means you're roughtly top 5, 2400 means you're as good as the typical best player. Etc.

Anyhow, sorry if that was longer than you wanted. There are some fine points that, I think, are necessarily to both A) understand the usefulness of Elo and B) understand its limitations. I don't see it as an absolute measurement, or even the best one - but it is very handy when comparing across generations and also considering things like player development.

No that’s brilliant, thanks! I’ll digest the implications and details of all that over the next couple days because it’s very technical but I’ve seen it mentioned a few times here.

What do the letters ‘Elo’ stand for?

Thanks again brother!

 

[Moxie]

No that’s brilliant, thanks! I’ll digest the implications and details of all that over the next couple days because it’s very technical but I’ve seen it mentioned a few times here.

What do the letters ‘Elo’ stand for?

Thanks again brother!

I agree...best explanation yet! And I had already been trying to understand it. I think maybe since @El Dude has had his head in it so long, he didn't realize how thorough and basic an explanation we needed. Well-done, Dude!

 

[El Dude]

No that’s brilliant, thanks! I’ll digest the implications and details of all that over the next couple days because it’s very technical but I’ve seen it mentioned a few times here.

What do the letters ‘Elo’ stand for?

Thanks again brother!

It is the name of the guy who created the system - Argad Elo, a Hungarian American physicist. No kidding!

 

[El Dude]

I agree...best explanation yet! And I had already been trying to understand it. I think maybe since @El Dude has had his head in it so long, he didn't realize how thorough and basic an explanation we needed. Well-done, Dude!

Thanks, Moxie. You know, I've only been learning Elo over the last year or less...I kind of avoided it for awhile, but then when I started researching it more, I was hooked. I think Jeff Sackmann's series got me into looking into it more.

 

[Moxie]

Thanks, Moxie. You know, I've only been learning Elo over the last year or less...I kind of avoided it for awhile, but then when I started researching it more, I was hooked. I think Jeff Sackmann's series got me into looking into it more.

You know I think it has an "averaging" quality that can be misleading. However, I do think your point about using it to predict youngsters seems useful. I'm still not sure about an ELO in 1985 means the same as one in 2023, but we can discuss that.