9/04/2009 12:30pm, #121
The problem that niethan pointed out above was a small part of a bigger problem: I was using an old render based on a corrupted dataset, thus the actual ratings are wrong. I'll update with a new render sometime later today.“Most people do not do, but take refuge in theory and talk, thinking that they will become good in this way” -- Aristotle, Nicomachean Ethics, II.4
9/04/2009 4:03pm, #122
Bad news: I found a couple of bugs in one of the libraries I'm using, so it might take me a bit longer to get the next generation of the visualization online. (Have to fix them and send patches in the meantime).“Most people do not do, but take refuge in theory and talk, thinking that they will become good in this way” -- Aristotle, Nicomachean Ethics, II.4
9/04/2009 5:18pm, #123
- Join Date
- Jan 2008
Thanks a lot for the charts. This helps to decide what to train.
The "big four" and some leg locks looks like a good choice.
9/05/2009 7:55am, #124
I've worked around the broken library (it was a design flaw, rather than a simple bug). New rankings and visualizations are up. Most rankings are similar to what they were before, but there have been some interesting changes.“Most people do not do, but take refuge in theory and talk, thinking that they will become good in this way” -- Aristotle, Nicomachean Ethics, II.4
9/05/2009 11:17am, #125
NY Combat Sambo
- Join Date
- Feb 2006
- combat sambo
Late to this thread, but really cool.
I can't seem to view most of Jack attachements though.One of the best Bullshido investigations ever written: http://www.bullshido.org/David_Kujawski_Investigation
"disgruntled ex student who couldn't hack training with Dave and his material and opted out (could be called pussied out) of training to go to Sambo" - Mor Sao
9/05/2009 11:37am, #126
here. I'm working on re-doing the submission frequency analysis in a better way, after which I'll post it.
9/06/2009 10:03pm, #127
Thanks again for this neat analysis, Jack. Though at this point I feel like I've sated the bulk of my interest in the actual submission/statistics topic, and am following the technical details more closely.
Was the design flaw in the 3rd party lib?What a disgrace it is for a man to grow old without ever seeing the beauty and strength of which his body is capable. -Xenophon's Socrates
9/07/2009 8:51am, #128
Also, I'm going to put up a little web-app that automatically publishes per-division current ratings based on the most recent fights.
and am following the technical details more closely.
multigraphs, the consequence of which was that repeat victories of one fighter over another were being rated like single victories (back-and-forths were handled properly). For some fighters this didn't matter much (Fedor's three defeats of Nog just pushed him more securely into 1st place), for others it made a huge difference (Wanderlei Silva's three muggings of Sakuraba, for example).
I will likely add multigraph support to the library in question at some point in the next couple of months, but in the meantime I calculate the values correctly in my own graph layer (which runs much faster anyway), then copy over the data I want to display.
10/09/2009 9:38am, #129
Thanks to G-Off's unbridled awesomeness we now have weight classes for all fights. I've run all-time top ten lists for each weight class, and they mostly turn out how one would expect, but with a few surprises mixed in.
I'll format the results nicely and make them available in the near future, but in the meantime here are the top players by weight in the modern (post UFC31, 2001-) era:
Lightweight — B.J. Penn
Welterweight — GSP
Middleweight — Anderson Silva
Light Heavyweight — Chuck Liddell
Heavyweight — Tim Sylvia (yeah, I know!)
10/09/2009 2:19pm, #130
How'd Couture make out?What a disgrace it is for a man to grow old without ever seeing the beauty and strength of which his body is capable. -Xenophon's Socrates