Common UFC Submissions: A Study By Numbers

**UpaLumpa** · Posted On: 6/30/2009 12:37pm

Originally Posted by Jack Rusher

Here's another pretty picture:

(Click for ~1300x1400 pixel version)

... a network graph of the top fighters in PRIDE. Nodes are fighters scaled according to (wins - losses), edges point from winner to loser with a label indicating how the fight ended. Fights in which the loser had never won a match or the winner hadn't won at least 3 more than he'd lost were culled.

Very nice.
Are you at all familiar with Neo Martinez's work using network theory and similar graphical approaches as what you presented here to characterize food webs?

**Jack Rusher** · Posted On: 6/30/2009 3:28pm

Originally Posted by [Tycho?]

Note that this is not the original spreadsheet. I don't currently have a way of sharing a document like that, but this should suffice for now.

Hey, I've got a Google Spreadsheet with all the UFC fights in it that I'd like to patch up with your observations about which submissions were used when. Any way I can get that from you?

**Jack Rusher** · Posted On: 6/30/2009 3:31pm

Originally Posted by UpaLumpa

Are you at all familiar with Neo Martinez's work using network theory and similar graphical approaches as what you presented here to characterize food webs?

Not until now. His ecological graphs are similar to some work I did last year on a graph theoretical analysis of systemic risk in the financial system.

The network diagram I posted before is the by-product of my looking into whether a graph-based approach could be used to auto-rate fighters based not only on how many fights they've won, but against whom.

**Matt Phillips** · Posted On: 6/30/2009 3:38pm

How will you handle the inevitable MMA-math cycles in the graphs? Hypersets?

**GHemmer** · Posted On: 6/30/2009 7:10pm

Never realize how many times the rear naked choke has been used, i thought the guillotine would be the most used for sure.

**Jack Rusher** · Posted On: 6/30/2009 8:59pm

Originally Posted by War Wheel

How will you handle the inevitable MMA-math cycles in the graphs? Hypersets?

I wasn't planning on doing the x beat y who beat z, so x must be able to beat z kind of thing. More like walking the graph of victories out from a given fighter, tallying up wins without double-dipping at any node.

**Domite** · Posted On: 6/30/2009 10:17pm

hey Jack, why does that first graph you posted taper off at the end? are there less submissions taking place recently, or did you include data from this current unfinished year and give that year it's own place?

I don't know the proper word for "place."

**Jack Rusher** · Posted On: 7/01/2009 8:18am

Originally Posted by Domite

hey Jack, why does that first graph you posted taper off at the end? [ ... ] or did you include data from this current unfinished year?

The unfinished year:

Originally Posted by Jack Rusher

The left taper reflects the relative infrequency of UFC events further back in time (5 in '04, 9 in '05, 17 in '06, 18 in '07, 20 in '08), while the right taper represents the end of the dataset (only 8 events so far in '09).

**Matt Phillips** · Posted On: 7/01/2009 9:06am

Originally Posted by Jack Rusher

Hey, I've got a Google Spreadsheet with all the UFC fights in it that I'd like to patch up with your observations about which submissions were used when. Any way I can get that from you?

Can you make this data available for others who would like to play with it. Like me ;)