I Watched Over 100 Fights on YouTube. Here’s What I Learned.

There are literlally thousands of fights on YouTube just waiting to be parsed through.

(Edit from author: thank you for all the interest in this article. If you’d like to see more data driven articles on fighting and training, I had a new website called highpercentagemartialarts.com)

In a former life, I was a data analyst at a college. I spent a lot of time combing through huge databases for information. We grappled with big issues of why students drop out, when they are most at risk, etc. This job impressed upon me the fact that data is a powerful argument. I found that people’s assumptions about how our college worked were often completely wrong (one time, I had to explain to a department head that it was mathematically impossible for her students to graduate on time). Sometimes long-standing theories everyone just assumed were true, weren’t. To my frustration, when confronted with hard data that refuted their claims, many people just ignored them. Or worse, they would try and find holes in our data. Despite that, the research findings of our small, but vocal, team started to slowly influence policy decisions on campus. Ultimately, we had the truth on our side. The cold, hard numbers. For the people who never wanted anything to change, no amount of evidence would convince them. But for the people who understood that all organizations have to move forward, they listened.

It’s a trend that is well underway in industries across the world. Politics, Investing, Sports, Gaming, everyone is looking at the numbers.

Like many things, the martial arts seem to be a step behind. We were slow to adapt to the internet, social media, better payment, and student information systems. After all, martial arts predates all those things. We were here before them and we’ll be here after them. That’s often the attitude. In all my years of training on the mats, I’ve never heard anyone cite a legitimate study on fighting behavior. Because they don’t exist.

But they can and they should. The data is there, it just needs to be compiled. Despite us living in one of the safest times in American history, people are still getting in fights. And more than ever before, they are recording and posting them on the internet.

So I took an early pass at gaining some information on how street fights actually unfold. I wouldn’t consider my work scholarly since it isn’t peer-reviewed. But I would consider it far and away more comprehensive than what we have now. And much more verifiable than that one guy in the dojo who has “probably been in a hundred street fights.”

What I Did

I watched 154 fights on YouTube and entered each one into a spreadsheet. I probably watched close to 200, but I had to restart my experiment a few times as I fined tuned what information was worth gathering. As it turns out, fighting is incredibly chaotic and it was difficult to answer what I thought were basic questions. How long does a fight last? Well, that depends on how you define a “start” and “end” The first punch? The first push? How do you define a clinch? How do you define a knockout? For most of these questions I had to come up with a subjective, but reasonable, definition.

Why 154 fights? This is an important point: I basically stopped when it was clear that the data was not going to shift dramatically anymore. I would have loved to have kept going, but I have a day job, too.

All of the fights I entered were real fights, none of them were competitions. All of them had to have been recorded in their entirety. I recorded males and females roughly evenly. Most encounters were males versus males or females versus females. None of them involved on-duty law enforcement, just “normal” people getting in fights. Here were some of my big takeaways:

1) Fights often have no clear winner

Some readers will cringe at my use of the word “winner.” Of course on a deeper level, no one really wins in a fight. But I have to define it somehow. When your standing over your unconscious opponent, you’ve “won.”

The most surprising outcome in fighting seems to be no outcome at all. 48.4% of the fights ended indecisively. In most cases, people simply got tired and stopped of their own accord. Bystanders tend to allow fights to play out, but would often step in when there was a lull in the action.

In fact, it seems that fights that drag past just a handful of seconds are unlikely to end in a clear way. Most people seem to have the energy for one, explosive onslaught of punches. If that fails to end the fight, a second onslaught just won’t have the same power. Turns out, fighting really doesn’t solve much.

2) Knockouts happen in the first ten seconds or not at all

23% of the fights ended in a knockout (which I defined as a single blow that incapacitated a participant). What was interesting about these is that more than half, 64%, occurred in the first ten seconds. After that, there was a sharp decline, but people still “lost” fights by being overwhelmed by punches. This was very likely to happen within the ten to thirty-second mark. After that, the likelihood of an indecisive fight rose dramatically.

3) Women always clinch

For the most part, men and women go about fighting the same way, with one major exception. While 55% of all fights involved a clinch, a whopping 79% of female fighters engaged in clinching. The reason is simple:

Women have hair.

bar chart showing me clinching 51% of the time.

pie chart with women clinching 79% of the time

When people aren’t winging punches in fights, they are grabbing whatever they can get ahold of. Women almost always use each other’s hair as a handle to steady their aim and keep the punches coming. They will also commonly use these grips to snap their opponent’s posture over. They might even get them to the ground completely with a well-timed jerk of the hair.

Such a minor thing seems to dramatically influence the strategy with women fighters. If you’re a woman doing martial arts, I would give serious attention to this aspect.

4) Bystanders usually let fights go on

The good news is that only 26% of fights involved a third party getting involved. This means that most fights that started between two people, stayed between two people. The bad news is that when other people did get involved, it was more often than not (68% of the time) the classic “friend jumps in” to join in the fight.

And in terms of someone facing multiple assailants, it’s a mixed bag. Like most fights, 37% of them had no clear outcome. In 26%, the outnumbered person was incapacitated. In another 22%, the person escaped or fled.

bar chart with various fight outcomes.
Feeling lucky? Only 11% of participants who faced multiple attackers were able to defeat even one of them.

But in only 11% of the cases of multiple attackers, was someone able to incapacitate any of his assailants.

The bottom line here is this, it’s highly unlikely you ever defeat multiple opponents. Run away instead.

5) Almost all fights will go to the ground and stay there

It’s an old cliche that “all fights go to the ground”. And basically, it’s true. Participants engaged in ground fighting 73% percent of the time. When you take out those ten-second knockouts that make up so many early finishes, the number jumps up to 83%.

What’s more, only 41% of grounded fighters were able to return to a standing position. Of those that did, more than half of them returned to the ground (57%).

In terms of outcomes, ground fighting has a major silver lining: violent knockouts drop dramatically on the ground, nearly by half. Only 29% of grounded fighters were knocked out or incapacitated by strikes. For standing fighters, that number jumped to 56%

The last thing worth mentioning is that 57% of the fights that went to the ground happened intentionally, meaning a participant made some sort of attempt at a takedown that worked. The rest was simply a result of people falling down.

6) No one uses “dirty fighting”

I was curious to see if I would witness “dirty fighting” techniques such as biting, groin strikes, fishhooking, etc.

Obviously, I did see a LOT of hair pulling among females. But as for the other things, not so much. Hair pulling aside, only 16% of the fights I recorded contained what I considered “dirty fighting” tactics. Interestingly, they were done by females about 80% of the time.

I read one time that males in our society tend to fight for dominance, and females tend to fight for survival. This generally seems to be the case when I observe fights. My theory is that men follow a loose set of rules when fighting in these “dominance fights” and seem to avoid certain tactics. Women do not.

I could be wrong on that but the bottom line is that dirty fighting is not common, especially among men.

In Closing

What do you think? Have you seen any other research on fighting that you’d like to share? Or maybe you’ve been in a hundred street fights and would like to chime in? While you’re here, check out a related article on our site about Developing an Independent Mind in Martial Arts. 

Also, if you like reading about martial arts and training, I have a book called The True Believers on Amazon. It’s about training in a martial arts school that became a pseudo-religious movement. It was some weird stuff.

 

[EDITORS NOTE: Another Martial Journal contributor has published a response article with his thoughts that can be read here.]

Louie Martin

Louie Martin

Louie is a black belt in Seibukan Jujutsu, Bujinkan Budo Taijutsu, and Enshin Itto ryu Battojutsu. He's certified in Gracie Combatives and is a Brazilian Jiu Jitsu blue belt under Daniel Thomas. He wrote a book about fanaticism in martial arts, called The True Believers.
Louie Martin

Latest posts by Louie Martin

    Louie Martin
    About Louie Martin 5 Articles
    Louie is a black belt in Seibukan Jujutsu, Bujinkan Budo Taijutsu, and Enshin Itto ryu Battojutsu. He's certified in Gracie Combatives and is a Brazilian Jiu Jitsu blue belt under Daniel Thomas. He wrote a book about fanaticism in martial arts, called The True Believers.

    67 Comments

    1. Well done. Some will say this information is obvious or common sense but truthfully it is very often ignored in training. We simply do not train well enough for real fights. So factoring this information in to how we train is important

    2. This mostly sounds like “social” fighting, not someone trying to seriously injure or kill the other party.

      • Hey Steve,

        Here was the breakdown of the fighting:

        40% were spontaneous fights between two willing participants
        27% were assaults were one participant was not willing
        5% were arranged fights with prior agreements
        27% of the time it wasn’t clear

        40% of arranged fights resulted in injuries that incapacitated an opponent.
        44% of spontaneous fights resulted in injuries that incapacitated an opponent.
        38% of assaults resulted in injuries that incapacitated an opponent.

        So, surprisingly, assaults had slightly less violent outcomes than fights. Maybe because more people in assaults simply fled.

    3. Thanks so much for this! From my observations of street fights in New York since the 70s, stories from bouncers and “tough guy” friends, and even my own few experience, you’ve captured the truth of the matter. A long-time martial artist and enthuseastic street brawler acquaintance used to like to say, “he who hits first wins.”

    4. Curious and probably can’t tell from random YouTube’s trained vs. non-trained fighters what the outcomes would be like regardless of discipline. And then Trained vs. trained varying disciplines, and I would think most trained fighter don’t engage in random street fights, its just not worth it.

      • I took a stab at that but it’s a little too subjective. The problem is the fights that last mere seconds. There’s not enough data to make an objective call on who might have had some prior training. I can tell you just my observations are that fighters with recognizable martial arts training did better on the whole, especially grapplers.

      • I was thinking the same thing, I watch alot of fight videos and noticed that when you see a trained vs non trained fighter most of the time they are teenagers. The trained fighters almost always come out dominant.

    5. Personal combat goes through predictable stages. Think of this as “testing the water”.

      Words, usually profanity, are followed by light touching. This is the finger jab that quickly escalates to the push.

      Within the first few seconds of contact, pushing escalates to some sort of punch.

      Punching also usually lasts for only a few seconds, for the exact reason stated; it’s a lot of energy and most people just don’t have that much endurance.

      Last stage is going to ground.

      There is a different dynamic between personal combat and mob, or gang combat. I propose that there is a different threat level and different outcome. I consider that latter form to be the more dangerous because of the potential for a frenzy style of assault.

      It also appears that there is a difference between daytime and night combat. The environment is also a critical consideration. Mid morning in on a city street is very different from an ally or under a bridge parking lot at night.

      The motivation for violence is an important factor in outcome. There is a difference between criminal gain – armed or strong arm robbery and an emotional response. Substance abuse in general and alcohol in particular has a strong correlation to violence.

    6. Very cool data collection and analysis! I’m curious how skewed the numbers may be because of the fact that someone went out of their way to upload and post the video on YouTube. Perhaps the video was uploaded because it was more interesting or had more action than the average fight? Interesting info nonetheless. Thanks for sharing!

      • That’s a great point and I think it ultimately does skew the numbers. But we can ultimately only work with the data we have, I can’t think of a better way to do it.

    7. I would agree women do fight dirtier than men. I was a CO in both a men’s supermax and women’s multi-level security prison and responded to plenty of fights.
      The fights between men certainly were motivated by specific causes; beefs, hits and truce beatdowns. The male fights were usually done in 20 seconds or when staff intervened; sometimes chemically, followed by “pig pile” applied mechanical restraints. But often it was done before the cops showed up and the bystanders usually functioned as refs and lookouts. Afterwards, the guys did their little vacation in the hole and you’d see the same fighters playing handball and chopping it up in the yard after their seg time was up.
      But the women! They were a lot more vicious. Whereas men tend to have a beef arise and immediately squash it, it’s not as predictable with women. They may settle it immediately or wait it out, make friends with the “bitch [they] hate” and out of the blue throw down on them in the dayroom 13 months later for some old shit that everybody (staff included) thought was settled.
      Another thing about the female inmates is they are very opportunistic fighters and will use closest weapon to target tactics. Usually fixed and unfixed environmental weapons; irons, staplers, three-hole-punches, closet doors, stair rails, picnic table edges, or my one time favorite a flying microwave oven.
      Whereas the men would usually cease fighting right after the cops showed up, the women would continue to fight after they were gassed and restrained and would sometimes require additional restraint (restraint board or the chair for the 300-500 pounders) after getting placed in seg.
      Conclusion, your data certainly verifies all of this about the ugly side of human nature.

    8. This points out the huge difference between fighting and self defense. These are all fights as far as I can tell. None are truly self defense situations based on the descriptions.

      • 27% of the fights I recorded were assaults where one participant was unwilling and unprepared. Violent outcomes were roughly the same between fights were two participants were willing and assaults where one participant was unwilling.

        • thanks for that addendum.I’m not sure what you mean though when you say the outcomes were all the same. I assume you mean that there’s usually not a clear winner, the knock out is in the first 10 seconds or not at all, they usually go to the ground, etc. My experience though tells me that pretty much everything about a self-defense situation is different from a mutual combat situation. Almost by definition, a self defense situation is “fighting dirty”. It might be true that the attacker gets a knock out right away or moves off if its a hard target — but that similarity belies a big difference in what it means to be a hard target in self defense as opposed to a good fighter.

          I’d think that maybe some different metrics could be applied to this analysis if you took the mutual combat situations out of the analysis. Just a guess.

    9. West Point did a study on incidents involving hand to hand combat in Iraq. They broke down the various ways that the action went, such as ground, standing, clinching, etc.

    10. Superb. Great research methodology and good results that speak volumes. I’m going to look for your book now and read more of your thoughts.

    11. Should put a disclaimer that, these are fights that Youtube allow on their website.

      You won’t see people getting attacked by a 3rd party with a machete or a gun on youtube,
      Or people knife fighting, or a gang of fighters knocking the opponent unconscious and proceeding to cut the others limbs off while their still alive.

      Youtube is PG rated fighting compared to what you’ll find elsewhere.

    12. Great article. Pretty much what I have been telling folks for decades.
      I was a boxer in HS, some utterly useless karate in college, and later a cop where I learned a little JUDO & BJJ.
      My 4 kids (2 girls then 2 boys) are all Black Belts in JUDO and all at least blue in BJJ. All have boxing and Muay Thai training and three of the four have done MMA. One of the things I have taught any who would listen was the critical issue of protecting themselves in a street fight. I tell them that their #1 goal is to take as little damage as possible for as long as possible. Most aggressors cannot sustain their attack for very long. I trained each of my kids to keep their hands up and never ever wince or turn away or close their eyes. “If you close your eyes they may never open again.” Every second they other guy is failing to hurt you he is losing any advantage he may have had. Soon he might be past the point of hurting you and being able to stop you from ending the fight.
      Here are my main trainees:
      1) http://bit.ly/1KzCTpq
      2) http://bit.ly/1KzCSlr
      3) http://bit.ly/1KzCV0z
      4) http://bit.ly/1KzCW4H

      • Good comment Mitch! Congratulations with your four champs! I would like to add: the better fighter you are, the less likely it is you’re being attacked or even challenged. You seem to radiate that you’re “tough to get”!
        In my experience as a “martial artist” the best fighting skills are boxing and BJJ. I combined it with Krav Maga, which is superior for defense against knives, baseball bats, gun threats and defenses against multiple attackers. But the base is to know how to use your fists and have the ability to throw powerful punches. And off course to avoid being punched, by dodging slipping, etc.

    13. Awesome work Louie! So fortunate to have someone looking at this from a scientific perspective. I am trying to do something similar when it comes to BJJ trying to find underlying reasons as to what makes a participant stay on the path versus the ones who quit. Interesting that you too started your path in BJJ through the Gracie combatives 🙂

    14. Super interesting, and great work man!
      Did you look at what attacks where most common as initiating attacks? Like the first physical attack that makes it kick off.

    15. Hello Louie,
      great work! I’m a blackbelt in some martial arts systems myself, I’m self employed in the wide field of violence prevention in Germany and I’m an instructor for law enforcement guys and instructors. I learned not to rely on what people told me on the mat but on scientific studies / or first hand experience from trustworthy people (mainly some police instructors I know). I watched many fighting / ambush / knife attack videos myself. I didn’t make a big data collection (due to the lack of time) but sorted some things out for myself (while discussing it with two other law enforcement instructors). If You like to exchange E-mails with me, feel free to contact me via my mail-adress: sk(at)gewaltpraevention.biz (and exchange (at) with @

    16. This is brilliant. I’m not sure the #1 most important lesson from this study is that all fights go to the ground. Of course we knew that already. I think the most important lesson (for street fighting) is: don’t get knocked out in the first 10 seconds! So footwork, boxing / KB defense, take-down defense.

    17. Louie, Great work! Are you familiar with the work by Rory Miller? I’m curious of how his definitions of different types of violence would apply to your data. Also for more criminally oriented videos, a good place is https://activeselfprotection.com/blog/. They’ve compiled a fairly big database of videos and may be willing to collaborate on the data mining.

    18. Nice work! Your data might be skewed by Youtube’s content restrictions though. Filmed fights that result in serious injury/death will likely be taken down. Therefore, “dirty fighting” may be happening a lot more than youtube fight statistics would suggest. Maybe do the same survey using http://www.reddit.com/r/watchpeopledie?

    19. Love this. Any conclusions on what were the most effective ways to “win” these fights? Did any commonalities arise from most “winners”?

      • I tried not to go there because I know people get passionate. To borrow from our friends at the Cobra Kai: “Strike first, Strike hard, No mercy.” isn’t the worst advice. I’m mostly kidding.

    20. Well done. I like this research report. It would be interesting to see the same research and data analysis done with a larger population, say 1,000+ video’s and also identifying outcome results taking into consideration the country & culture where the fight took place. Maybe treat each as separate groups and look at the variances between each. In that case, you can see if for instance Russian, Brazil, Japanese, American cultures, etc, etc. . . for instance have different habits and outcomes when it comes to the fighting process from initiation to end of the conflict. I know however, that you don’t get paid to do this kind of research, but I find it fascinating for someone who is also a practitioner in Martial Arts. I think what you have done so far is great!

      • Thanks a lot, I’m look for ways to improve. I’ve been doing a lot of research on sample size. With my number of fights (154) my margin of error is 7.8%. A 5% margin of error is considered baseline. For me to hit that, I should have recorded 377 fights. The good news is, there’s no need to record much more than that. Eventually the percentages will essentially stay flat. I’m working on another one right now, but it’ll take awhile.

    21. Very interesting, be great if there was more breakdown, Do people lead with there left? hand,or right Do people kick/? more,or less do the same size people fight each other,Is this across the board, country to contry, Im sure the military colleges must have this information, and if not I hope they pick up on your work , like you say you do have a day job, so thank you for your study ,

    22. These results are interesting, but the biggest drawback is that there’s no way to avoid some kind of selection bias.

      The most obvious selection bias is that in almost all cases, the fights on YouTube happen when third parties are present to record the fight with a video camera. Whether things would be different if no one were around is impossible to say. That’s not your fault, but whether the results would be different in a fight with no witnesses could be an important question.

      The other selection bias is whether someone thinks that the fight is worth putting on YouTube. While the world is full of stupid people who will record a friend doing criminal harm to someone else and then put the event on YouTube as evidence, some people are smart enough not to publish a video of a friend committing battery. That bias could reduce the number of conclusive fights that are shown on YouTube. On the other hand, fights that are too inconclusive might also be deleted because they just aren’t that exciting. Again, this selection bias doesn’t mean that you didn’t do an interesting study, but the selection bias could influence what conclusions we should draw from this survey of published YouTube fights.

      Obviously, you can’t know the background of all of the people in these fights, but I’d be curious about some general observations. For instance, were people who went to the ground but were able to get back to their feet people who appeared to have some formalized training? Did the ones who could stay on their feet after going to the ground appear to have more training than the ones who were brought down again? Some self-defense people seem to teach ground game with the idea of ground fighting being inevitable while others teach that the primary purpose of ground technique is to get off the ground. I’d be curious whether any of that showed in the fights. Did the people who threw knockout punches appear to have more formal training or did some other factor lead to the knockouts?

    23. Interesting study! Thanks. Surely you know Major W.E. Fairbairn’s book “Get Tough”, he was a street cop in Shanghai with tons of real world experience.

    24. I know there’s already enough people congratulating you for your exceptional work, but God you did a good job.
      Do you have the parsed data and would you be willing to share it? I would love to check your conclusions and maybe try to make some of my own, too! (I will mostly search for correlations between the different features you measure)

      I know the info was biased in various ways, but as you already said, unfortunately YouTube is the best source available (didn’t know about the other sites people mention) and it’s still a comprehensive analysis. I will definitely check out your book.

    25. Interesting research.

      However here are a bunch of factors that may compromise it.
      -most video taped violence is social violence.
      -criminals or people killing or maiming someone rarely film it and put it on youtube.
      -social violence that results in someone getting stomped to death also rarely makes it on the internet.
      -a fight between two willing parties is NOT self defense.
      -the background and experience of the participants might influence a lot what kind of tactics they used. In certain places they might be unskilled or act a certain way in other places they use completely different tactics and might be highly skilled.
      -on the video you also do not see the aftermath, people suffering damage and going to jail.

      Just that most fights go down like what you saw does not mean that all situations will allways go down like this. There is a huge difference between fighting someone, defending yourself against a drunk, getting jumped by a crowd or getting attacked by an armed criminal trying to kill you.

      So yeah your research might apply to certain social violence in certain groups however in other cases it might be going a lot different.

      Stay safe

      • Daniel, Those are the same points I was going to mention. Since it has been mentioned before, I am just going to add to the small choir here. The fact that someone was shooting a video means it was a certain type of engagement. So your research cannot be generalized to all interpersonal violence. My comments are in no way to take away from the excellent exploratory research that was made. Thanks for sharing!

        • Agree here and would love to see a comment on the sample characteristica somewhere in the actual article to add to both credibility and useability.

    26. Is anyone here familiar with the book “Verbal Judo”Could a study be made to help teach people to talk their way out of fights? Also in these comments two strategies are brought up 1) hit hard and hit first vs 2) protect yourself and let your opponent punch themself out. What are everyone’s thoughts on which approach is preferable?

    27. Nice to see numbers in street fights. Helps prioritize tactics where the numbers seem to go, for example invest in ground game and cardio primarily. Thanks for the effort, some of us appreciate the curiosity and hard work. Keep it up!

    4 Trackbacks / Pingbacks

    1. Gracie Combatives Student Watched Over 100 Fights on YouTube And Came Away With 6 Key Takeaways
    2. BJJ and Self Defense by the Numbers – Capital Mike
    3. You watched Over 100 Street Fights on YouTube, But I've been in them! - Martial Journal
    4. Weekend Knowledge Dump- June 29, 2018 | Active Response Training

    Comments are closed.