UFC 271 features the final fight for one of the true pioneers of Women's MMA, the incomparable Roxanne Modafferi. But in today's article, I wanted to attempt an exercise in data mining to retire her with the lone finish in her UFC career. Her opponent this week is the rising star "King" Casey O'Neill, an undefeated prospect blazing her way through the UFC Women's Flyweight division. Let's first talk about Roxanne, then Casey, and dive into some math.

Roxanne started her professional MMA career allllll the way back in 2003! I meant it before when I say she is a true legend of the game, Roxanne fought for the inaugural UFC Flyweight championship unfortunately losing to Nicco Montano. As I stated before, Roxanne has never been finished in her UFC career and across 44 professional MMA fights she has only been finished a total of four times! 3 submissions, 1 KO by slam. In terms of fighting style, Roxanne is an old school BJJ practitioner who has developed her striking over time but never really to a level where should could compete with the top of the Flyweight division. Taking a look through her stats is an incredible experience, I watched her fights and was struck by how incredibly hard her guard is to pass. The stats back this up:

Take particular note of just how little time she has spent in side control, mount or back control in her 11 fight career. A truly incredible accomplishment. Let's look at some quick stats on her output to back up my contention that she is essentially a control grappler:

To summarize above, Roxanne averages more ground control time per takedown than 70% of Women's FLY in UFC History but ranks sub 50th percentile in every other measure along with an abysmal 9th percentile in Clinch Strikes. Roxanne wants to grab you and hold you to ride out a decision victory. So with the data dump out of the way let's talk about the star of this article: Casey O'Neill.

Where Roxanne is an aging legend retiring after this fight, Casey is a fighter on a meteoric rise up the FLY rankings. Casey is undefeated in her UFC career and has put up genuinely prodigious grappling statistics along with finishing all of her opponents. As a side note, I have a tendency to over use the term prodigious in my writing and videos but Casey is truly a prodigy, look at these stats:

In three UFC fights spanning 7ish rounds of total round time, Casey has more ground control time than 76% of Women's FLY in UFC HISTORY. Not only that, in that same time frame she has more Mount and Back control time than 90% of Women's FLY in UFC history. Casey's grappling dominance is savant like. Let's transition into some scouting notes I took. Casey is a highly pressure oriented grappler. She is constantly in her opponent's face pressuring with strikes and takedown attempts. She is highly finish oriented in her grappling, she expertly switches between strikes and submission attempts based on what her opponent give her. One key thing I noticed watching Casey was how she uses her locked down knee in top half guard. Casey loves to use a knee on belly power pass from top half to mount. She uses it over and over on her 3 opponents. She also really likes to transition from a traditional top side control to knee on chest position. Casey has a real feel for transitions between traditional and high mount, the best fight to see this is her battle against Antonina. Casey could be described as a bit reckless in her pressure but she is only 24 so she has plenty of time to capture that aggression and put it to more efficient use.

But enough talk, let's get to the true reason for the article. Let's mine some data friends. I am going to give a quick overview of the math in the following paragraphs so if you don't care skip to the next one after the big "Discussion and Conclusion" and check out the results.

Alright for those that have stuck around for the math, time to dive in. Essentially, what I am doing is using a technique called supervised machine learning. More specifically: linear regression. The idea is that if I take a number of variables, I can attempt to train a machine learning model to predict a specific variable or variables. In English, I took all of Casey and Roxanne's control and controlled time in different positions by round for their entire careers as well as their ground strikes landed, ground strikes absorbed, opponent submissions attempted and submissions attempted by them those rounds and taught a model to predict those values. In essence, the model doesn't really do anything. Why would anyone want a model that predicts statistics that have already happened? No one would! But if we think about it, if we peel back the layers of that model and investigate the math it is using to make those predictions we can make some assumptions about the fighter's styles. Each of those variables I described above (control and controlled time by position) gets assigned a predictor value by the model. If I look at those predictor values and analyze them further I can begin to understand where Roxanne and Casey want to go in grappling to strike, get struck, attempt to submit or opponents attempt to submit. Have I lost you? Let's look at the numbers, it will help to see examples!

Let's start with Roxanne and her top 2 and bottom 2 in each category. Remember that a positive predictor value means its a hot spot to target and a negative value is a cold spot to avoid targeting:

So let me break down the above. What I see as an MMA Data Analyst is that Roxanne wants to control her way into mount (her top hot spot for landing strikes) and then rain holy hell down on her opponents. She also is incredibly active off of her back in half guard which is her second greatest hot spot. As far as her cold spots (the places that Casey is going to want to keep Roxanne in to avoid absorbing damage) Bottom Full Guard jumps off the screen followed by Scrambles in second place. So let's now look at where Roxanne takes the most damage:

Roxanne's grappling defense is so good it is actually pretty tough to find hot spots to target against her. Bottom Mount is a statistical anomaly (1.02% of total time) so really you have to attack her from top Half Guard. Roxanne's incredible guard is also a huge cold spot for landing damage on her, considering it is nearly a quarter of her total ground time. She really does an incredible job of avoiding ground strike damage. Getting a ground strike KO/TKO for O'Neill is going to be really tough looking at these numbers. Before we transition to Casey, let's check out Roxanne's sub defense hot and cold spots:

Same exact story as before, Roxanne has a liability in bottom half and her guard is TITANIUM. Now let's talk Casey and come up with a gameplan:

Really simple breakdown here, O'Neill has a very active striking bottom side control but the most important takeaway is O'Neill wants to get to top mount to land strikes.

Alright, math is finally out of the way. Let me summarize the table and lets get to the predictions and analysis. Casey has two main hot spots she wants to attack you with submissions: scrambles and back control. She is 2x more likely to attempt a sub in scramble time than in back control however. Most interestingly from our previous analysis, she doesn't seem to want to attempt subs from mount instead preferring to rain strikes.

So what do I see and what do I predict? Based on the data and tape study, here is what I would recommend to Casey if I could speak with her. Casey. Fight as you normally would but when you get into top half guard as you always do, attack arm triangles to hide your knee on belly power pass to mount. Roxanne statistically has a liability in her bottom half sub defense, take advantage of that to get to your favored control striking position. Mount. Easier said than done though Casey. Roxanne is the greatest guard I have studied statistically so mounting her would be a truly incredible feather to add to your already very feathery cap. This to me is the key to setting up the finish on Roxanne. Casey uses great clinch trips and scramble situations to establish her top half guard. Roxanne essentially will be more willing to give this up due to her propensity to strike actively from bottom half. From there, Casey needs to ride out Roxanne's mount defense, attack the rear naked choke if Roxanne gives her back off the mount or rain strikes down if Casey can secure the mount control and keep Roxanne from rolling over.

And finally, my prediction. I think Casey O'Neill will be the first fighter in UFC history to finish Roxanne Modafferi and I project it occurs by submission. The main area I left out above is Casey's biggest submission hot spot. Scrambles. Casey has a great feel for leg lock opportunities and is incredibly aggressive in these positions. Casey is also a very talented positional no gi grappler in my visual study. I think she will be tremendous at opening up these opportunities on Roxanne and finishing them when given the opportunity. Modafferri statistically hasn't really been tested by a talented no gi grappler in the mold of Casey. On the other hand, Casey's incredible knee mount pass could very well lead to mount control and a ton of damage to Modafferi that she eventually succumbs to or rolls over giving up her back leading to a finish by RNC. I personally see the scrambles as a huge opportunity for O'Neill but don't discount the RNC in this one.

Interested in learning more or seeing some other data related content? Check out our video on Robert Whittaker for 271 here and consider signing up for our newsletter here. We release a weekly data breakdown of the past weekend's fight every Monday morning! It is well worth it I promise!