| |
|
 Subscribe in a reader
FightMetric on Yahoo!
Check out our article in Yahoo! Sports' MMA section breaking down some Anderson Silva stats and discussing what they might mean for Dan Henderson. Highlights include:- Silva's career striking accuracy
- Why Silva thrives in the Octagon while CroCop failed
- Silva's takedown defense stats
- The two ways opponent have figured out to beat Silva.
Much of this draws from the data in our TPR Report on the career of Silva. Stay tuned tomorrow for a TPR Report containing Dan Henderson's career TPR Report.
Volunteers?
We're looking for four volunteers to test a new feature that we have in the works. Time committment is about 15 minutes and it should be a lot of fun. ***UPDATE: Thank you to all who volunteered ***
The Complete Anderson Silva
FightMetric has gone back and scored every* one of Anderson Silva's fights. You can now see a TPR Report for all 23 bouts, going all the way back to the Meca Vale Tudo days in 2000. We will be publishing a detailed analysis of Silva's career profile with a new content partner next week.*There is apparently no existing video of Silva's victory against Waldir Dos Anjos. That one is not included.
Revisiting St. Pierre-Penn
Over the course of several weeks, we have converted all of our fight reports into the dynamic Flash format except for one. Our most controversial fight report to date is still our first, the one between Georges St. Pierre and BJ Penn. A polarizing fight in itself, the score that FightMetric arrives at (St. Pierre wins 29-28, but the overall score is a draw) has left many feeling unsatisfied.
The problem is damage. As it stands, the fight is too close to call only because of the level of damage inflicted credited to Penn. Currently, that level stands at "moderate" on a scale between none, light, moderate, and heavy damage.
Since scoring that fight months ago, we've seen more and more comparative evidence that the level of damage inflicted should, in fact, be marked as "light." St. Pierre had a bloody nose (not a broken nose, as it says in the report) and a swollen eye. We have seen similar levels of damage in fights like Sanchez-Parisyan and Ortiz-Griffin and those cases were classified as "light."
Changing the damage level would, of course, change the conclusion of the fight. Right now the overall score is 234 for Penn against 233 for St. Pierre, meaning a draw. Changing the damage level to "light" means St. Pierre would have the clear advantage 233-202.
Things change, the system improves, and we have no problem admitting when we've made a mistake. In this case, we'll leave this up to you. What do you think? Try to think objectively (i.e., don't consider the fact that this choice changes the outcome) about the correct damage level. Feel free to leave your thoughts in the comments or email us at info@fightmetric.com. We will be revisiting this decision in the next few days and will modify the report as needed.
Fight Matrix
No, that's not a typo. The similar-sounding site, fightmatrix.com, serves a similar purpose as FightMetric: to bring serious analytics to mixed martial arts. Their aim is to provide computer-generated fighter rankings that are more objective than the usual rankings you'd see done by opinion poll. It took college football and basketball decades before they started using computer rankings. Thanks to Fight Matrix, it's taken MMA not nearly as long. Check it out at www.fightmatrix.com.
Explaining TPR: Volume IV - The Time Component
Among the six components that make up the TPR calculation, there are two that have a set number of points that are always shared between the two combatants. These are Time and Dominance. This edition of Explaining TPR will focus on Time.
In addition to being the first component we're discussing with a mandatory shared value, it is also the first with a variable weighting. Under normal circumstances, the Time component is worth a maximum of 10 points, split between the two fighters. This assumes that both fighters have attempted 20 techniques, the minimum number necessary to get a meaningful measure of Accuracy. If a fighter attempts less than 20 techniques (i.e., strikes, takedowns, guard passes, and submissions), Accuracy must be discounted as a portion of TPR and two of the other categories become more valuable. In the case of Time, the maximum number of points increases from 10 to 20.
The decision to make the default value of Time so low (it's tied with Method as the second-smallest contributor) was based on the survey we conducted. Time ranked last by a significant margin among the contributors to fight quality. The message seemed clear that fans don't really care how long it takes for a fighter to finish.
As it turned out, this was only true to a point. People didn't ascribe that much value to finishing a fight in eight minutes compared to four, or even 14 minutes compared to four. Where time becomes a bigger factor in assessing quality is when a fight finishes fast. There is a big difference in perceived quality between a fight that finishes in four minutes and one that ends in 30 seconds. For evidence, consider the explosion onto the scene of Rameau Thierry Sokoudjou. The fact that he knocked out Ricardo Arona and Antonio Rogerio Nogueira was certainly impressive, but what catapulted him in the rankings was how quickly he dispatched both opponents. The same can be said about Houston Alexander, and to some degree, even Anderson Silva. Being a very quick finisher will do wonders for your career.
The calculation works the same way whether the maximum is 10 points or 20 points. To start, the winning fighter gets five points no matter what. The remainder, either five or 15 points, get split between the two fighters. The principal between the split is that the winning fighter should be rewarded for finishing quickly, while the losing fighter should be rewarded for lasting a long time.
In a 15 minute fight, there are check marks every three minutes. So a fight lasting less than three minutes means the losing fighter gets no points, a four minute fight means the loser gets one point, a seven minute fight means two points, etc. The end-result is that a fight that goes all 15 minutes has the winner getting his five automatic points and the loser getting all of the remaining five points. The Time component rewards each fighter equally for doing equal work, i.e., not finishing the fight.
The Time component is adaptable for any fight that has a defined end-point, no matter how long. Fights with unlimited time (like the early UFC's) cannot be measured.
Explaining TPR: Volume III - The Method Component (Cont.)
Part Two
(Read Part One here)
With the hard question of finished fights vs. decisions out of the way, we set the total number of points for the Method component at 10 based on survey results and our own calculations. The next issue was that of apportioning points for the different victory methods. Method is unlike win/loss, in which the binary choice means one fighter gets all the points and the other gets nothing. Different victory methods say different things about each fighter's performance.
We categorized all fights (except no contests) into these categories:
KO/TKO
Submission
Decision
Disqualification
Doctor's Stoppage
The first two indicate a decisive victor. When a fighter gets a knockout or submission, he receives all 10 possible points and the defeated fighter gets nothing. This is his reward for deciding the fight without putting it in the hands of any outside parties.
Decisions of all forms - unanimous, split, and draws - mean both fighters failed equally in their goal to end the fight. Since it's impossible to generalize about the quality of decisions, the fighters split the 10 points, receiving five apiece.
A disqualification means one fighter won, but didn't beat his opponent. In this case, the winning fighter gets five points; he didn't finish, but wasn't finished himself, same as in a decision. The losing fighter gets nothing. In a sense, his illegal actions mean that he finished himself.
Doctor's stoppages, which usually mean cuts, are difficult to value because they come in two forms. Some injury stoppages are the result of a quality performance, like Rich Franklin's stoppage of Evan Tanner. Some represent a single good technique in an overall poor performance, like Kenny Florian's elbow on Chris Leben in the first season of The Ultimate Fighter. The common denominator, though, is that the decision to end the fight was made by an outside party. Even in a case of accumulated damage due to domination, fighters will still argue that they could have continued. Just like in a decision, it is impossible to generalize about the validity of a stoppage. Therefore, the points are awarded the same, with both fighters receiving five.
As a point of clarification, fights that end by TKO because an injured fighter could not defend himself - think Mauricio Rua's loss to Mark Coleman - are considered TKO's. Despite the fact that the winning fighter capitalized on an injury he did not cause, the stoppage was as a result of his succeeding strikes and not the decision of a doctor.
UFC 81 TPR Cont.
The Nogueira-Sylvia fight is a good example of how TPR evaluates performance. In this case, Sylvia walks away with a higher TPR despite losing the fight. The important thing to remember is that TPR is not a comparison between the two fighters in a fight. It is a comparison between a fighter's performance and every other fighter's performance in every fight. So the fact that Sylvia had a higher rating has no bearing on the outcome. But it is a good indication that Sylvia fought a well above-average fight, despite losing. Nogueira's performance, meanwhile, was only slightly above average.Brock Lesnar got the highest rating among losing fighters with a 65, well above the loser's average of 45 and higher than two winners. An impressive debut.Frank Mir continues his feast or famine ways. With the exception of his decision win over Dan Christison, Mir has never had a TPR less than 90 in a win and never more than 20 in a loss.Nice comeback from Marquardt who had the third highest TPR this time after getting a 9 in his loss to Anderson Silva.Griffin-Tibau was a bout that defined average. Griffin gets a 55, exactly the average for winners and Tibau gets a 42, slightly under the loser's average of 45. Tim Boetsch came out of nowhere and exploded to a TPR of 94. Most impressive was his 71% striking accuracy, a rare feat.When a fight lasts 30 seconds and one fighter lands no strikes at all, you get a TPR like Lytle's and Bradley's: 100 and 0.Keita Nakamura walks away with a TPR that is nine points higher than Rob Emerson, who got the split decision nod. The reason? Nakamura should have won that fight, based on FightMetric's values. The first round goes to Nakamura 107-100, and turned entirely on the knockdown toward the end of the round. The second round was also close, with Emerson taking it 76-65. Finally, Nakamura wins Round 3 by a score of 144-99. That gives Nakamura the 29-28 win (as Abe Lefkowitz scored it) and the overall win by a total of 316-275.
UFC 81: Total Performance Roundup
Today is the release of the TPR report for the eight UFC 81 fights broadcast on Saturday's pay-per-view. The TPR's for the fighters are, in descending order:Chris Lytle - 100Tim Boetsch - 94Nate Marquardt - 93Ricardo Almeida - 88Frank Mir - 85Brock Lesnar - 65Keita Nakamura - 63Tim Sylvia - 62Antonio Rodrigo Nogueira - 57Tyson Griffin - 55Rob Emerson - 54Gleison Tibau - 42Rob Yundt - 19David Heath - 17Jeremy Horn - 14Kyle Bradley - 0Stay tuned for a more detailed write-up, including detailed effectiveness scores for Emerson-Nakamura.
Explaining TPR: Volume III - The Method Component
Part One
In the previous installment of this exploration of the TPR formula, we spent a considerable amount of time diminishing the importance of wins and losses. The premise is that a win or loss is a component of performance quality, not an arbiter of quality in itself. We even put forward an example in which a losing fighter received a higher performance rating than the fighter who defeated him.
The result of these conclusions is that the Win component is the smallest contributor to total performance quality. While that is true, we will see that winning or losing plays a part in determining the point value of other components. This installment will focus on the next-smallest contributor, the Method component, which is closely tied to win/loss.
Like the Win component, the weight for Method was heavily influenced by the findings of the survey we conducted. Just as one has to make a hard decision about the relative quality of wins vis-à-vis losses, one has to make a similar value judgment about finished fights versus decisions. This choice is a little harder. The win/loss question forces a valuation of winners and losers. As discussed, not all wins and losses are the same. It's easy to conclude that the general, idealized "loss" can be better than the general, idealized "win" because there's not enough information to make a firm, black-or-white decision. The fighters themselves are often not in control of the outcome. Maybe the judges awarded a bogus decision or maybe the referee stepped in for an unjust stoppage. If the goal of every fighter is to win his fight, the point is that you can't always blame the loser for failing to meet that goal.
Method is the opposite. Every decision has the common inconclusive characteristic. That a fight goes to a decision is entirely in the hands of the fighter. If the goal of every fighter is to finish his fight (as is the first principle of the FightMetric system), we would have to say that a fight that goes to a decision means failure to meet that goal every single time. The principled stance is that a finished fight is always better than a decision.
This is nice in theory, but it doesn't hold up in practice. Some of the most cherished performances in MMA history were in fights that went to a decision. To pick just one example, Randy Couture's performance against Tito Ortiz was five rounds of dominance to the point of embarrassment. It takes a hard line to say that performance was worse than Gabriel Gonzaga's knockout of Kevin Jordan, despite the fact that Gonzaga finished and Couture did not.
The survey results showed this in practice. The majority of respondents - exactly two-thirds - took a nuanced approach and said that decision victories can be better than finished fights. It is interesting to note that the percentage of respondents that chose the hard-line approach - that finishes are always better than decisions - increased compared to the win/loss question. Only 24% said that a win is always better than a loss compared to 33% for this question, a testament to the added level of complexity in making this choice.
|
|
|
|
