All Time Rankings - with numbers and formulas

AllTimeTotal = iOpponentWin + iOpponentAll + iHistQtrSum + iHistQtrBest + iHistQtrDominance + BasicCarScr + ChampPts + FightPctMark

i really like how you gave equal weight to historical, long term and short term factors. since i started coming here in 2002 there's been a debate about how long rankings should take into account. "2 years" seems to be the most common answer, for a ranking of one time, which is not necessarily the same combination of variables for overall rankings. not unlike (let me get geeky here) quantum mechanics and relativity.

i, personally, think historical relevance should have some account in todays rankings, but many others disagree. which leads me to ask....

have you also done other updates for "today's" (or, any one point in time) rankings that you can share? for example, here is A Silva ranked today, and how heavily does historical relevance factor into it, considering he's 0-2 in the last 2 years?
 
Great work OP. Really lays it all on the table quite nicely.
 
You are correct about lower weight category fighters. And that is easy to explain - it's all about opponent level.
Level of competition in Flyweight is still very low and even though Mighty Mouse is dominating that division, his iOpponentAll index is not that high.

This graph shows how division (weight Category) strength was changing historically:

gCtgr_HistLn_Ctg8A.jpg



Demetrious is also relatively young inMMA (less than 8 years fighting) and he will be going up rapidly.
He will probably get additional 60 - 90 points in 2015 (depending on the fight results) and will pass Hayato Sakurai, Rich Franklin, Ken Shamrock and Renato Sobral.

Yep, thanks for the thorough response and what he'll likely look like in the near future. It's just weird to see him below guys who've never been champion at all or really established themself as a dominant #1 in their division. I understand why due to the way it's scored but realistically I don't think anyone will say someone like Yuki Kondo belongs above MM in an all-time ranking. I'm not trying to critisize the system and I'm sure if MM continues to win or at least gets some big wins in the next couple of years he will shoot up the rankings, it's just weird to see that's all lol
 
i really like how you gave equal weight to historical, long term and short term factors. since i started coming here in 2002 there's been a debate about how long rankings should take into account. "2 years" seems to be the most common answer, for a ranking of one time, which is not necessarily the same combination of variables for overall rankings. not unlike (let me get geeky here) quantum mechanics and relativity.

i, personally, think historical relevance should have some account in todays rankings, but many others disagree. which leads me to ask....

have you also done other updates for "today's" (or, any one point in time) rankings that you can share? for example, here is A Silva ranked today, and how heavily does historical relevance factor into it, considering he's 0-2 in the last 2 years?


Yes, it is all about balancing numbers. If we put too much weight on a certain factor, then overall scores might look odd.
That is why I come to this site to see what "experts" think. It is not about what just I or you personally think, but it has to be accepted by a majority of people.
This final version of ATT looks very promising and most of the people like it.


Regarding how historical relevance factor in "today's" rankings.
Current Rankings are determined by the Current Score. One of the ingredients of that Current Score is same BasicCarScr that is used for AllTimeTotal.
CurrentScore = BasicCarScr *60% + MidTermScore + RecentScore + Last4Fights + StreakPoints

BasicCarScr is a rough version of ATT (doesn't take into consideration the time period and most measures are very generic).


This is a good example of balanced numbers and listening to what experts have to say.
Few years ago we used 100% of BasicCarScr to calculate CurrentScore.
And many people complained that fighters with a big Career were ranked too high.
Then we cut it to 75%, then 50% and finally last couple of years we have it at 60%.
Now we rarely have those kind of complaints.
There is always individual preference, but we need to find a good solution in the middle.
 
LMAO at the DISRESPECTFUL number Jones GETS.

Lets be honest with each other for once you placed Jon Jones in number 9 behind big nog, wandy, dan henderson and matt hughes thats BS and you know it.
 
Last edited:
That is an excellent question and a pretty good answer.
I will give you my detail analysis in about one hour (after I finish my 9 AM meeting :rolleyes:)

When you speak of their ranking(as in "this is a summary index that shows how high fighter was ranked in each quarter") are you talking about media ranking or is it a system you devised on your own? Sorry if you already included that info and I overlooked it
 
Yes, it is all about balancing numbers. If we put too much weight on a certain factor, then overall scores might look odd.
That is why I come to this site to see what "experts" think. It is not about what just I or you personally think, but it has to be accepted by a majority of people.
This final version of ATT looks very promising and most of the people like it.


Regarding how historical relevance factor in "today's" rankings.
Current Rankings are determined by the Current Score. One of the ingredients of that Current Score is same BasicCarScr that is used for AllTimeTotal.
CurrentScore = BasicCarScr *60% + MidTermScore + RecentScore + Last4Fights + StreakPoints

BasicCarScr is a rough version of ATT (doesn't take into consideration the time period and most measures are very generic).


This is a good example of balanced numbers and listening to what experts have to say.
Few years ago we used 100% of BasicCarScr to calculate CurrentScore.
And many people complained that fighters with a big Career were ranked too high.
Then we cut it to 75%, then 50% and finally last couple of years we have it at 60%.
Now we rarely have those kind of complaints.
There is always individual preference, but we need to find a good solution in the middle.


which puts A Silva at #9 (i will shamelessly put in a plug for your website since you've been reluctant to do so - HERE).

top to bottom, all time and current, these are the best rankings i've ever seen. and on a side note, as a math guy, i love the thought and work you've put into this. kudos.
 
Last edited:
LMAO at the DISRESPECTFUL number Jones GETS.

Lets be honest with each other for once you placed Jon Jones in number 9 behind big nog, wandy, dan henderson and matt hughes thats BS and you know it.
LOL

Show us your algorithm that proves you're right then, fanboy. JJ should be happy to reach number 9.

TS: I love your ranking. Very well done!
 
which puts A Silva at #9 (i will shamelessly put in a plug for your website since you've been reluctant to do so - HERE).

top to bottom, all time and current, these are the best rankings i've ever seen. and on a side note, as a math guy, i love the thought and work you've put into this. kudos.

Thanks a lot.
That is exactly my goal - to have the best and most objective rankings possible.
My main problem is how to keep my team together. This application is becoming so big that we are spending at least 20 hours each week on maintenance and new development. And is all for free.
So, if you want to participate in any way, just let me know.
 
Fulton over Rickson.

antoniobanderas.gif

and all of them are above Bruce Lee!!! :eek: OMFG HOW IS THAT POSSIBLE!!! not to mention dude from Blood Sport :icon_neut

i imagine your post is mostly tongue in cheek, but the fact is, Rickson had only a handful of official fights. and the algorithm shows it. it's often necessary to understand a tool's limitations before fully grokking what it's all about. every ELO ranking i've ever seen has had severe limitations. that's why this one is so impressive.

frankly, i'm curious if Phillip Miller has a higher ranking than Fulton and Rickson....
 
Thanks a lot.
That is exactly my goal - to have the best and most objective rankings possible.
My main problem is how to keep my team together. This application is becoming so big that we are spending at least 20 hours each week on maintenance and new development. And is all for free.
So, if you want to participate in any way, just let me know.

huh. let me sleep on that. PM me more details. cheers
 
When you speak of their ranking(as in "this is a summary index that shows how high fighter was ranked in each quarter") are you talking about media ranking or is it a system you devised on your own? Sorry if you already included that info and I overlooked it

We have an algorithm that calculates Current Score each week. Right now we are ranking about 10,000 fighters weekly.

For Historical Quarterly Rankings we use same algorithm and this is the process:
1. Go to the previous quarter-end, by setting the RankDate=QuarterEndDate and removing all data > RankDate
2. Run algorithm for Current Score
3. Store historical data and go back to #1

After repeating this 86 times, we have the whole history. There is no point in doing this for periods before 1993 Q3.

It takes us about 3 days to recalculate everything. History recalculation is done only when something in algorithm is changed (Current Score algorithm wasn't changed in the last 6 months), or when we have new/changed fight related historical data.

Point that I want to make is that same algorithm is used for all fighters, for all time periods, for all Weight Categories, and all genders.
That is the essence of objective ranking.
 
Tell me how a fighter with 6 losses is ahead of a fighter with 2 losses, and almost an equal amount of wins?

I don't think any algorithm can explain this.

Other than that I find the rankings good. I like how Nogueira and Hughes are ahead of Jose and JJ, as it should be, but no way is Anderson ahead of GSP, statistically speaking. Yes, some may argue that he was a finisher with a killer instinct, but it's purely subjective, and as far as statistics go GSP is the #1 -- better opposition, less losses, most fights won in the UFC.
 
Look how close top 3 is. Then when you factor in that Fedor was a MW fighting at HW, he automatically becomes #1.
 
Back
Top