Introduction, UFC 258 Stats Package, a Khabib-Usman Stat, the PFL and More
Welcome back!
Welcome (Back) and Introduction
First off, welcome back to everyone who had subscribed to the first iteration of my newsletter, and welcome to everyone new!
I’m a firm believer that there’s not much value for anyone if one writes something they’re not excited about. If I’m not excited to write a newsletter, why should you be excited to read it?
That’s why I fizzled out writing the old newsletter as quickly as I did. I tried to keep it to a pretty specific area, and just wasn’t excited.
So now I’m going to take more liberties in writing about whatever happens to be exciting me in a current week. It will mostly be MMA related, and when it’s not I’ll usually try to tie the topics in to MMA or sports analytics or sports betting. Those are things I’m passionate about, so I’m often making those connections anyway.
That means the topics will be more eclectic, and the newsletter will probably take on more of my personality than it had before.
I know that’s not for everyone, but I expect it will bring a lot more value to the people who are interested than it would bring anyone if I tried to force something I wasn’t feeling.
So what kind of topics get my creative juices flowing? Here are some thing I’ve been thinking about or excited about in the past month that may have been topics if I had been writing:
Conor McGregor simultaneously being over-rated and under-rated.
NBA Top Shots (Dapper Labs is also doing a UFC NFT, which I may write about in the future)
Context around Max Holloway’s significant strike record vs. Katter. Legitimately one of the most absurd records in sports right now.
The changes in DraftKings UFC scoring for 2021 and the rates at which different stats become more/less important
Betting markets for a high-profile fight (McGregor) vs. an undercard fight
How losing bettors convince themselves they’re winning
Diving into how much striking accuracy matters relative to volume (a.k.a “does Alistair Overeem’s record-setting accuracy rate matter?”)
The PFL Gets Me
The PFL holds a much bigger place in my heart than Bellator.
Bellator offers essentially no stats from their fights (at least to the public). The PFL has been tracking stats since the start of their first season. They get me.
The PFL having an even more exciting roster this year is a nice added bonus, and I’ll definitely be producing a decent amount of PFL content when they start back up again.
The stats they keep aren’t perfect (i.e. no way to differentiate a non-power clinch strike from a distance strike), but they’re a step in the right direction.
Some fun PFL stats so far:
the most standing strikes landed in PFL history belong to Rashid Magomedov (427).
The most ground strikes belong to Lance Palmer (570).
Across 5 PFL fights, Ali Isayev has landed 423 more strikes than his opponents.
Natan Schulte’s 35 takedowns are 10 more than anyone else has landed in PFL history
Kayla Harrison has 14 takedowns on 17 attempts (82.4% accuracy) and her 25 dominant ground positions are the most in PFL history.
The Highest Significant Strike Differential in a Decision Loss
Carlos Condit out-landed Robbie Lawler by 84 significant strikes in his split decision loss when they fought. Only one other fighter in UFC history has lost a decision with a differential higher than even +50 (Maldonado vs Pokrajac). I’m not a judge and I’m not saying Condit won the fight — just a stat I find #neat.
Julian Marquez by Submission …
… is still available at +900 (as of writing), if a long-shot bet is your kind of thing.
Identifying Context Around a Stat - Usman and Khabib
As is always the case when a Tweet picks up some steam, this one is getting some interesting feedback.
I’m always happy to defend stats or discuss limitations/concerns with people who are legitimately interested in a conversation, but I ignore people whose digs show they’re not actually interested in discussing it.
Behind every dig though is some truth, though, and some issues that are worth addressing.
A lot of people throw around a lot of stats so far removed from any context that they are at best meaningless, and often even misleading. I’m guilty of this myself. Sometimes I’ll share a stat I just find funny/interesting without adding enough context around whether it’s really meaningful or not.
So here I want to use this Usman/Khabib Tweet to highlight some important pieces of context for when we are looking at a seemingly impressive stat.
What the Stat Is Actually Measuring
The more complicated a metric, the harder it is to fully get this across in a Tweet. It’s absolutely crucial to understand both what a stat is measuring and what it isn’t. So in the example, it’s important to convey that this isn’t a rate from their entire fight time (which would be “control rate”), but a rate of the grappling time.
Sample Size and Relevance
Sample size is pretty well understood, so I don’t think I need to spend much time on it. If someone has 1 fight, scores a takedown and finishes it quickly, they’d be at 100% in this stat. But that wouldn’t mean much. So here, I use a pretty lofty sample of 1 hour grappling time.
Having the sample related to the stat is also important. We could set a minimum of 1 hour fight time (or 5 fights, or anything related to overall time) and we wouldn’t be out of the woods for sample size. Someone who has spent 55 minutes at distance and 5 minutes grappling isn’t overcoming the small sample problem above. The sample needs to be relevant to specific stat.
Selection Bias - Who is Being Compared
I don’t think selection bias is doing any damage with this example stat, but it’s still worth getting into.
Who isn’t included in this sample? Obviously anyone who hasn’t been in the UFC long (so they can’t hit 1 hour of grappling). But this will also keep out fighters who don’t grapple at all.
That’s not going to punish any strong offensive grapplers, but it could keep strong defensive grapplers out of the sample. For example, Merab Dvalishvili doesn’t reach the 1-hour grappling threshold, but his control time (2,953 seconds) would rank 101st among the 233 qualifying fighters.
A more obvious case, for the sake of example, is career significant strikes absorbed.
Getting hit with significant strikes is obviously not great, but that doesn’t mean that getting hit the most makes you the worst fighter. This is, in part, because if you’re the worst fighter in the UFC you simply won’t last long enough before being cut to rack up the most significant strikes absorbed (an honor that belongs to Donald Cerrone, with Max Holloway in second).
Arbitrary Cut-Offs and Competition
The 97% cut-off in my Tweet is totally arbitrary. I could just as easily have said they rank 1st and 2nd.
If the fighters ranked 3rd through 15th in the stat were all between 96% and 97%, hiding how close things are behind an arbitrary cut-off would hide some important context from the Tweet.
That isn’t the case here (the rest of the top-10 ranges from 89.3% to 96.1%), but it’s something to keep in mind when you see a cut-off like that.
Again, I’m not perfect here. On Twitter where I’m just trying to share interesting as much as I’m getting into in-depth fight analysis, I’ve certainly shared my fair share of stats that are technically true. But by giving some thought into the context around the stat, you can usually figure out just how much weight you should be putting into the claim.
As promised … UFC 258 Stats Package
As advertised, here’s my stat package for UFC 258. If you have any questions, shoot em my way, but it should be pretty self-explanatory. You can click file > make a copy if you want to be able to edit/play with the data yourself.
If you’d be interested in a paid subscription product that includes an even more comprehensive stat package (my advanced metrics and more) and/or one that is specifically designed around DraftKings UFC stats — please reply to this email or DM me and let me know.
That’s all for this edition, thanks so much for reading!
Any thoughts/comments/questions/topics you’d like to read about, don’t hesitate to reach out (by replying to this or by DM on Twitter/Instagram)!