MMA Ranking Talk

It never ceases to amaze me how little accurate information some people need to start opening their mouths. The amount of wrong information I see regarding the MMA rankings on this site and the ranking system used are insane. How about we wipe the slate clean and start from the very beginning…

Requirements for our new MMA Ranking System

In order to build something we should generally have an idea of the purpose. In terms of an MMA Ranking system, the purpose should seem obvious…to rank MMA fighters. Where this gets interesting is that there are several ways to rank MMA fighters.

Many sites seem to factor popularity into it heavily. Some sites like to focus on the very last fight someone ever had. For this “new” mma ranking system here is the base criteria.

1) Respect the fighters (they are professional athletes) and reflect on MMA as the sport that it truly is.
2) Be unbiased, consistent, predictable and accurate

That shall be the basis. To get away from the hype, from the subjective, from the inconsistent and in the process shift some power away from the matchmakers and over to the fighters.

Any ranking problems thus far?

Any problems thus far? Any glaring issue with either of those points? Anyone have more to add or think one of those points should be omitted from this ranking system?

Implementation of our new MMA ranking system

Since we know what we are aiming for we now set about implementing our new ranking system. Here is how the system works:

All MMA fighters start on equal footing

What this does is remove hype, popularity. Yes some MMA fighters might gain an advantage from being a BJJ Black belt or from K-1 Striking experience, but history shows that MMA is a different ballgame. As such, all mma fighters start at the same place.

Outcome determines rating changes

No getting signed to a new org and jumping 5 places, no getting into a title fight and climbing the ranks. What you actually do will determine how your mma rating and ultimately how your mma ranking changes.

Quality opponents are worth more

The better the person you beat, the more you gain, the more outmatched, the less you lose. We aren’t looking at pure mma fight records where someone 100-2 is automatically superior to someone who is 99-2.

That’s it!

Now take a seat, pull out some scrap paper, throw on the old thinking cap and let’s go through that one more time. The system we just built ignores hype, it rewards success and it factors in quality of opposition faced.

Yet again I pause and yield to questions from the audience. Who sees a glaring flaw with the above? Which of the three points mentioned above should be gone? Is there another point missing that should be included?

How the MMA Ranking system performs

We thought up the system, we adhered to our requirements and lets see what we got. Looking at the actual numbers we see our system is basically as unbiased as possible. The only cases where “human” interaction plays any part is on selection of values and those were deliberately chosen for accuracy.

The system is perfectly consistent and predictable. No more #1 losing to #5 and #5 moving to #1 while elsewhere #2 beats #1 and still sits at #2.

In terms of accuracy the MMA ranking system we’ve created above is quite strong. Overall prediction numbers are significantly better than a “fair coin toss” (roughly 33% better) and system integrity scores higher than some got in high school gym class.

Problems with our new MMA Ranking system

Pretty much nothing is perfect and obviously we can see some flaws with our results. This begs the question…are the flaws because of the system we built or some other factor? Looking back at the ground rules for our ranking system we see very few stipulations and almost nothing anyone has a problem with.

When we take a look at the actual ranking system basics and the design we see more sound thought and few issues. Turning to the results we see numbers that show high accuracy and remain true to our unbiased and consistent requirements previously set forth.

Puzzling…if the system is built on sound philosophies, and the numbers prove accurate, why is it we see some “glaring” issues in the results?

A possible answer

Someone much funnier than me and dearly missed once said, “Garbage in, Garbage out!”. They were talking about politics, but the theory can be applied in many places. In this case I’m talking about matchmaking. Let’s consider this situation…

An MMA Fighter is ranked too high/too low

Let’s build this from the ground up…

We know all MMA fighters in this ranking system start at the same ranking. That means there is no chance the fighter started off higher than someone that should be ranked below them.

We know fighters only get points when they win. This means the fighter had to have won some fights. Additionally we know fighters lose points when they lose matches, so that means in order for a fighter to be higher in rank they had to have gained more points than they lost.

We also know that quality of opponent matters. To the extreme that in some cases an MMA fighter will literally get zero points for beating an opponent. As above, we also know the system will take significantly more points from a fighter that loses to a much weaker fighter.

We also know the system has no idea who a fighter actually is. Our system is as unbiased as possible and pays no attention to name, or popularity.

What this leaves us with is a situation where the root cause of our problem is matchmaking. The good news is we can identify it, the bad news is that the solution is better matchmaking.

Think about a case where one Fighter is “too high” and another fighter is “too low”. If the matchmaker paired them against each other and the “proper” fighter won, then the MMA rating for the “too high” fighter would drop significantly and the MMA rating for the “too low” fighter would raise significantly.

This is just a simple case of the ranking system being able to self-correct itself. Now obviously we might not get the perfect matchup, and we might not get it right away, but whenever and however it happens, the adjustment will still happen. Eventually that “too high” fighter will lose and when they do they’ll drop even more rating for it. At the same time, the “too low” fighter is bound to turn it around and when they do, they’ll gain even more rating for it.

The honest truth

Here’s the deal. The MMA Ranking System on this site is very good. I don’t say it because I built it. As a matter of fact, if I hadn’t built it I’d probably be more positive towards the site and the ranking system. I say it because it’s true. The system has very sincere and respectful goals in mind and despite the few “eyesoars” that some like to focus on it does a fantastic job.

I’m someone who is incapable of being happy, so you can rest assured I’ll keep trying to make it better. What would be great though is if people could focus on what exactly the system sets out to do and how well it actually performs.

I’m really not one to point fingers, but re-read the above points. That is all the system sets out to do and that is all it does. When you realize how accurate the system actually is it becomes clear the problems many have actually lie elsewhere.

If you hate the site, you hate the site. That’s fine, stay away and if readership goes to 0 I can go back to building boats in bottles. I’m ok with that. All I’d really like though is some honest discussion. If there is really a problem with the few very specific points above then lets hear it. The above MMA ranking system is not perfect, but it’s also not as flawed as some insist on saying. Just stop and think about the points mentioned above and consider these situations:

- The Seahawks get to play the Raiders 12 times this year. The Seahawks go 12-4 and make it into the playoffs over the 49ers who go 11-5 (with 8 games each against the Steelers and Giants). Do we blame the NFL Playoff system because it said the 12-4 team goes or do we maybe look at the matchmaker that setup the schedules?

- The commissioner of baseball declares that NY is a major market and as such at least one NY team will *ALWAYS* get a playoff berth. The Mets do slightly less horrible than the Yankees (not this year, grats yanks on the birth) and get a playoff berth where they actually manage to win a few games past August 1st and take out their first round opponent.

If the ranking system in place had said the other team did better than the Mets up to that point, do we blame it? Should we really be flaming the ranking system in this case? Instead, maybe we focus on the fact that the series shouldn’t have happened at all.

Those are just two examples, but many more could be made. Players can get suspended, MMA fighters can linger on the undercard, MME can prevail over MMA at times, etc.

See the thing about me is I don’t mind being blamed for what I actually do. Look at the situation below and understand that if I were the older brother I’d be ok with whatever punishment I got. As the younger brother…not so much.

Consider an older brother hitting their younger brother in the head with a baseball bat and the younger brother getting yelled at because they bought the bat in the first place. *shrugs*

Tags: mma rankings, nfl, site, system

This entry was posted on Wednesday, September 23rd, 2009 at 5:52 pm and is filed under Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.