Big cage / small cage statistical analyisis

cs10

Blue Belt
@Blue
Joined
Oct 3, 2012
Messages
619
Reaction score
0
Hey guys, was busy studying for a big finance test for the past several months but finally have some free time on my hands now.

I was thinking of trying to do statistical analysis of sorts and thought looking at a comparison of big cage / small cage events would be an interesting place to start.

Basically my thinking is to collect data on a couple hundred recent fights. Look at the odds for "goes the distance" and "finishes ITD", and then see if based on the implied probability of those odds decisions are happening either:

-more often (on average) then implied by the odds
-less often (on average) then implied by the odds
-and also break down the results based on if it was a small cage or full size cage.

I can find the odds and the results of the fights easy enough, but does anyone know where i might be able to find info on which events had which size of cage?

Any thoughts / opinions / suggestions are most welcome as well.

-Thanks
 
The problem is you will be comparing apples to oranges. On average, fights that occur in the smaller cages will be between lower-level guys. I would expect finishing rates to be higher because of crappier defensive skills more than cage size.
 
Hey Waiguoren, thanks for the input.

My thinking was by using the odds to adjust for the "implied probability" it should somewhat correct for that.

Haven't fully got my head around the best way to interpret it but here's a very basic example of what I'm thinking.

Lets say we collect data for 100 fights in the small cage. Of those 100 fights the average odds of a fight finishing ITD is say.... +125 (just an example), This would be an implied odds of 44.44%.

Now if over our sample fights in the small cage actually resulted in finishes 55% of the time. We might assume that, on average, the odds underestimate the probability of a finish when the fight is in a small cage, and we should skew are bets likewise.

something along those lines, let me know what you think.
 
The problem is you will be comparing apples to oranges. On average, fights that occur in the smaller cages will be between lower-level guys. I would expect finishing rates to be higher because of crappier defensive skills more than cage size.

This was my first thought too. I forget the correct term since it was a while ago I studied statistics but I think there will be a systematic error in the sample size since both groups are not similar. So even if you find out that fights in small cages ends in ITD more often it doesn't mean that the small cage is responsible for the difference(though it certainly will explain a part of the difference since you only need to use simple logic to understand that a smaller cage results in ITD more often).

Edit: It's good initiative by you TS but if you do happen to find an inefficiency(which is possible since MMA is not a mainstream sport and you are looking at a specific condition) the perfect way of getting rid of that inefficiency is to publish the results here so the sharps can move the line to a more accurate price. It doesn't require much for a tiny market such as fight ends ITD between random unknown fighters in a small cage to be priced more accurately.

If I were to do this analysis and happen to find a gold mine to be frank with you guys I wouldn't share the info and just keep it for myself. No offense.
 
Last edited:
Hey UAK, what if I only look at fights in the small cage. Then just compare the implied probability (based on the odds) of a fight going to decision with the actual results?

This way i would not be comparing apples to oranges per se, but looking at the implied probability of decision based on the odds, measured against what actually occured (i.e. the realized probability).

What do you think?
 
Hey UAK, what if I only look at fights in the small cage. Then just compare the implied probability (based on the odds) of a fight going to decision with the actual results?

This way i would not be comparing apples to oranges per se, but looking at the implied probability of decision based on the odds, measured against what actually occured (i.e. the realized probability).

What do you think?

Yeah that's a good idea. I would be surprised if the betting odds hasn't adapted(since we are talking about adjusting for a simple variable) to this yet because I think even a great deal of shertards are aware of the fact that small cage results in ITD more often but who knows, maybe most punters are completely ignoring facts. It also takes a while for new knowledge to become common knowledge and baked in to the betting odds.

100 fights is not big enough sample size though. Without doing any calculation I would say your margin of error is going to be too big for the results to be reliable.
 
Hey UAK, what if I only look at fights in the small cage. Then just compare the implied probability (based on the odds) of a fight going to decision with the actual results?

This way i would not be comparing apples to oranges per se, but looking at the implied probability of decision based on the odds, measured against what actually occured (i.e. the realized probability).

What do you think?
let's say we take fighter A vs fighter B. the fight took place in a small cage. the line for it to go to decision was -150 but fighter A finished the fight by TKO. what does that prove? the implied probability based on the betting line would be 60% to go to decision but one occurrence doesn't prove anything because 40% of the time, the fight WILL end ITD as the odds predict. unless you can see the results of 100 fights between the two fighters, you cant confirm your theory. and you cant take another fight that's -150 to go to decision and use it as a sample because the variables are totally different. does that make sense?

in my opinion, what you could do is look at # of strikes thrown by a fighter in a small cage versus his average in a normal sized cage. again, its tough because the fight is a one time occurrence and it could be a fluke of sorts, but i think if you can find that fighters throw/land like 2-3 more strikes per minute in a small cage versus their strikes per min in a normal sized cage, we can then infer that a smaller cage facilitates action. but you would need to have a big sample size to prove that and you'd have to check all the fighters individually
 
Last edited:
but what does that prove? let's say we take fighter A vs fighter B. the fight took place in a small cage. the line for it to go to decision was -150 but fighter A finished the fight by TKO. what does that prove? the implied probability based on the betting line would be 60% to go to decision but one occurrence doesn't prove anything because 40% of the time, the fight WILL be end ITD as the odds predict. unless you can see the results of 100 fights between the two fighters, you cant confirm your theory. and you cant take another fight that's -150 to go to decision and use it as a sample because the variables are totally different. does that make sense?

No you can't see the same 2 fightes fight eachother infinite number of times but you can generalize. TS is trying to figure out if the betting odds is consistenly underrating the probability of ITD occuring in a small cage fight. Not looking at specific fighters.

If you see that the implied probability of ITD is on average 45% for fights in small cage but they keep hitting at 53% then blindly betting on all future small cage fights will result in a positive ROI(if the betting market doesn't adapt and fighters suddenly don't start chaning their style or getting better however MMA is a fast changing sport, on average there are way less subs now than there were 10 years ago).

There have been cases of this before. Double digit hometown underdogs in NFL historically showed positive ROI but not anymore.
 
No you can't see the same 2 fightes fight eachother infinite number of times but you can generalize. TS is trying to figure out if the betting odds is consistenly underrating the probability of ITD occuring in a small cage fight. Not looking at specific fighters.

If you see that the implied probability of ITD is on average 45% for fights in small cage but they keep hitting at 53% then blindly betting on all future small cage fights will result in a positive ROI(if the betting market doesn't adapt and fighters suddenly don't start chaning their style or getting better however MMA is a fast changing sport, on average there are way less subs now than there were 10 years ago).

There have been cases of this before. Double digit hometown underdogs in NFL historically showed positive ROI but not anymore.

that is 100% false lol. there is WAY too much variance in an between UFC fights (namely specific match ups and fighters' styles) for you to make a profit, long term, by doing that.
 
Hi IGnP, thanks for the input.

I tend to mostly agree with you, based on the likely sample size we'll get and the variability it might be difficult to end up with a "statistically significant" outcome that would be usable to make bets on.

So...likely....even if i put this together the results could be mostly meaningless. I understand that is the most likely outcome but am still somewhat set on giving it a go at least out of interest.

However, I have to disagree with your last comment;

"that is 100% false lol. there is WAY too much variance in an between UFC fights (namely specific match ups and fighters' styles) for you to make a profit, long term, by doing that."

If we assume (i know big assumption, but lets make it theoretically) that what UAK about the implied and true probabilities is accurate:

"If you see that the implied probability of ITD is on average 45% for fights in small cage but they keep hitting at 53% then blindly betting on all future small cage fights will result in a positive ROI"

then I think he would be correct, you would make a profit by blindly betting on all future small cage fights (until those changes that UAK mentioned occur or the bookies adjust). Especially in the long term. As there may be short term variance where you lose several in a row, but in the long term if you have an edge of 8% on every bet you make you will come out a winner (i believe)

It would be similar to flipping coins getting even odds on heads or tails (but you know tails is actually 8% more likely). Over hundreds of bets you are vastly more likely to come out on top than behind.

Like i said though, in the grande scheme of things, getting a big enough sample size and discovering a large enough edge to make it worthwhile is highly unlikely so in that respect you are right.

I'm still hoping someone can point me to some info which lists which fights took place in the smaller cage so I can perhaps give this a shot!

-Thanks
 
Hi IGnP, thanks for the input.

I tend to mostly agree with you, based on the likely sample size we'll get and the variability it might be difficult to end up with a "statistically significant" outcome that would be usable to make bets on.

So...likely....even if i put this together the results could be mostly meaningless. I understand that is the most likely outcome but am still somewhat set on giving it a go at least out of interest.

However, I have to disagree with your last comment;

"that is 100% false lol. there is WAY too much variance in an between UFC fights (namely specific match ups and fighters' styles) for you to make a profit, long term, by doing that."

If we assume (i know big assumption, but lets make it theoretically) that what UAK about the implied and true probabilities is accurate:

"If you see that the implied probability of ITD is on average 45% for fights in small cage but they keep hitting at 53% then blindly betting on all future small cage fights will result in a positive ROI"

then I think he would be correct, you would make a profit by blindly betting on all future small cage fights (until those changes that UAK mentioned occur or the bookies adjust). Especially in the long term. As there may be short term variance where you lose several in a row, but in the long term if you have an edge of 8% on every bet you make you will come out a winner (i believe)


It would be similar to flipping coins getting even odds on heads or tails (but you know tails is actually 8% more likely). Over hundreds of bets you are vastly more likely to come out on top than behind.

Like i said though, in the grande scheme of things, getting a big enough sample size and discovering a large enough edge to make it worthwhile is highly unlikely so in that respect you are right.

I'm still hoping someone can point me to some info which lists which fights took place in the smaller cage so I can perhaps give this a shot!

-Thanks

You've got the theory right but you're trying to fit a square peg in a round hole. First of all, the odds for each fight are going to be different. The example of 10 point home underdogs in football is a bad example/comparison because getting the home underdog at +10 or more is always a -110 proposition. Every fight is going to have different ITD odds because of the match up and capabilities of the fighters involved. There is just way too much variance with MMA. Some fighters go in there with the aim to knock the opponent out while others look to jab for three rounds. Some want a sub while others want to lay and pray. And frankly, there isn't a big enough sample size for you to prove anything.
 

Forum statistics

Threads
1,234,841
Messages
55,311,599
Members
174,734
Latest member
Bob Gnuheart
Back
Top