Using Data Analytics to observe the Evolution of the Fight Game (UFC Data)

Purpose of Article

The sole purpose of this article is to assess a bold hypothesis, that is, in the game of Mixed Martial Arts (MMA), the grappling arts have now surpassed the striking arts as its predominant form of fighting. This hypothesis can be challenged in a number of ways. We shall explore this more under the "Data Understanding" and "Data Exploration" sections.

The Recent & Brief History of Fighting

In the mid-'80s, Mike Tyson was undoubtedly considered the "baddest man on the planet" and rightfully so, he managed the odd feat of knocking out 44 opponents in 50 bouts. It wasn't just his record that earned him the nickname, but the way he accomplished those victories - through vicious knockouts and in some cases, within seconds. Before Tyson (and he, himself, would attest to this), it was Muhammed Ali who took the world by storm with his flamboyant in-ring style. His exceptional ability as a boxer even got him the nickname "The Greatest". They were considered the best fighters of their time.

However, not all fighters are the same, and stay the same. There is a reason why people would say one is the best "of their generation", but if we place them in an octagon today, how would they fair? Granted, Boxing and MMA are both completely different sports and even the striking (especially distance management, etc) in the ring and the octagon vary significantly. But will we perhaps witness a rise of "baddest men on the planet" from non-striking backgrounds? Fighting has evolved tremendously, one would certainly have to take up more than one discipline to increase their chances of survival, let alone, standing victorious. A fighter who manages to integrate a few disciplines allows himself the flexibility to create more than one method of dramatically negating an opponent's strength. Additionally, having more than a single fighting style significantly increases the unpredictability aspect in a fighter's game.

Sidenote

As a practitioner of Brazilian Jiu-Jitsu though the Gracie Jiu-Jitsu Academy, I can almost certainly say that, in my opinion, striking isn't the first sport I would get into if self-defense is of upmost importance, and especially in Singapore's context. I say this with the backing of the strict enforcement parameters of what would be considered self-defense. In other words I shouldn't, at any rate, risk breaking my knuckles for self-defense. Running, on the other hand, doesn't require you to take up martial art lessons.

Data Understanding

Allow me to first justify the title of the article, specifically "UFC data". "UFC" stands for the Ultimate Fighting Championship and is the largest & fastest growing American MMA promotional organisation today. The UFC roaster features top level fighters in the sport and has a partnership with the United States Anti-Doping Agency (USADA) to consistently test their athletes for fair play. Therefore, I believe it to be appropriate that bouts taken place throughout the company's history, serve as a representative sample to accurately test the hypothesis. This data can be found on Kaggle.

Our data-set spans 27 years from 1993 to 2019. It comprises of a total of 5,144 and 41 rows and attributes respectively, with a total of 1,334 distinct fighters. Having sympathy for my readers, I will only be highlighting a few attributes that are key for the purpose of this article (I have not used the definitions as defined by the Unified Rules of Mixed Martial Arts to over-complicate things, Wikipedia does just fine);

1. KD is number of knockdowns (Wiki Definition: A situation in full-contact combat sports where a fighter is down or vulnerable)
2. TOTAL_STR is total strikes 'landed'
3. TD is number of takedowns (Wiki Definition: a takedown is a technique that involves off-balancing an opponent and bringing him or her to the ground with the attacker landing on top)
4. SUB_ATT is no. of submission attempts (Wiki Definition: A submission is a term for yielding to the opponent, and hence resulting in an immediate defeat. The submission can also be referred to as a "tap out")

These are the same attributes used by Rajeev Warrier, who contributed the data to Kaggle. I will be conveniently skipping the data cleaning stage in this article, but if you are interested, please send me a DM. The data cleaning and visualisations were done in Python & Excel. Alas, my student Tableau license expired a few months ago.

Data Exploration

Total number of last rounds by year

This graph depicts the total number of last rounds by year. This means that each bar represents the total number of rounds taken for all bouts collectively in a single year. The graph not only supports my earlier statement that the UFC has grown tremendously throughout the years, but for the majority of the bouts, the number of rounds taken to conclude a bout is 3. This information may not be useful at this stage but is worth-noting. Non-title fights in the UFC are all 3 rounds unless otherwise stated in certain bout agreement contracts (Eg: BMF title). I suspect most of these 3 last-round bouts are concluded by way of judges' decision. More importantly, we know the number of 1 round finishes have increased throughout the years. What we need to determine is what proportion of these finishes were via knockouts or submissions, and has the proportion changed? For simplicity sake, we will consider doctor stoppages as knockouts.

*Clearly there is a limitation in our data-set: 2019 is not complete and only collected mid-way (somewhat). Therefore to maintain the data accuracy, we will omit data tagged to the year 2019.

Average knockdowns

In understanding the line graph above, I should first distinguish what Red Corner and Blue Corner means. The colour distinctions are merely to let a fighter know where he or she should go after each round, that's it! Hence, since every fighter may have fought more than once in a given year, it can be that the data of red and blue corner strikes were thown by the same fighters (with similar MMA backgrounds because fighting styles take years to change). However, the key is this: these same fighters may or may not have similar approaches in their bouts. As the saying goes, styles make fights. Therefore, we can assume to take each individual red/blue corner fighter as a separate individual competing. Hence, it can be treated that both line graphs are representative.

The graph does prove that striking has been advantageous for fighters who contested in bouts from the beginning of the 21st century, with data suggesting a surge in knockdowns from 2000 to 2003/4. Despite fluctuations through to 2018, it suggests that in the long run - striking still remains very much important in the octagon. If we were to perform predictive modelling primarily based off historical average, we should expect to see more knockdowns in the coming years.

Average Total Significant Strikes (Landed)

I should first preface this analysis with a more in-depth definition of the term "strike" in an MMA bout. It refer to all strikes thrown at a distance, in the clinch and on the ground. Essentially they are attacks (legs, fists, elbows, knees) that come in without being blocked. The graph provides clear enough evidence to conclude that there is insufficient evidence to conlcude that the above hypothesis is true. Grappling arts may not have succeeded the striking arts, at least not yet? Well, like the average knockdown graph, it points to more significant strikes in the coming years.

Total Number of Takedowns & Submissions Attempts

In this quick section, I did not populate the takedown data to its average, but used the total sum instead. Since figures are always increasing, it does not affect our analysis.

From both graphs, we can move towards an expectation that future bouts will entail more submissions & takedowns. In other words, we can infer that fighters of the future would be much more competitive than ever before, possessing more complete skill-sets from different martial arts backgrounds. Although it is true that the stand-up game is where every martial art discipline begins, there have numerous debates that wrestling remain the most important discipline in MMA. Fighters who are armed with great wrestling pedigree will almost invariably have good submission defense. Knowledge of good submission defense usually entails good submission offense.

Fight Outcome in numbers by year & Concluding Statement

It is arguable that the days, when boxers are generally the most skilled fighters, are long gone. Though it may not be the case that grappling arts have surpassed striking, as the data showed, I believe with enough time (with just a mere visual inspection of the graphs' slopes), grappling will turn out to be more important one day.