An important first step for anyone looking to analyze MMA statistics is acknowledging the shortcomings of the widely available stats. On their own, most of these stats don’t end up reflecting what you want them to measure.
A prime example of this is significant striking accuracy rate. By name, that sounds like it’s a measure of the most accurate strikers. In practice, however, it’s largely a measure of what kind of significant strikes a fighter throws — as “significant strikes” includes power strikes thrown in the clinch and on the ground.
So as of writing, the top-5 in significant strike accuracy rate in UFC history includes names like Tatiana Suarez (2nd, 65.8%), Anthony Hamilton (3rd, 65.6%) and Josh Barnett (4th, 61.5%). Those three have also all had the majority of their significant strike attempts come from in the clinch or on the ground: 59.6% for Suarez, 60.5% for Hamilton and 55.3% for Barnett.
If we look back to UFC-wide stats since 2010, it’s no surprise that there’s a relationship between those rates. The average significant strike from distance has a 37.4% accuracy rate, the average significant strike in the clinch has a 68.8% accuracy rate, and the average significant strike on the ground has a 70.1% accuracy rate.
So yes, Suarez has landed more of her significant strike attempts than Anderson Silva (59.9%), but not many people would try to argue that she’s a more accurate striker than The Spider, who barely trails her despite the fact that just 25.1% of his significant strike attempts come from the clinch and on the ground.
Most Accurate Strikers in UFC History
|Significant Strike Accuracy||Adjusted Accuracy|
|1. Alistair Overeem||1. Alistair Overeem|
|2. Tatiana Suarez||2. Alexander Volkov|
|3. Anthony Hamilton||3. Justin Gaethje|
|4. Josh Barnett||4. Kevin Holland|
|5. Anderson Silva||5. Paulo Costa|
|6. Kevin Holland||6. Fabio Maldonado|
|7. Montel Jackson||7. Anderson Silva|
|8. Nikita Krylov||8. Chris Gruetzemacher|
|9. Justin Willis||9. Vicente Luque|
|10. Justin Gaethje||10. Fabricio Werdum|
|11. Evan Tanner||11. Jon Jones|
|12. Chris Gruetzemacher||12. Alexander Volkanovski|
And of course, we can take these ideas in the opposite direction for striking defense. The average defense rate on a distance strike is 62.6%, while that falls to 31.2% in the clinch and 29.9% on the ground. So an overall defense rate can be skewed in exactly the same way.
Clinch and ground striking is a skill, however, so the solution to this problem is not to just look at distance striking numbers if we want to paint a complete picture of efficiency.
There’s also another quirk to account for: striking pace — especially striking pace relative to an opponent. Someone who is incredibly picky about which strikes they throw can end up really inflating their accuracy rate, while someone who pushes the pace is almost necessarily forced to be throwing strikes that will generally be less accurate. This is reflected in the stats — as distance strike attempt ratios go up (i.e. a fighter’s pace relative to their opponents), efficiency numbers go down. Now I can see the argument for not adjusting here — if we’re talking pure efficiency, then someone who chooses efficiency over pace should rank higher. But in the interest of trying to capture fighter skill, and make this stat as useful as possible for analyzing fights, I choose to make a pace adjustment.
So where does that leave us with adjustments?
We want to account for the different accuracy rate of different kinds of strikes. We want to account for how often a fighter is throwing (or facing) each kind of strike. We want to account for how much a fighter is out-pacing (or being out-paced by) their opponent.
Using these adjustments, we can look at the UFC averages to see what accuracy or defense rate would be expected given the pace and types of strikes that were thrown. This “expectation” level would be the same for any fighter in the same circumstance, as we’re just looking at overall averages. Say for example a fighter attempts 50% distance, 25% clinch and 25% ground strikes while attempting 1.5 times as many significant strikes as their opponents. We can use the UFC averages to calculate the “expected” accuracy a UFC-average fighter would have there.
If we take the actual accuracy rate they posted and compare it to the average expected rate, we get accuracy vs. expectation. Doing the same for defense and combining the measurements, then, gives us Striking Efficiency vs Expectation (SEvE for short).
20 Most Efficient Strikers in UFC History by SEvE
|Rank||Fighter||Acc Rank||Def Rank|
|5||Germaine de Randamie||114||1|
Note that the goal of SEvE, as you can probably tell looking at the ranks above, is not to identify the most well-rounded strikers. Taking an example like Alistair Overeem, his adjusted accuracy is so strong that even a low adjusted defense rate doesn’t keep him off the list. Being well-rounded doesn’t hurt, though, as you can see with Demetrious Johnson who doesn’t rank top-25 in accuracy or defense, but still managed to rank 10th overall.
Those accuracy and defense ranks should also help explain some of the names that may not seem like they belong on the list. Liz Carmouche may not have a reputation as a striker, but an incredibly high 74.8% distance striking defense rate gives her a huge boost on the defensive side.
Something that really stands out here is just how good the all-time greats are. Even though we’re not adjusting for competition, Anderson Silva, Jon Jones, Demetrious Johnson and GSP all rank in the top 11. These are guys who spent large portions of their UFC careers fighting the top contenders in their weight classes, and still their efficiency numbers held up.
Future Improvements to the Metric
Also keep in mind that this is my first draft of the metric, and I believe there is still room to improve. Here are a few thoughts on where I could see that happening.
First, I’m still not satisfied by the pace adjustments. As mentioned above I do believe they’re worth applying, but I will be doing more research into how best to handle the adjustment, seeing if there are any better stats to use to quantify pace.
Second, I have something of a gut feeling that this metric is overrating defense. I have a suspicion that fighters have more “control”, so to speak, of their accuracy rate than they do their defense rate. That’s something that I will be researching in the future, and depending on the results there could be some value in de-emphasizing defense so it’s not weighted as heavily as offense in this metric.
Third, and back on the topic of adjusting for opponents. I’ve not completely given up on the possibility, but the big issue I see here is twofold. First, some fighters just don’t have big enough career samples to be reliable and worth adjusting for. But if we ignore adjusting for opponents with small samples and only adjust for opponents with big samples, then we run into some serious survivorship bias — we’d generally only be adjusting for disproportionately tough competition but never disproportionately weak, since weak opponents aren’t likely to last in the UFC long enough to amass a big sample.
Fourth is a possible weight class adjustment. Different weight classes carry different average accuracy rates. For example, Heavyweights land a much higher percentage of their distance strike attempts than Strawweights. By not adjusting for this, the offensive and defensive ranks don’t give a true “pound-for-pound” comparison. Now necessarily accuracy and defense rates in a given division will move by the same amount (if 40% of the strikes are landed, 60% have to be defended to get to 100%), but if I start weighing offense and defense differently, then weight class will certainly need to be accounted for.
Another thing this metric isn’t taking into account is knockdowns or power. Because knockdowns don’t inherently affect accuracy or defense rates (beyond the fact that it may be easier to land accurately on an opponent who has already been hurt in the fight), there’s no simple way to factor knockdowns in here, or to adjust for them. I also don’t think doing so would add much value to the metric — a separate metric to measure would be more valuable than trying to combine that into this one.
And finally, the big possible adjustment would come from more complete data being publicly available. The official stats break out significant strikes by distance, clinch and ground, as well as head, body and leg. If these were combined (i.e. distance head, distance body, distance leg, clinch head, clinch body, etc.) then we could far better control for differing accuracy rates on different strikes. Leg kicks are almost assuredly more accurate than head strikes from distance, for example, but to control for head/body/leg means not controlling for range, and again this stat would just showcase grapplers at the top.