As a competitor, isn't it nice to know when the other guy has tendencies? What does a dip of their shoulder mean when sparring? How do they set up particular combinations? What do a pitcher's facial and body movements tell us about what he's about to throw? How does a basketball player set up his jump shots or drives to the basket? This is all extremely important information in the heat of battle. While few MMA fighters will readily admit to gearing their fight strategy towards the judges, wouldn't it help to know if the pre-assigned judges tend to score certain things higher or lower than normal?
Judges are human beings tasked with grading a complex sport in real-time with potential visibility problems while using guidelines that leave things open to interpretation ("effective" striking and grappling). It would be absolutely shocking if they all behaved exactly the same and didn't have individual tendencies. All one needs to do is watch Jamie Varner vs. Melvin Guillard from UFC 155 to quickly and permanently dispel that notion.
By evaluating over 6,000 scored rounds, we already know a good amount about how MMA judges typically value landed and missed strikes, takedowns and ground control. We've got a robot judge who, without having watched a single fight last year with human eyeballs, matched the real judges in 214 of 248 UFC decisions (and also disagreed with Diego Sanchez over Ross Pearson). Instead of the typical judge, wouldn't it be nice to know the individual tendencies of specific judges?
UFC 189, one of the biggest fight cards of the year, is right around the corner and we have the judging assignments for the main and co-main events. Junichiro Kamijo, Chris Lee and Marcos Rosales will man the scorecards in the main event for Conor McGregor vs. Chad Mendes, and Adalaide Byrd, Derek Cleary and Sal D'Amato will score the Robbie Lawler vs. Rory MacDonald co-main event.
Kamijo, Rosales, Byrd and D'Amato each have a good amount of data, so we can try to document some of their observed tendencies. We don't have thousands of observations for each judge, meaning some of the performance metrics like ground control have to be aggregated more than I'd normally like (i.e., if you make your performance statistics too detailed without enough observations, results get screwy). So think of this as more of an interesting experiment as opposed to hard core MMA analytics.
Conor McGregor vs. Chad Mendes
Raise your hand if you're not pumped to watch McGregor vs. anybody? No hands up? Good. McGregor has utterly dominated in his five UFC fights, but three of those dominations came against people named Brimage, Brandao and Siver. Not that each of them wouldn't kick the crap out of yours truly, but they don't exactly round out 60 percent of the best resume heading into a title fight. Either way, Mendes is a game opponent and more than a suitable replacement to get the fight blood flowing tomorrow night.
The two judges we can evaluate are Junichiro Kamijo and Marcos Rosales. Kamijo is mostly a Nevada judge with 116 decisions to his name according to MMA Decisions. Rosales has 172 decisions while spending much of his time in California and Nevada, but also doing a decent amount of judging in other areas. I had the pleasure of shadow judging next to Rosales at Bellator 136 earlier this year as he called out Chad George's Von Flue choke in spite of our poor viewing angle and few other people in the building noticing it (possibly including the referee).
Each judge will have a chart and each chart shows how that judge tends to value striking and grappling relative to the typical judge. So if Head Power Landed is 40 percent, it means the judge in question tends to put more weight than usual on power shots to the head; 40 percent more weight than usual, to be exact. If Leg Power Landed is -60 percent, it means the judge is 60 percent less influenced than usual. For whatever reason, strikes to the leg just don't do it for them. Think of zero percent as meaning the judge in question tends to behave like most others.
Generally speaking, the farther the numbers are from zero the more statistically strong the results. A difference of 10 percent or -10 percent could easily be due to noise. We're mostly interested in the more substantial changes or noticeable patterns.
For a typical judge, a power shot to the head is the highest-scoring strike. It takes five jabs to the head or two power shots the body or legs to equal one powerful shot to the skull. Kamijo is about average when scoring power shots, although he may not weigh leg kicks as much as most. He seems to be more heavily influenced by jabs that touch, but don't blast, the face. Power shots still matter and are always preferred to non-power jabs as the end result of a combination. But, relative to other judges, Kamijo really likes to see you active and touching the other guy's face.
When examining grappling for Kamijo, submissions and the clinch clearly stand out. At -56 percent, Kamijo doesn't give nearly as many points for simply attempting a submission. But if you get it locked in and tight (and aren't able to finish), you're almost guaranteed to win the round. Tight subs are usually worth roughly twice as much as just attempting any legitimate submission. For Kamijo, they're worth eight times as much! Put another way, when he sees a tight submission, he's very likely to think effective grappling trumped effective striking that round.
A typical judge usually gives one minute of ground control a little more than twice the points that a minute of clinch control earns (recognizing that ground control can be a mixture of different positions). Kamijo values any type of control (clinch or ground) more than the usual judge, but is super affected when you press your opponent against the cage. Let's see, so head jabs, clinch and ground control are good to Kamijo? Paging Darren Elkins. Darren Elkins please come to Nevada. You want Junichiro Kamijo as your judge.
One final thing to note about Kamijo is that he doesn't seem to be influenced at all by missed strikes - and he's the only one of the four judges today who can make that claim. Misses to the head, body and legs, whether jabs or power shots, don't seem to faze him. This is impressive considering some of the viewing problems I experienced shadow judging when the action is in the corners or right in front of you on the fence. And not all misses are whiffs. Consistently being able to distinguish not only whiffs, but blocks and things like off-target shoulder blows in the heat of the action is pretty impressive.
Did we just explain Kamijo's 30-27 scorecard for Nam Phan in Garcia vs. Phan 1? Perhap we did. If anything, it looks like Kamijo penalizes fighters a little for winging to the head and missing, and that surely wouldn't have done any favors for his perception of Garcia.
Marcos Rosales was an incredibly nice guy to work with - as were Big John, Mike Smith and Mike Bell. But that doesn't keep him from having tendencies to analyze.
The striking story for Rosales is that he doesn't seem to have nearly the appreciation for power leg shots and jabs to the head as the typical judge. He appreciates a good body shot but, as we'll see in a moment, might have difficulty being able to tell when they connect.
With grappling, Rosales is mostly a typical judge. As my Von Flue story suggests, he's very aware of submission attempts. Perhaps this explains why he values them over twice as much as the usual judge.
It's not shown on the chart, but Rosales is affected like a normal judge by fighters swinging and missing to the head or having their strikes blocked. But he values missed power shots to the body about 150 percent more than normal, which may say something about his ability decipher action against the cage. If you have Rosales as your judge and you're against the cage, go to town with power to the body and, make or miss, you tend to score about as many points as if you connect to the head.
Main Event Implications: What does all of this mean for McGregor and Mendes? Our two judges are consistent in both placing less weight on attacking the legs with power and more weight on controlling the opponent in the clinch.
Neither McGregor nor Mendes attacks the legs much with power anyway, but McGregor has been more susceptible to leg strikes (and Mendes' numbers include almost 30 minutes in the cage with Jose Aldo). Judges are likely to give leg attacks less weight in this fight...but I'm pretty sure both guys really just want to punch or kick each other in the face anyway.
In the clinch, both guys tend to have the same amount of control, but it's probably because Chad Mendes goes for a bazillion takedowns a minute and lands at a solid rate. If McGregor's takedown defense is truly as solid as his current 5-for-5 success rate, then the tendency of our judges to favor clinch control would seem to edge for Mendes.
Overall, I've got McGregor at 67.9 percent to capture the interim featherweight title on Saturday night. It'll take some serious Mendes clinch control to overcome those odds...or a counter knockout...or a guillotine.
Lawler vs. MacDonald: The amazing scrap that nobody's talking about. At least more people are talking about it than the poor TUF Finale fighters on Sunday. We've got your back and haven't forgotten about you guys.
Lawler and MacDonald met a year and a half ago with Rory coming out on the wrong end of the decision. Count me as one of the people who believes it was MacDonald's fight to lose and we're likely to see a different outcome when they run it back. Both guys are hard to finish and they've had one close decision already, so judge tendencies could very well play a role on Saturday.
The two judges we can evaluate are Adalaide Byrd and Sal D'Amato. Byrd is exclusively a Nevada judge with 131 decisions and the unfortunate distinction of being on the ass end of Garcia vs. Phan 1 and Varner vs. Guillard. D'Amato is a veteran of 277 decisions and has judged all over the world.
Byrd is pretty typical when it comes to scoring strikes landed, if anything edging towards valuing each strike a bit more than normal. Subs are interesting as she pretty much gives no weight to submission attempts, but if you lock it in nice and tight, you're golden in her eyes. She also weighs cage pressing in the clinch twice as much as usual and ground control about 50 percent more.
If you're noticing that most of her stats are blue and positive and are getting confused as to how someone can weigh almost every element of striking and grappling more than the typical judge, don't worry, you're probably not alone. It means that small differences in strikes, takedowns or control times are more likely to be seen by Byrd as bigger advantages or disadvantages whereas the other three judges examined today would tend to see the same thing as being closer. * Fans of Nam Phan nod their head while saying, "But it wasn't even close!" *
Byrd mostly does a good job of not being influence by misses, especially power misses to the body. Her one area of weakness is possibly the most important, missed power strikes to the head, where she behaves like a typical judge and gives fighters 1/10th the credit of a landed strike.
Sal D looks mostly average but the scale is somewhat skewed by the 388 percent value for tight submissions. The results for striking suggest that D'Amato tends to be relatively more affected by touching someone than by connecting with power. Volume strikers should have a slight advantage with D'Amato as opposed to low output power punchers. At distance this edges for MacDonald who tends to land five more head jabs and 1 ½ more power strikes per five minute round than Lawler, and also has excellent defense.
In grappling, D'Amato's less influenced by takedowns and more affected by all aspects of submissions. It could be that he's more aware of submissions or simply that he thinks subs are a more significant component of effective grappling, especially when locked in tight.
Co-Main Event Implications: Byrd and D'Amato both value an active head jab more than usual which tends to favor MacDonald. Rory's takedowns probably won't get quite the credit they'd normally receive, but the fact that both judges put significant weight on tight submissions could end up a big advantage for MacDonald, especially since the last time Lawler attempted a documented submission was...well...never.
It's not clear who Byrd's penchant for clinch control favors as neither guy's exceptional at cage pressing and if we look back at their first fight, they spent a grand total of around 20 seconds in the clinch with only two seconds of control.
I didn't show it, but Byrd and D'Amato both punish fighters 50-75 percent more than usual for missing a takedown. This is probably meaningless to Lawler's offense but could matter on the defensive side as MacDonald went for six takedowns in 15 minutes during their last meeting. Throw in that Lawler has since trained twice with a focus on stuffing Johny Hendricks' takedowns and this judging tendency could mean even more.
I've got Rory Mac at 66.4 percent to dethrone the champ and take home the strap Saturday night and am starting to get really excited. I see the judging tendencies as edging MacDonald's win probability even higher.
One big thing that jumps out is the areas we tend to see the most variation in values. Look back at all four charts. UFC 189's judges all pretty much know how to score a power strike to the head or body. We see much more variation in how they value lighter, jabbing touches to the head and power strikes to the possibly more foreign area known as the legs. How to value less damaging strikes is a tough one, but the leg results are intriguing. Should every judge get kicked in the legs by each fighter prior to the bout so they can appreciate the damage? What do you say Andy Foster and Bob Bennett? Let's make it a thing at the next Nevada and California commission meetings.
We also seem to have much wider variation in how the judges value grappling relative to striking. Many MMA judges have a boxing background and may be more tuned-in to striking, but I think there's more to it than that. Does grappling do damage? What makes it effective? If one guy strikes a little and the other guy grapples a lot, it can be hard to say the grappler won. Even though we know how tough grappling is, sometimes it just feels like the strikes had to win out. But not everyone necessarily feels that way.
Effective grappling was one of the bigger points of disagreement in the California commission's shadow judging exercise at Bellator 136. Mike Bell and I both scored rounds 1 and 3 for Alexander Volkov because, in my opinion, Tony Johnson was mostly cage pressing and throwing annoying mosquito shots. I did my best to give hardly any consideration to cage pressing and tried to focus on other, seemingly more effective techniques like Volkov's knees.
Then there's the issue of did my 15 seconds of mount overcome your minute of guard control or 40 seconds of half guard? What if you had 45 seconds of side control with a kimura attempt and I had 1 minute of guard control and 15 seconds with your back and a quick, failed rear naked choke attempt? And what exactly is a "tight" submission? It's one of the more subjective components of the FightMetric scoring process. Comparisons with striking are hard enough, but throw in all the combinations and permutations of grappling and it's no wonder we observe a wider variation in what our four judges consider to be effective.
Remember, tendencies may change when they're put out for public scrutiny and people are made aware of them. If there's interest, you may see more judge tendency articles in the future when big fights come around.
Be sure to follow all the UFC 189 coverage right here at Bloody Elbow throughout the weekend.
Enjoy the glorious fights!