| |
|
 Subscribe in a reader
Vindication
One of the interesting things about applying the FightMetric algorithm to extremely close fights is the unknown quantity. With bouts as close as the ones we've analyzed so far, it's been impossible to tell who FightMetric will call the winner based on fight tracking alone. It's only once we've plugged all the numbers into the equation that we get to see how things turn out. Part of the fun is in not knowing what the outcome might be.But sometimes it's fun to try using the algorithm for a fight where you know what the result will be, just to see how right you are. Such is the case with the report released today on the fight between Stephan Bonnar and Keith Jardine. This is one of those fights where everyone seems to agree that the decision was a bad one. What makes matters worse was that Bonnar got a unanimous decision. Most of the recent controversial decisions people like to talk about have been split decisions. In that case, there's a judge supporting you no matter who you thought won those fights. This fight just goes to show that three judges can be unanimously wrong.Two other notable things from this fight: The first relates to damage. Though people clamor for damage to be a deciding factor in judging fights, vicious leg kicks did not seem to count for much in the eyes of the judges. It's important to remember that damage doesn't always have to appear on the face. The other notable factor was that Jardine takes the third round despite suffering a knockdown. Some have suggested that MMA judging can be improved with mandatory point deductions for events like knockdowns or tight submission attempts. The third round of this fight is a perfect argument against that idea. A five-minute MMA round is too dynamic for the winner to be decided based on a single occurrence. The winner of the round should be the fighter who fought better for five minutes, not for five seconds.
Introducing HiPer Strike
Much of the focus on FightMetric has been on the fight score our algorithm provides for bouts that go to a decision. This powerful statistic has produced unique insight into some of the closest fights in MMA history. But FightMetric is more than just one statistic; we are constantly analyzing and refining our research to provide the MMA community with the highest quality MMA analysis in the world.Today, we're happy to reveal HiPer Strike, the newest measure to be used in the FightMetric system. One of the cornerstones of the FightMetric system is to count every significant occurrence in a fight. But a modicum of perspective is lost with that level of detail. By including every strike, it is sometimes difficult to establish meaningful volume and accuracy observations. Even measuring power strikes alone can provide a confused picture of the striking battle. To this end, we've established HiPer Strike (HiPer is short for high-percentage). The strikes counted in this statistic are only the ones that result in the highest percentages of fight endings. In other words, while every strike matters, these are the strikes that one would consider to be the most important when viewing a fight. Comparing two fighters' HiPer Strike totals will give a much clearer picture of striking effectiveness than comparing total strikes, power shots, or jabs.Starting with this Wednesday's report (the match between Stephan Bonnar and Keith Jardine), our By the Numbers page will no longer report on total power strikes versus jabs. Instead, the more meaningful HiPer Strike and LoPer Strike totals will be used.If you're interested in seeing exactly which strike categories are included in HiPer Strike, take a look at our proprietary scoresheet that we've made available to the MMA public for those that would like to engage in research of their own.The real power of the HiPer Strike statistic is still to come. Based on your responses to our Fight Quality Survey, we have nearly completed the next phase in our analysis: A rating system that can compare the quality of any fight with any other. Stay tuned to FightMetric for this revolutionary new method of evaluating fights!
You Can't Judge a Fight By Its Scores
If you saw that the official scoring of a fight was 29-28, 29-28, and 30-26, you'd probably assume that this was not an especially close fight. While two judges gave a round to the losing fighter, the other judge saw all three go to the winner, including one that was 10-8. How close could a fight be with those kinds of scores?
The fight in question was the one between Diego Sanchez and Karo Parisyan. The FightMetric report released today takes an in-depth look at a fight that many considered to be the 2006 Fight of the Year. The FightMetric decision notwithstanding, a fight like this heightens the feeling that simple win-loss or official scoring are not enough to give a clear picture of how close or how good a fight was. Karo Parisyan deserved better (and not just because the loss set-back his title shot dreams even further).
FightMetric is currently engaged in developing a statistic that will help create clarity around how well a fighter performed in any given fight. Thanks to those that participated in our survey, we have a better understanding of what you feel makes a performance a good one. With that knowledge, we're confident we can present some truly meaningful statistics around fight quality in the near future.
Join the Statistical Revolution
The very first step in developing the FightMetric system, before any data was collected or fights scored, was to think. The question was: Out of all the things that can happen in an MMA fight, what are the things that really matter. After careful consideration, FightMetric arrived at 67 things that truly affect the outcome of a fight. Step two was to create a scoresheet so that fights could be tracked along those 67 categories. That sheet went through several iterations before arriving at the final version that has served FightMetric so well. Today, we are making that scoresheet available to you.Our goal all along has been to further the understanding of this sport through clear, objective study. From the emails and comments we've received so far, we can tell that many others share the same goal. To help others who would like to try their hand at fight scoring and to further the understanding of how fight stats work, we are providing the right tool, free of charge, to anyone who wants it.The scoresheet is pretty self-explanatory. Strikes are tracked in three areas (striking distance, clinch, ground) to three targets (head, body, legs) as both power strikes and jabs. Grappling advances from top position are listed with a + (e.g., +Side means an advance to side) and improvements on the bottom are indicated with a - (e.g., -Side means an improvement from side). The other thing you'll notice is that some areas are shaded. These indicate a new statistic we've developed called HiPer Strikes. These are the strikes that result in the highest percentage of fight endings. We will be coming out with a longer explanation of HiPer Strikes in the near future, but the basic result is that strikes in these areas should be weighted more heavily.FightMetric will shortly be going digital, switching to a proprietary input device instead of paper. The scoresheet has been integral to FightMetric's research. We hope it will he helpful to you as well.As always, please send your comments on the scoresheet or any other matters to us at info@fightmetric.com.
FightMetric's Analysis of Evans-Bisping
After the controversy of Hamill-Bisping and the unsatisfying result of Evans-Ortiz, the second-to-last thing either of these fighters needed was another decision. The last thing they needed was a split decision. Unfortunately , that's what they both got.
The first round seemed clearly won by Evans. He took the fight to the ground three times, passed to side twice, and outstruck Bisping. Evans' totals for the round included four head power strikes (three on the ground, one in the clinch) to only one for Bisping, though Bisping landed eight other power strikes in the clinch (four to the body, four to the legs). In all, Evans landed 17 total strikes to 13 for Bisping. Using FightMetric's proprietary effectiveness algorithm, the round goes to Evans by a score of 69-34.
The second round started out well for Evans, landing strikes on the feet and scoring a double-leg pickup that turned into a slam. Bisping was able to completely neutralize Evans' ground attack and managed to get back to his feet. The tide turned in Bisping's favor, as he landed more huge body knees in the clinch and scored with combinations at distance. The overall strike total favored Bisping 27-8, which included six head power shots and eight stiff jabs at distance. FightMetric sees this round 89-55 for Bisping.
The third round was the closest, and was surely the round that caused the split decision. A back-and-forth battle on the feet seemed to favor Evans slightly, as he landed four head power strikes to Bisping's two. Evans also scored two takedowns and a pass to side, though he was again frustrated by Bisping's defense and landed no strikes of consequence (four small jabs to two jabs for Bisping). Overall, it was Evans' strikes that landed to greater effect, opening a small cut at Bisping's right eyebrow. FightMetric gives this round to Evans by a score of 87-63. That makes the fight 29-28 for Evans, in-line with the majority decision of the judges.
Viewing the fight as a whole favors Evans as well. The total StrikeScore was 135-131 for Evans, which FightMetric considers a statistical tie (there's a four-point margin of error), but the total GrappleScore favors Evans 76-55. Total scores for both fighters are 212 for Evans compared to 185 for Bisping, in a match that goes to Evans on the strength of his takedowns and passes to side control.
Combined striking stats (strikes at distance, in the clinch, and on the ground) are depicted in the charts below:
 
Keep Answering and Keep Your Eyes Peeled
Thanks to all the people who filled out the survey in the post below. Because of the response (and the fact that you only get 100 responses per survey unless you pay), I've created a copy of the same survey, which you can still fill out. The link in the post below has been changed to accomodate people still using the permalinkStay tuned next week for for stats and analysis of UFC 78, the release of Sanchez-Parisyan, the introduction of a new striking statistic, and a bit of a surprise.
FightMetric on Sherdog Radio
Tune into the Jordan Breen show today at about 3:15 for an interview discussing the FightMetric system and previewing the Evans-Bisping fight this weekend.
Thinking Outside the Guard
When devising the FightMetric system, we did a comprehensive analysis of both striking and grappling to see what techniques and positions matter most in ending fights. For the grappling component, we operated on the basis of comparative advantage: Has the technique you've executed put you in a better position to end the fight than before. In most cases, this is patently obvious. Everyone knows that advancing from half-guard to mount is an advantageous transition.But what to do with scrambles and unorthodox grappling positions? This problem is evident in the FightMetric report released today on the fight between Tyson Griffin and Clay Guida. The non-stop action in this fight led to several unusual ground positions, some of which did not make immediately clear which fighter had the advantage. What then is an objective statistical system to do? The answer, unfortunately, is nothing. With no historical evidence that these positions confer any comparative advantage, the system is forced to view them as neutral. While FightMetric gives the overall nod to Guida, it is Griffin who takes the ten-point must decision on the strength of two very close rounds. Would the result have been different if there was an established way to quantify the value of non-traditional grappling moves? It's impossible to tell, as some of the positions might just as easily have favored Griffin.Perhaps this same aspect of the incredibly complex world of grappling paralyzed the fight's real judges just as it did FightMetric. Without the ability to evaluate non-traditional positions - to think outside of the common, guard-based positions - fight judges are likely to grade these maneuvers as neutral too. FightMetric is a numbers-based system that relies on cold, hard historical data, and so has difficulty accomodating rare exceptions. What's the judges' excuse?
Tell Us What You Think
Over at FightMetric.com, we're working on the next phase of our analysis of MMA and we need your help. We want to know what you consider to be the most important factors in determining a fighter's performance. To figure this out, try this experiment: Think of the most perfect performance you've ever seen. For example, take Anderson Silva's fight against Chris Leben. What about that fight made it a "flawless victory?" Was it the fact that Silva landed all his strikes? That he took no punishment from Leben? Was it the knockout? The speed? If you'd like to help out, you can fill out this three question survey and provide your thoughts. The survey is limited to 100 participants, but if more people are interested, we can put another one up. Thanks!
10-8 Rounds, a Reality Check
One of the biggest compaints that fans have about the ten-point must system is the lack of clarity surrounding rounds that are scored something other than 10-9. Close rounds are almost never called 10-10 and there have only been two 10-7 rounds in UFC history (Forrest Petz against Sam Morgan and Alessio Sakara against Elvis Sinosic). That leaves the elusive 10-8 round. No one knows what it takes to convince a judge that a round was dominant enought to deserve a 10-8, but we do know how often it happens. Based on a in-depth study of judging using publicly available NSAC records, we have found that less than 4% of MMA rounds are scored 10-8.
Scoring a round 10-8 is not something to take lightly in a three round fight, as demonstrated in the FightMetric report released today on the match between Tito Ortiz and Forrest Griffin. To do so means that the fighter who received eight points must dominate or end the fight if he wants to win. If all he can do is win the other two rounds by typical margins, the best he can hope for is a draw. That is precisely the situation that FightMetric ascribes to Griffin. Having been so totally dominated in the first round, Griffin's remarkeable comeback to win the final two rounds was not enought to award him the victory.
We have gotten several questions regarding FightMetric's requirements for a 10-8 round. Some have suggested that the firstround in the Hamill-Bisping fight should have been 10-8 in favor of Hamill; a decision that would have spared him his first loss.
FightMetric awards a 10-8 round when a fighter achieves a round score that is 100 or greater and is more than six times his opponent's round score. The 6x number is based on data relating to average round scores and is premised on the fact that a 10-8 round is essentially like winning two rounds at the same time.In the case of Hamill-Bisping, Hamill won the first round by a score of 207-55. By contrast, Ortiz won the first round in his fight against Griffin by a score of 177-20. The important point was not the total score (Hamill had a higher number), but rather the differential. The 10-8 round is a measure of domination, not sheer volume of attacks. No matter how well a fighter performs in a round, he would not get a 10-8 if his opponent was similarly effective. Thus it is Ortiz who FightMetric awards a 10-8 round for tallying more than eight times his opponent's score, while Hamill receives a regulation 10-9 for a total that was not even four times Bisping's round score. One of the things that will be interesting to see as the FightMetric project progresses is exactly how often 10-8 rounds should be scored. Maybe 4% is really the correct number and the judges have it right. Only time will tell.
Matyushenko-Schoenauer By the Numbers
The second fight broadcast live from the IFL Grand Prix was the light-heavyweight title match between Vladimir Matyushenko and Alex Schoenauer. The match was not nearly as close as Horodecki-Palaszewski, and all three judges gave the win to Matyushenko. The striking stats for the match looked like this:
Unlike Horodecki-Palaszewski, this fight did have asignificant grappling component. The grappling action breaks down this way:
The FightMetric round scores match up with the scores granted by two of three judges. Matyushenko won rounds one and two by scores of 93-34 and 51-20, respectively. Schoenauer took the close third round 46-39 on the strength of his three submission attempts. The third judge, who scored the fight 29-27, seemed to be too generous with round one.The overall FightMetric score unsuprisingly favors Matyushenko 183-100, making him the clear, true winner of the first IFL individual title.
FightMetric Takes on Horodecki-Palaszewski
Though FightMetric was not on-site in Chicago for the opening round of the IFL Grand Prix, we were watching the broadcast on television and scoring from home.
The opening match between Chris Horodecki and Bart Palaszewski ended the exact same way as their first meeting; in a contentious split decision. The opinion of most was that Hordecki definitely won the first round and that Palaszewski won the second. The third round was anyone's guess.
FightMetric tracked the fight, which was nearly entirely composed of striking. The only significant grappling actions were two takedowns by Palaszewski, one of which led to a sweep by Horodecki. The striking stats for the match break down like this:
The FightMetric round scores tell an interesting tale. The first round goes decisively to Horodecki, who wins 80-26. The second round goes to Palaszewski by an almost equal margin, winning it 104-49. The final round, which was anyone's guess, it still is. That round was dead even, 50-50. Using the ten-point must system, this fight is a draw. Too bad MMA judges seem morally opposed to scoring 10-10 rounds.
The overall FightMetric scores for the fight are 181 for Palaszewski to 178 for Horodecki. Since FightMetric maintains a two-point margin of error for each fighter, any fight within four points is considered a draw. The overall FightMetric scores produce the same result as the ten-point must system. Call this match even.
Come back tomorrow for the results from the light-heavyweight title match between Vladimir Matyushenko and Alex Schoenauer.
Wanted: Bad Decisions
Several people have written in asking if we are soliciting ideas for fights to analyze. The answer is a resounding yes! There is no shortage of close fights to analyze, so if people have a preference for specific matches, it makes our job easier in choosing which to do.
Feel free to toss out ideas. Any organization is fair game, with the only condition that the fight has to be available on DVD. If we can't watch it, we can't score it.
There are several fights that we've already done that are waiting in the queue. Stay tuned for reports on:
Tito Ortiz vs. Forrest Griffin
Diego Sanchez vs. Karo Parisyan
Stephan Bonnar vs. Keith Jardine
Ricco Rodriguez vs. Antonio Rodrigo Nogueira
And the mega-troika: Tyson Griffin against Frankie Edgar, Clay Guida, and Thiago Tavares. (Not at the same time, of course. Not even Rio Heroes allows that.)
Any fight is fair game, but we're focusing the custom reports on the really close and controversial decisions. Leave your requests in the comments or email us at info@fightmetric.com.
|
|
|
|
