What Numeric Facts should be used in the All Time Ranking discussions

Discussion in 'UFC Discussion' started by zec1234, Jan 6, 2015.

  1. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    For the All Time discussion we first have to identify what facts should we used and how.
    Those facts should describe fighter's accomplishments in some numerical way.
    Most often discussed "facts" are:
    - where they were ranked and for how long?
    - who faced better opponents?

    Simplest way to get Numeric Fact is to do: Count, Average or Sum.

    In this thread I am trying to have a conversation on how to use MMA numbers, not to argue who should be on the top.
    We can use fighter names as example for some facts and measures, not to root for them.

    So, numbers for All Time Ranking.
    What should we measure? What should we count and how to use Avg/Sum?
     
  2. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    Most people use ranking position as a description of how good fighter was at a certain point in time.
    That same number (ranking position) can also be used to describe opponents.
    Since we want to compare fighters across weight categories, we should use P4P in addition to divisional ranking.

    But there is a difference between dominant #1 (like GSP in 2010) and non-dominant (like Hendricks/Lawler).
    Also there is a difference between barely top 5 fighter and undoubtedly top 5.
    Furthermore there is a difference between being top 10 Flyweight and top 10 Lightweight.
    For those reasons, it is also good to use some other numbers (scores, ratings) that describe how good fighters are.

    Other things that people use here:
    - how many championship fights they have? Problem here is that not all belts are the same.
    - how dominant they were in victories? It is hard to express this in a numeric way.
    - did they avenge all their losses? is this really that important and does beating somebody at the end of their career counts?

    What else would you use and how would you numericize that?
     
  3. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    To support this discussion with examples, I will use the approach that we at ScoreCardMMA use to calculate All Time Score.

    In most of our analysis we use following rankings and scores:
    • RC = RnkCtg - Ranking in weight Category (Division)
    • RP = RnkP4P - P4P Ranking (some call it division dominance)
    • CS = CurScr - Current Score, used to create current rankings (represents fighter
     
  4. tenniswhiz

    tenniswhiz Steel Belt

    Joined:
    Oct 29, 2009
    Messages:
    32,000
    Likes Received:
    11,228
    Location:
    Santa Monica
    Different things matter to different people.
    The sport existed in a different form than what is going on today, for several years.
    Rankings are flawed, claiming 'x' factor is more important than 'y' factor is a matter of opinion, and some guys had to suddenly change everything to fight under different circumstances (and some guys just grew up under those circumstances).

    Arguing GOAT is a pointless exercise.
    There have been several great ones who are all qualified.
     
  5. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    Here is an example of how some of those rankings and scores look historically for Fedor:

    [​IMG]


    I will provide detail numbers for different counts.
    Let me know what do you want to see and what fighters you would like to compare?
     
  6. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    Here is another example where graph can help in comparing careers of two fighters:

    [​IMG]

    This compares Category Ranking and Current Score for JBJ and Aldo.
     
  7. smead

    smead Yellow Belt

    Joined:
    Apr 1, 2007
    Messages:
    182
    Likes Received:
    20
    Awesome site! Never knew it existed. JBJ has passed or will pass Fedor for #1 in my opinion. Things that cannot be quantified which make Fedor #1 to me are: unlike almost ALL of the other fighters in the top 10, Fedor did not cut weight. He was an undersized HW without question (but so was much of his competition); his fighting style cannot be quantified - he always fought to finish (not to win). I don't know if you use % of finishes as a variable?
     
  8. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    Here are some measures that we use.
    I used numbers for
    A. Silva .|. Fedor .|. Jon Jones |

    15 .|. 13 .|. 7 .|. (cOpRC5) = Number of fights with Opponent in top 5 Ctg
    12 .|. 9 .|. 7 .|. (cwOpRC5) = Number of Wins with Opponent in top 5 Ctg
    4 .|. 7 .|. 1 .|. (cOpRP10) = Number of fights with Opponent in top 10 P4P
    17 .|. 14 .|. 7 .|. (cOpRP30) = Number of fights with Opponent in top 30 P4P
    15 .|. 14 .|. 9 .|. (cwOpRP50) = Number of Wins with Opponent in top 50 P4P
    5 .|. 1 .|. 8 .|. (mwOpRP) = Win - opponent best P4P ranking (before fight)
    501 .|. 472 .|. 510 .|. (awOpCS) = Win - opponent average Current Score
    760 .|. 807 .|. 828 .|. (mwOpCS) = Win - opponent highest Current Score
    524 .|. 506 .|. 512 .|. (aOpCS) = Opponent average Current Score
    72 .|. 60 .|. 28 .|. (cntHF) = Number of ranked quarters
    27 .|. 26 .|. 17 .|. (cntR1C) = Ranked #1 in Category (at the end of quarter)
    14 .|. 11 .|. 9 .|. (cntR1P) = Ranked #1 P4P (at the end of quarter)
    31 .|. 30 .|. 17 .|. (top3ctg) = Ranked in top 3 in Division
    27 .|. 28 .|. 17 .|. (top5p4p) = Ranked in top 5 P4P (at the end of quarter)
    34 .|. 49 .|. 25 .|. (maxStreak) = Best Win Streak (points not #of fights)
    42 .|. 45 .|. 21 .|. (gt6CS) = Current Score over 600 (count quarters)
    971 .|. 938 .|. 969 .|. (maxCS) = Best Current Score
     
  9. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    There are a lot of things that cannot be quantified and unfortunately I cannot use them in any calculation.
    We can say that somebody has more heart, or better technique, or is most athletic.
    That might be just our opinion. And how important that really is?
    I love Fedor, but looking at the numbers, there is no justification to make him #1. And it is not his fault, but his management.

    Finish percentage is something that we can calculate and I use that in our formula.

    Here is Finish % for Big 5:

    79% (33) - Anderson Silva
    76% (34) - Fedor Emelianenko
    71% (21) - Jon Jones
    64% (25) - Jose Aldo
    52% (25) - Georges St. Pierre

    In brackets you have number of wins.

    Finish % = (NumberOfWins - NumberOfDecisions) / NumberOfWins

    Simple formula.
     
  10. FallonFoxGOAT

    FallonFoxGOAT Banned Banned

    Joined:
    Oct 21, 2013
    Messages:
    7,688
    Likes Received:
    74
    Location:
    The secret Eye Poke Army headquarters.
    I think an average number of fights per year and what the average ranking of opponents faced per year/average ranking of opponents overall (as in their entire career) would be some things to look at.
     
  11. Jeff Goldblum69

    Jeff Goldblum69 Brown Belt

    Joined:
    Jan 6, 2012
    Messages:
    3,939
    Likes Received:
    1,090
  12. estranged

    estranged Brown Belt

    Joined:
    Feb 1, 2014
    Messages:
    4,502
    Likes Received:
    5,398
    cool stats here
     
  13. ThePewp

    ThePewp Green Belt

    Joined:
    Feb 16, 2013
    Messages:
    1,476
    Likes Received:
    6
    I also think that context should matter a little bit regarding the nature of losses.

    Jones lost to Hamil. No questions. But, he did dominate that fight and lost a sketchy DQ, similar to how Anderson lost by DQ for delivering a groin kick. Imagine if DC claimed to be crippled by the eyepoke, refused to continue fighting and demanded a DQ Title Win. I think that the stats of how a fighter performed should matter; Condits victory over Diaz, as we all know, was not quite the way of the warrior.

    If there were a good means to include at least time dominated, strikes/td's landed per fight lost, that could help shed context on things, especially for people who are kinda new and maybe didn't see some of the fights, just see the records.
     
  14. MMun

    MMun Black Belt

    Joined:
    Nov 3, 2011
    Messages:
    6,777
    Likes Received:
    387
    Location:
    Copenhagen, Denmark
    Do you have any insight into how the Fightmatrix rankings differ from your method? Comparing your all-time list with theirs, there are a lot of similarities and some key differences. The top 10 are very similar, with 9/10 fighters ranked at most one place away from their position on the other list. But you have Chuck at #8 and Randy at #12, while FM has Randy at #7 and Chuck at #16. Outside the top 10, some fighters are similarly ranked while others change wildly. For example, FM has Frankie at #12 while yours has him at #23.
     
  15. Magnum1978

    Magnum1978 Silver Belt

    Joined:
    Mar 8, 2007
    Messages:
    11,989
    Likes Received:
    3,124
    Location:
    Melbourne, Australia
    People look at different things, there are a lot of intangibles, in the end its always personal preference/opinion


    - domination (not losing rounds)
    - finishes (its complete, undeniable victory)
    - fighting the opponent strengths
    - aggression
    - come backs (mental toughness, etc)
    - weight given up to opponents (or going up divisions, etc)

    etc
     
  16. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    Using "average number of fights per year" would benefit fighters like Travis Fulton, Shannon Ritch, Dan Severn and Jeremy Horn. We don't use that measure directly, but indirectly is used in some other measures.

    Average ranking of opponents is a good suggestion. The only issue is that ranking position is not always good for averages.
    You can have opponents that were ranked #4 , #7 and #200 (like Hong Man Choi for Fedor) and average would be #70.
    For that type of calculation we prefer to use Scores instead of Rankings:
    A. Silva .|. Fedor .|. Jon Jones |
    524 .|. 506 .|. 512 .|. (aOpCS) = Opponent average Current Score
     
  17. FallonFoxGOAT

    FallonFoxGOAT Banned Banned

    Joined:
    Oct 21, 2013
    Messages:
    7,688
    Likes Received:
    74
    Location:
    The secret Eye Poke Army headquarters.
    IMO beating someone ranked so low should reflect negatively on your score (especially for All Time Status), anyways.

    I'll sleep on it and see if I can come up with any other ideas.
     
  18. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    Problem with Fightmatrix rankings is that they don't publish their formulas or detailed numbers.
    They have some generic statements on how they do things, but never show exact formula or how do they come up with those numbers.
    For AT Rankings they don't even have points published. Just order of fighters.

    Our methodology, formulas and detailed numbers are all presented on our website.
     
  19. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    Here is more statistic in the graphical presentation.

    This is Historical P4P ranking for Big 5.

    [​IMG]

    Graph is inverted (1 at the top) and on the logarithmic scale.
     
  20. zec1234

    zec1234 Brown Belt

    Joined:
    Feb 22, 2008
    Messages:
    2,573
    Likes Received:
    268
    Location:
    Texas
    That is what I'm trying to avoid - intangibles and personal preferences.
    I am trying to use things that are defined and can be described using numbers.
    Just a definition of intangibles - "incapable of being realized or defined" , is enough to steer clear of it.

    Let's go through your list:

    - domination (not losing rounds)
    This is something that can be measured and we are using domination as a parameter for individual fights.

    - finishes (its complete, undeniable victory)
    Also not intangibles. We are using that, but is not a major factor.

    - fighting the opponent strengths
    This is something vague, but we have a fight mark that can cover some of those cases. It might be also sign of dominance.

    - aggression
    As oppose to be a smart and technical fighter?

    - come backs (mental toughness, etc)
    That can be defined on individual fight basis. Sometimes you dominate and get submitted, and other time you can be dominate and then KO that guy.

    - weight given up to opponents (or going up divisions, etc)
    Also something that can be done, but in the individual fight basis. Fedor was typically smaller heavyweight, but was also fighting LHW and MW.
    We have a size advantage flag for the fight, but that is used only if that is drastic.


    So, most of those are not actual intangibles, but a pretty good suggestions.
    Some of those we already use.
     

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.