Tapology Does Major Study of the Implications of Going Beyond 3 Rounds
by GregS123 on Oct 16, 2009 9:31 AM EDT
Email
Print
In most modern MMA promotions, a title bout means a chance to not only watch the highest caliber of opponents step into the cage, but to see their talents on display for a full 5 rounds instead of 3. Tapology’s analysis of bouts from 2008 and 2009 shows that the so-called “championship” rounds 4 and 5 dynamically alter MMA bout results. A large majority of non-title fights are finished either in the first round, or via decision. Yet title fights, which force fighters to gameplan for bouts that could be a full 66% longer, end up with a far broader distribution of outcomes, including more round 2 and 3 stoppages alongside stoppages in rounds 4 and 5, and fewer fights finished in the opening stanza. This suggests the current mixed martial arts landscape could look very different if non-title fights were extended to 5 rounds.
Excluding title fights, 40% of MMA bouts end in the first round. Only 17% are concluded in the second round, and a rare 6% finish in the third. The remaining 36% go to decision.
With title fights that go to 5 rounds there is a notable increase in matches ending in the middle frames, and a decrease in first round finishes. Round 2 or 3 endings happen in 42% of title fights, versus only 23% for non-title fights. This suggests the prospect of additional title rounds has fighters approaching championship fights differently, conserving energy and adapting their strategies and aggression as a result of having more time to work with.
Without advocating for or against a change, proponents of extending non-title fights to 5 rounds should take note. Not only would 3-round decisions get extended, but there might also be a significant impact on fights that otherwise would have ended in round 1. With fighters having the option of employing strategies intended to unfold over five rounds, all fights might become longer, not just the 36% that currently go to decision.
| Bout Type | Fight Count | Round 1 | Round 2 | Round 3 | Round 4 | Round 5 | Decision |
|---|---|---|---|---|---|---|---|
| Title Fights | 50 | 28% | 26% | 16% | 8% | 2% | 20% |
| Non-Title Fights | 752 | 40% | 17% | 6% | 36% |
Source: Tapology.com
The FanPosts are solely the subjective opinions of Bloody Elbow readers and do not necessarily reflect the views of Bloody Elbow editors or staff.
31 comments
|
3 recs |
Do you like this story?
Comments
but good post..
the issue is really worth taking another look.. I would like to see how much it changes when we get more title fights.
by Anton Tabuena on Oct 16, 2009 11:12 AM EDT up reply actions
A population of 50 is large enough to draw statistics from. The minimum size for good data is 35.
Keep firing Assholes!
Never trust a man with no shirt on.
It depends on the variance of whatever parameter you’re looking at. You might only need 8-10 samples in each set to draw a statistically significant conclusion…
Yeah, this.
If you took a random 35 AB sample of a baseball player, I don’t think you could draw any meaningful conclusions from it.
Twitter: @Mike_Fagan_13
http://www.sackmikegoldberg.com
Does this include WEC fights?
Those can provide even more 5-round fights for data.
Ever tried. Ever failed. No matter. Try Again. Fail again. Fail better. -Samuel Beckett
by Scott C. Broussard on Oct 16, 2009 4:45 PM EDT reply actions
Five rounds is too long for most fights.
What this data is most useful for is betting—it appears clear that if you’re betting on a stoppage in a 3-round fight, you should bet it will happen in round 1, and if you think it will get out of round 1, you should bet a decision.
Also something to consider: Championship fights are usually more evenly matched than the aggregate of non-championship fights, so having two veteran, well-rounded fighters may depress the number of fluke first round knockouts.
My guess is fighters are more cautious in the first round of a five-rounder but come to bang in the second round when the feeling out process is over, whereas if you give away 1 round of a 3 round fight, you’ve put yourself at a serious disadvantage to win by points, so you’ve gotta go out and bang from the start.
█♣█
A wise man told me don't argue with fools
Cause people from a distance can't tell who is who -- Jay-Z
Also something to consider: Championship fights are usually more evenly matched than the aggregate of non-championship fights, so having two veteran, well-rounded fighters may depress the number of fluke first round knockouts.
yes definitely. hard to actually try and measure or quantify this, but it’s good to debate.
www.tapology.com | twitter.com/tapology
“What this data is most useful for is betting—it appears clear that if you’re betting on a stoppage in a 3-round fight, you should bet it will happen in round 1, and if you think it will get out of round 1, you should bet a decision.”
The books don’t offer the same odds for each round.
Twitter: @Mike_Fagan_13
http://www.sackmikegoldberg.com
Tapology Spam
The volume and manner of the posts do not sit well with me.
it’s BE’s site and i am sure they will tell us if we are upsetting them. if so we will gladly listen. we love BE, hang out here all the time, and are contributing our stuff.
this article was submitted with more mellow design, but then was edited by BE to make it more prominent. i’m guessing that means they liked it.
we don’t think it’s spam but we’re sorry to hear it doesn’t sit well with you.
www.tapology.com | twitter.com/tapology
i think also, in a non-title fight, guys are working to make their way to a title shot, ie: lay n’ pray is a more viable option to work your way up the chain. in a title fight, guys are more willing to let it hang out and get the win, as you tend to have to really take it from a champ to win a decision.
Gatti. Dekkers. Pele. Aoki. Kang. Vanderlei.
http://theworldsoldestsport.blogspot.com/
by theworldsoldestsport on Oct 17, 2009 3:37 PM EDT reply actions
Ugh
I commend you for trying to think about MMA statistically, but you have to do a better job before publishing. To start, your sample size is probably too small to be statistically significant. Of course, I have no way of knowing, because you didn’t reveal whether you had done any tests for statistical significance and what the results were. Without that level of confidence, these numbers represent anecdotal evidence. Statistical analysis should always be a cut above anecdotal evidence.
Second, I think you need to be more careful distinguishing between correlation and causation. Even if the statistical significance tests come back favorable, it would not prove a causal link between title fights and early finishes. As an example, I would hypothesize that title fights tend to be larger mismatches than normal fights. I would want to see the finishing percentages for fights with similar odds to title fights. Perhaps it’s the greater differential between perceived fighter skill (as measured by odds) that is the key to earlier finishes, independent of title.
My point is not to discourage you, just to ask for a greater level of analysis. And if you need a great lesson in correlation vs. causality, check out this brilliant XKCD comic:
The sample sizes are listed on purpose. You can go to town and do all the number crunching you’d like. We used words like “suggest” and “might” because obviously there are a lot of factors to think about.
As an example, I would hypothesize that title fights tend to be larger mismatches than normal fights. I would want to see the finishing percentages for fights with similar odds to title fights. Perhaps it’s the greater differential between perceived fighter skill (as measured by odds) that is the key to earlier finishes, independent of title.
Yup, now that’s a good, constructive question.
www.tapology.com | twitter.com/tapology
“The sample sizes are listed on purpose. You can go to town and do all the number crunching you’d like. We used words like "suggest" and "might" because obviously there are a lot of factors to think about.”
Why wouldn’t you list the sample size? The point Annie is making is that these numbers are near worthless without any sort of test for statistical significance. And you may use words like “suggest” or “might” to seperate yourself from any sort of commited stance, but then what exactly is the point? And why is the term “major study” in the headline?
Twitter: @Mike_Fagan_13
http://www.sackmikegoldberg.com
someone from BE edited this post heavily
i didn’t write major study
www.tapology.com | twitter.com/tapology
what exactly is the point?
the point is not to rock your world. maybe you are over-analyzing what we’re doing.
all we did is show that the 50 title fights over the past 2 years have had different outcomes than the 750 non-title fights. the results are pretty interesting, and pretty basic.
a lot of people on this site in the past several months have talked about whether or not all fights should go from 3 to 5 rounds. here is some information that is interesting that you could look at when you talk about that issue.
www.tapology.com | twitter.com/tapology
It sounds like you’re backing away from your findings by admitting that it’s basic and using weasel words like “might” or “suggest.” Don’t do that! You’re so close to really having something to say. If you could just go that last little bit, you might be able to show something that no one has seen before. My point is to say “just do it.” If you own a copy of Excel, you already have all the tools you need to do tests for statistical significance and basic regression analysis. If you need help doing it, there are numerous tutorials online for running stats using Excel.
My word of caution is around putting out preliminary or incomplete analysis. In general MMA analysis, people might criticize, but no one can eviscerate, given that all analysis to this point has been subjective. You just have to be a little more careful when you start doing statistical analysis because people can truly and objectively refute you. And if I know stats people, they’re just itching for a chance to tear someone a new one.
haha wait…. aren’t weasel words like “some people even say he’s got a bad ground game” without naming any sources. i didn’t know that being non-committal also counted.
you make good points though i do stand by the original write-up. and i really like your questions about the odds for title fights vs non title fights.
i am sure that a big statistical analysis would confirm that “yeah, 50 is so-so but no’t great.” the purpose is to share some interesting info and create some conversation. (again for the record… i never called this a major study. original post title that i wrote was far more subdued).
www.tapology.com | twitter.com/tapology
What were trying to say is that the information isn’t interesting to anyone with a basic handle on statistics because there’s really no useful data being shown. It can’t be used in any discussion of three/five round fights because it doesn’t tell us anything (outside of the direct observations of the fights).
Twitter: @Mike_Fagan_13
http://www.sackmikegoldberg.com
I would hypothesize that title fights tend to be larger mismatches than normal fights. I would want to see the finishing percentages for fights with similar odds to title fights. Perhaps it’s the greater differential between perceived fighter skill (as measured by odds) that is the key to earlier finishes, independent of title.
OK here it is… just for you my friend!
Same sample of bouts used in the above post, except many of these non-title fights did not have odds posted. So it is n=50 title fights but only n=692 non-title fights.
In the title fights, the average favorite was -308 and the average underdog was +242.
In the non-title fights, the average favorite was -295 and the average underdog was +223.
Pretty similar, though this is just a straight, unweighted average and I’m not sure if there is a better alternative way to calculate it. Also, every single fight was split into favorite vs underdog, even if it was almost even. So, a guy who was -110 but going up against a guy -120 would still be classified as the dog.
www.tapology.com | twitter.com/tapology
It’s tricky to use averages because you’d also need to know the standard deviation to get a sense for what the variance was. Remember that 5 and 7 have an average of 6, but so does 11 and 1. It might be more useful to use the median. The median should be very informative in the larger sample, but not quite as much in the smaller one. Still, I think the median is a better metric here than average.
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
MMA Rankings
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
