All Time Rankings - Updated (Fereder no. 1) [Archive] - MensTennisForums.com

All Time Rankings - Updated (Fereder no. 1)

amirbachar
03-19-2011, 04:51 PM
As I promised, I'll now post the rankings from 1978 to 2010.
The methodology is the based on KRACH, plus the set factoring I presented here:
http://simtheworld.blogspot.com/2011/03/new-rankings-concept-set-importance.html
The rankings are made in a way, that the expectancy of sets won will be the same as the number of sets won in the real world, of course factored by importance.

Update:
I got a suggestion from LocoPorElTenis (thanks!) to factor the results by years, so when players are at their peak, their results will count more.
I've implemented that in the following way:
After each iteration I've ranked the years for each player by their (sets won)/(expected results) in the current rankings.
Then for each player I set a year factor, where the factor for the i-th best year was (average results of the i-th year/expectations for all players)^4.65.
Then each match was factored by the minimum factor for both players in that year.
I admit the rankings makes a lot more sense now.

Full details in:
http://simtheworld.blogspot.com/2011/03/all-time-atp-rankings-update.html

Here are the top 20:

Name Ratings
1 Roger Federer 15740
2 Bjorn Borg 15554.06922
3 Rafael Nadal 14453.64331
4 Ivan Lendl 12515.61249
5 John McEnroe 12346.31073
6 Andre Agassi 11517.9761
7 Pete Sampras 11120.58706
8 Jimmy Connors 10552.42997
9 Boris Becker 9989.140947
10 Novak Djokovic 9839.656444
11 Stefan Edberg 9469.054288
12 Lleyton Hewitt 8529.975095
13 Andy Roddick 8386.351809
14 Andy Murray 8365.692677
15 Mats Wilander 7986.889212
16 Jim Courier 7439.320297
17 Juan Martin del Potro 6853.467182
18 Michael Stich 6845.329504
19 Michael Chang 6788.132475
20 David Nalbandian 6718.077676

*Players still active in bold

castle007
03-19-2011, 04:54 PM
Djokovic, murray and roddick in the top 10??

gulzhan
03-19-2011, 04:54 PM
Tsonga is here and Safin is not?! :lol:

Filo V.
03-19-2011, 04:59 PM
Pandora's box has just been opened.

OLYMPICLEGEND
03-19-2011, 05:01 PM
Anything that says Murray is better than Connors I refuse to take even remotely seriously. In fact, nearly every player on that list is in an unworthy position. LOL Tsonga? :mad:

asmazif
03-19-2011, 05:06 PM
anyone who says tsonga isn't the 15th best player of all time is obviously lying.

Blackbriar
03-19-2011, 05:07 PM
You can make the numbers say anything, even that Tsonga is an all-time best (:haha:) or Connors is greater than Federer (:rolls:) but you can't convince people that this :bs: is true

HDW
03-19-2011, 05:08 PM
dick number 9, :explode::scared::cuckoo::spit::spit::haha::haha:

amirbachar
03-19-2011, 05:08 PM
Anything that says Murray is better than Connors I refuse to take even remotely seriously. In fact, nearly every player on that list is in an unworthy position. LOL Tsonga? :mad:

He is better, not more accomplished... Today's level is clearly better.

JimmyV
03-19-2011, 05:08 PM
I modified the new "set importance" rule you made for these rankings and created a "statistical importance" rule. According to this "statistical importance" rule, all the new stats you made up are total BS and completely irrelevant.

GlennMirnyi
03-19-2011, 05:09 PM
Most pathetic assessment of tennis' top players I've ever seen.

Action Jackson
03-19-2011, 05:11 PM
Nadal vs God gets deleted and then there is this gem.

JimmyV
03-19-2011, 05:12 PM
I created a new stat called "Serbian". After rigorous quantitative analysis, according to my findings only Djokovic, Troicki, and Tipsarevic fulfill the requirements, making them the 3 greatest players ever. This cannot be disputed. Their level is just too great, players from past generations just don't stack up.

Johnny Groove
03-19-2011, 05:14 PM
Djokovic better than Sampras?

Roddick better than Connors?

The methodology is the based on KRACH

Cocaine is a hell of a drug.

http://images.t-nation.com/forum_images/1/a/1acbd_ORIG-normal_CocaineRickJames.jpg

Sunset of Age
03-19-2011, 05:14 PM
Too funny. This thread will become legendary! :yeah:

barbadosan
03-19-2011, 05:29 PM
Djokovic better than Sampras?

Roddick better than Connors?



Cocaine is a hell of a drug.

http://images.t-nation.com/forum_images/1/a/1acbd_ORIG-normal_CocaineRickJames.jpg

That last comment just about sums it up. Nuff said :worship:

barbadosan
03-19-2011, 05:31 PM
On reflection, the best response to a thread like this may simply be "Moving right along"

Sonja1989
03-19-2011, 05:32 PM
Nadal 1? Great! :D :D :D

TMJordan
03-19-2011, 05:33 PM
Tsonga 15th best player of all time :haha: :haha: :haha:

Is he even the 15th best right now>?

Sonja1989
03-19-2011, 05:35 PM
Yeah, everybody won GS event or his best ranking was at least #3. Tsonga is interesting here...

ExcaliburII
03-19-2011, 05:37 PM
Vilas isnt even there :spit:

TMJordan
03-19-2011, 05:38 PM
Tsonga vs Vilas at RG would be nice to watch.

fabolous
03-19-2011, 05:43 PM
well, your seasons ranking seemed ok, but this...not so much.

Blackbriar
03-19-2011, 05:44 PM
Roddick better than Connors?

:spit::haha::rolls:

SetSampras
03-19-2011, 05:45 PM
Djokovic ahead of Agassi, Sampras, and Lendl? What the fuck

Arakasi
03-19-2011, 05:54 PM
Why is everyone hating on the OP? Fair enough, the rankings are stupid but he's just showing us what they would look like using a different methodology. Criticise that, not him. This probably took a while to do.

Just at a quick glance, I'm guessing that a player will fall down these rankings quite a lot as they decline? That would explain Nadal and Borg being so high and it partially explains why Djokovic and Murray are there. It doesn't explain Roddick though :scratch:

silverwhite
03-19-2011, 05:58 PM
Could you do this for the WTA too? I'm guessing Wozniacki would be #1 :haha:

LocoPorElTenis
03-19-2011, 05:58 PM
amirbachar, I know you put a lot of effort into this... but you do realize the results don't make the slightest sense whatsoever, right? Since you have some good ideas, perhaps you could try to understand what went very seriously wrong, so you can ammend it and come back with a more serious ranking :yeah:

habibko
03-19-2011, 06:00 PM
first I felt sorry for the time you wasted on making all of this

then I kept reading on down the list and I don't feel sorry anymore

Blackbriar
03-19-2011, 06:01 PM
Where is Fabio Fognini?
:)

tennishero
03-19-2011, 06:01 PM
Tsonga is here and Safin is not?! :lol:

i guess that shows you how overrated safin was

Sonja1989
03-19-2011, 06:03 PM
Where is Fabio Fognini?
:)

OMG :haha: :haha: :haha:

dombrfc
03-19-2011, 06:03 PM
Djokovic better than Sampras?

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA

samanosuke
03-19-2011, 06:09 PM
Absolutely agree !! Everyone who hasn't lost a single match through carrier deserves to be called as as a Goat

barbadosan
03-19-2011, 06:14 PM
Why is everyone hating on the OP? Fair enough, the rankings are stupid but he's just showing us what they would look like using a different methodology. Criticise that, not him. This probably took a while to do.

Just at a quick glance, I'm guessing that a player will fall down these rankings quite a lot as they decline? That would explain Nadal and Borg being so high and it partially explains why Djokovic and Murray are there. It doesn't explain Roddick though :scratch:

I'm sorry Arakasi, but this time I have to disagree with you. In the end, human beings make the judgements, not things (i.e. methodology). It is beholden to a statistician to assess the credibility of the methodology he attempts to use; not just to throw it out there, devil be damned.

When the methodology produces ludicrous results, then the statistician is obliged to go back to the drawing board, and re-examine the precepts. Results have to be within shouting distance of reasonable; these aren't. Small examples, no way Djoko or Tsonga could be within yodelling distance - far less spitting distance - of Sampras, and if he claims it to be all time, where the heck are people like Laver. They've been reduced to figments of our collective imagination?

tealeaves
03-19-2011, 06:15 PM
Tsonga is better than Del Potro :lol:

CCBH
03-19-2011, 06:17 PM
Why does Nalbandian factor on the list with such a terrible rating? Everyone else is ranked from high to low.

Clearly, another variable in the analysis needs to be the importance of the matches won, not just the quality of the opponent (GS final >>> Delray Beach 1st round).

Apophis
03-19-2011, 06:26 PM
The accuracy of up to 10 significant digits is so astounding that it cannot be wrong.

Blackbriar
03-19-2011, 06:30 PM
Xavier Malisse must be proud to have beaten the 15th best player of all-time.

Action Jackson
03-19-2011, 06:30 PM
This reminds me of Turkmenbashi when he had 99,9% approval rating. These numbers have as much credibility.

BodyServe
03-19-2011, 06:31 PM
How does it work? It takes into account time taken between points? Injury excuses?

Johnny Groove
03-19-2011, 06:32 PM
Marcos Daniel should be #1.

barbadosan
03-19-2011, 06:39 PM
There's hope for Johnny Groove and Xristos yet :)

BigJohn
03-19-2011, 06:40 PM
jXZ_8QnjinE

"Well a very, very heavay a heavy derit, burtation tonight—we had a very dairst Darrison…b*uyet let’s go hit tarish taszin, licsth ta bit to hed the pedt."

dodo
03-19-2011, 06:43 PM
Maybe you could explain a bit more how this thing works. I cant really figure out what youre saying on your page. Because clearly there is way too much disparity between the results and just about everyones (somewhat) informed opinion.

Ilovetheblues_86
03-19-2011, 07:05 PM
Coria and Ferrero better n clay than Kurten :lol:

amirbachar
03-19-2011, 07:06 PM
Why is everyone hating on the OP? Fair enough, the rankings are stupid but he's just showing us what they would look like using a different methodology. Criticise that, not him. This probably took a while to do.

Just at a quick glance, I'm guessing that a player will fall down these rankings quite a lot as they decline? That would explain Nadal and Borg being so high and it partially explains why Djokovic and Murray are there. It doesn't explain Roddick though :scratch:

That may be true in some cases. The rankings are an average of all the matches played. So when a player declines too much (because he probably lost many matches in the beginning of his career it's not certain), he will decline in my rankings too.

amirbachar
03-19-2011, 07:09 PM
Clearly, another variable in the analysis needs to be the importance of the matches won, not just the quality of the opponent (GS final >>> Delray Beach 1st round).

It is - read the full details in the link.

amirbachar
03-19-2011, 07:16 PM
I'm sorry Arakasi, but this time I have to disagree with you. In the end, human beings make the judgements, not things (i.e. methodology). It is beholden to a statistician to assess the credibility of the methodology he attempts to use; not just to throw it out there, devil be damned.

When the methodology produces ludicrous results, then the statistician is obliged to go back to the drawing board, and re-examine the precepts. Results have to be within shouting distance of reasonable; these aren't. Small examples, no way Djoko or Tsonga could be within yodelling distance - far less spitting distance - of Sampras, and if he claims it to be all time, where the heck are people like Laver. They've been reduced to figments of our collective imagination?

Maybe some people just don't realize that the level of play today is a lot higher that it used to be.
Are you saying that you are certain that Sampras would beat Djokovic for example? In my rankings Djokovic is ranked a bit higher, but they are so close, so it's basically a 50-50 match according to the stats.

Matt01
03-19-2011, 08:08 PM
Djokovic better than Sampras?

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA


Rafa better than Fed, Djoker better than Sampras...I really like this list :lick:

LocoPorElTenis
03-19-2011, 08:15 PM
Maybe some people just don't realize that the level of play today is a lot higher that it used to be.
Are you saying that you are certain that Sampras would beat Djokovic for example? In my rankings Djokovic is ranked a bit higher, but they are so close, so it's basically a 50-50 match according to the stats.

How does your method distinguish between eras? Do you automatically give more points to more recent wins?

amirbachar
03-19-2011, 08:20 PM
I forgot to updated the rankings algorithm with non-tiebreaks taken into account more. Read details here:
http://simtheworld.blogspot.com/2011/03/adding-tiebreaks.html

Now the first post is updated.

amirbachar
03-19-2011, 08:35 PM
How does your method distinguish between eras? Do you automatically give more points to more recent wins?

Not at all - that's the good thing about my rankings.
Although players from different eras didn't play against each other, they are connected through players they both played.
So every player has an equal starting rating, and by iterating the results over and over, they ratings converge into the ratings I presented.
In theory all players from one era could be ranked above players from another era...

Commander Data
03-19-2011, 09:08 PM
Djokovic above Sampras -> your work :worship::worship::worship:

Pirata.
03-19-2011, 09:25 PM
:haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha: :haha:

LocoPorElTenis
03-19-2011, 09:33 PM
Not at all - that's the good thing about my rankings.
Although players from different eras didn't play against each other, they are connected through players they both played.
So every player has an equal starting rating, and by iterating the results over and over, they ratings converge into the ratings I presented.
In theory all players from one era could be ranked above players from another era...

I'm sorry but you're losing all credibility. The reason new eras get higher points is simply because players who face greats from two eras (say Agassi and Federer), play the ones from the earlier era when they are already declining, and the ones from the newer era when they are at their peak (the decline period of a player is usually much longer than the development period to reach the top). The fact that your ranking assigns the same points regardless of the form of the players when they met each other (so that a victory over Federer when he was 17 counts the same as one when he was dominating the world) is another problem.

You've come up with a ranking that doesn't make any sense whatsoever, no matter how you slice it. Even looking at current era for example, it's absurd that Nadal is ahead of Federer, I guess because he dominates the matchup especially on GS. But the main issue is you're unwilling to have an objective look at your methodology in the face of some very fair criticism.

born_on_clay
03-19-2011, 09:38 PM
Statistics don't lie
Face it haters :haha:
Great thread :yeah:

v-money
03-19-2011, 09:43 PM
Not at all - that's the good thing about my rankings.
Although players from different eras didn't play against each other, they are connected through players they both played.
So every player has an equal starting rating, and by iterating the results over and over, they ratings converge into the ratings I presented.
In theory all players from one era could be ranked above players from another era...

I think the fact that you have players connected through eras by common opponent they've played, in itself favors the current era. Players get beat up on during the end of their careers or even during rough patches in their careers and this system penalizes them for those losses rather than rewarding them for the achievements in the primes of their career.

amirbachar
03-19-2011, 09:57 PM
I'm sorry but you're losing all credibility. The reason new eras get higher points is simply because players who face greats from two eras (say Agassi and Federer), play the ones from the earlier era when they are already declining, and the ones from the newer era when they are at their peak (the decline period of a player is usually much longer than the development period to reach the top).
The fact that your ranking assigns the same points regardless of the form of the players when they met each other (so that a victory over Federer when he was 17 counts the same as one when he was dominating the world) is another problem.
That's a fair point that I haven't considered. Maybe I can define somehow a peak of let's say 5 years, and consider only matches where both players are at their peak.

You've come up with a ranking that doesn't make any sense whatsoever, no matter how you slice it. Even looking at current era for example, it's absurd that Nadal is ahead of Federer, I guess because he dominates the matchup especially on GS. But the main issue is you're unwilling to have an objective look at your methodology in the face of some very fair criticism.
As I said in the blog, Nadal's distnace from Federer would be higher if Federer lost his clay semi finals and hadn't lost to Nadal in the final.


I am willing to accept criticism. Some posters just doesn't say what's wrong with the system. They just say "no way that player A is better than player B", maybe without even seeing player B play...
Thanks for your reply. I will try to address the problems you presented.

Edit: what do you think about defining the peak as the 5 seasons with the most wins (factored by importance of tournament)?

amirbachar
03-19-2011, 10:40 PM
Perhaps a better suggestion - defining the peak year like before, and decrease each year by 80%, so each result after 2 years equals a max of 64%.
Then the factor for a match would be would be the minimum of the 2 player's percentage.

LocoPorElTenis
03-19-2011, 10:45 PM
Perhaps a better suggestion - defining the peak year like before, and decrease each year by 80%, so each result after 2 years equals a max of 64%.
Then the factor for a match would be would be the minimum of the 2 player's percentage.

why don't you use iterations of your own ranking to determine form? since some players form' lasts much longer than others, a fixed amount like 5 years is not a good idea. You could use the form of a player in each year relative to its career as a weight. In that form, they wouldn't be heavily penalized for their declining years as they are now.

Also, you should definitely weight the matches not only by tournament but also by round. A GS final >>>>>>>>>>>> a GS first round.

n8
03-19-2011, 10:53 PM
Why is everyone hating on the OP? Fair enough, the rankings are stupid but he's just showing us what they would look like using a different methodology. Criticise that, not him. This probably took a while to do.

Just at a quick glance, I'm guessing that a player will fall down these rankings quite a lot as they decline? That would explain Nadal and Borg being so high and it partially explains why Djokovic and Murray are there. It doesn't explain Roddick though :scratch:

This. The intention was good and, when the results weren't consistent with reality, at least he had the guts to show them.

I wish I could get this many responses to my article threads. I think a lot is in the title :).

Dyraise
03-19-2011, 11:13 PM
Statistics don't lie
Face it haters :haha:
Great thread :yeah:
:lol::lol: nice

amirbachar
03-19-2011, 11:35 PM
why don't you use iterations of your own ranking to determine form? since some players form' lasts much longer than others, a fixed amount like 5 years is not a good idea. You could use the form of a player in each year relative to its career as a weight. In that form, they wouldn't be heavily penalized for their declining years as they are now.
Wow, that's a great idea. That will definitely improve the rankings.

Also, you should definitely weight the matches not only by tournament but also by round. A GS final >>>>>>>>>>>> a GS first round.
I thought about it, but it means that it may be possible that losing the semis would be better than losing the final.


Again, thanks a lot for your tips. It's nice to see that there are posters who contribute.

duong
03-20-2011, 12:23 AM
This. The intention was good and, when the results weren't consistent with reality, at least he had the guts to show them.

I wish I could get this many responses to my article threads. I think a lot is in the title :).

I will say it more clearly : the research itself is interesting,

but as many researchers, there's this effort to produce a striking title and definitive conclusions to catch the attention which yes, catches both attention and critics :shrug:

That's not my habit as I'm not a researcher, my work as a statistician is one of a civil servant who works for the government, we don't have to "sell ourselves", we just produce stats to help the government and people.

Oh well ... even some people in my job sometimes make a "striking title", I mean derive conclusions which are not directly related to the results, but personally I don't like doing that.

Now on the main topic :

1. if I understand well, you start from 1978 : in that case it seems difficult to me to rate Connors and Borg ; besides the title "all times" well ... as Laver still is the main contestant for after the war, I mean.

2. Locoporeltennis talked a little bit about it, and you also tried to address it from what I see with your "peak years", but I think generally that where many methods give too much importance to longevity (like for instance the stats for the total number of slams : I can't avoid thinking that Borg could have won more of them if he had kept on ... then Borg has a disadvantage in numbers), I think that your method doesn't give enough importance to longevity. I mean if a player keeps on long after declining, of course his stats will decrease :shrug: For instance, I don't know if it's possible but if Fed had stopped his carreer in the end of 2007 or 2008, what would his result be ? For the same reason, Borg has an advantage in % : because he stopped early his stats not in numbers but in % (like % wins in slams) are unbeatable :shrug:

Then well maybe you can try to put many tricks in your method to improve it but well, when a method is too complicated, it's usually tougher to sell ;)

Anyway, I congratulate you for your efforts, I could never be brave enough to do that but that's interesting indeed ;)

amirbachar
03-25-2011, 08:35 PM
updated...

Lleyton_
03-25-2011, 08:58 PM
Please stop and save your time for something more useful. This isn't working.

n8
03-25-2011, 11:41 PM
The new rankings look a lot nicer. Good work!

duong
03-26-2011, 12:49 AM
I would like to tell you something, Amir, but frankly speaking I can't understand the method :confused:

maybe it would need a little bit more explanation to present to the people :shrug:

amirbachar
03-26-2011, 01:30 AM
I would like to tell you something, Amir, but frankly speaking I can't understand the method :confused:

maybe it would need a little bit more explanation to present to the people :shrug:

Well, you can't have a system which is both simple to understand and produce meaningful rankings when comparing players from different times...
That was really fun making this, and it didn't take me as much time as you would think (about 10 hours in total).
I hope the rankings make sense to people now, and if not, maybe it's because they overate their childhood stars more than their real ability...

duong
03-26-2011, 11:10 AM
Well, you can't have a system which is both simple to understand and produce meaningful rankings when comparing players from different times...
That was really fun making this, and it didn't take me as much time as you would think (about 10 hours in total).


yes I totally understand, Amir, it's an observation I've often made in my work (although we can seldom put it into practice because it's usually forbidden to display too complicated methods)

However, at least for people like me who are able to understand those things, it would be nice of you to maybe explain a little bit more explicitly your method because it's really hard for me understanding (well I'm tired at the moment but yet ;) )

amirbachar
03-26-2011, 06:12 PM
yes I totally understand, Amir, it's an observation I've often made in my work (although we can seldom put it into practice because it's usually forbidden to display too complicated methods)

However, at least for people like me who are able to understand those things, it would be nice of you to maybe explain a little bit more explicitly your method because it's really hard for me understanding (well I'm tired at the moment but yet ;) )

I can give it a shot, but if you have a concrete question, you can ask me here:

I'll use the notation A to speak about a player with current ratings A.
I start by assigning arbitrary rankings for all players (let say everyone start with 1000).
Then, for each set between A and B, the expectancy for A is A/(A+B). For each player we can calculate the expectancy sets won according to the sets he actually played.
If the expectancy is lower than the number of sets he actually won, we increase his ratings.
If the expectancy is higher than the number of sets he actually won, we decrease his ratings.
That system is called KRACH.

Now, to adjust it to all time tennis rankings:
I assigned set factor, according to the importance of the set n the match (so final sets worth more).
I assigned tournament factor, so GS for example worth more.
And finally, the latest addition - for each player I ranked all the years he played in according to the sets won/expectancy (like before).
For the best year for each player, there is a factor of about 1.9, for the 2nd best about 1.38 and so on...
So I factored each match by the minimum of both players factors in that year.

duong
03-26-2011, 07:57 PM
I can give it a shot, but if you have a concrete question, you can ask me here:

I'll use the notation A to speak about a player with current ratings A.
I start by assigning arbitrary rankings for all players (let say everyone start with 1000).
Then, for each set between A and B, the expectancy for A is A/(A+B). For each player we can calculate the expectancy sets won according to the sets he actually played.
If the expectancy is lower than the number of sets he actually won, we increase his ratings.
If the expectancy is higher than the number of sets he actually won, we decrease his ratings.
That system is called KRACH.

Now, to adjust it to all time tennis rankings:
I assigned set factor, according to the importance of the set n the match (so final sets worth more).
I assigned tournament factor, so GS for example worth more.
And finally, the latest addition - for each player I ranked all the years he played in according to the sets won/expectancy (like before).
For the best year for each player, there is a factor of about 1.9, for the 2nd best about 1.38 and so on...
So I factored each match by the minimum of both players factors in that year.

does the expectancy only depend on the past or also on the future ?

amirbachar
03-26-2011, 08:41 PM
does the expectancy only depend on the past or also on the future ?

It depends also on the future, since it is calculated as A/(A+B) where A is the ratings in the current iteration.

amirbachar
03-27-2011, 07:38 PM
Just to show there is no inner advantage in the system to the current players, here are the top 50 up to 2002:

http://simtheworld.blogspot.com/2011/03/all-time-atp-rankings-up-to-2002.html

amirbachar
03-27-2011, 10:15 PM
Rankings by surface:

http://simtheworld.blogspot.com/2011/03/all-time-atp-rankings-by-surface.html