It’s here because the current MLS Elo page is pretty awful. I’m meaning to do something about that in the near future when I get a couple days of free time.

Also, these ratings tend to not mean much until every team has played about 10-12 games (strictly anecdotal experience, no mathematical reason why) so yeah there’s some funny stuff. But just in case someone was wondering what was going on recently…

MLS Elo Ratings
1564.6888 – Seattle Sounders
1549.3225 – Sporting Kansas City
1545.2690 – Real Salt Lake
1537.6687 – San Jose Earthquakes
1530.5509 – New York Red Bulls
1525.1645 – Los Angeles Galaxy
1522.0902 – Houston Dynamo
1515.8164 – Colorado Rapids
1506.4407 – Chicago Fire
1504.5356 – DC United
1490.4999 – Montreal Impact
1487.3706 – FC Dallas
1485.6442 – Vancouver Whitecaps
1480.2150 – Columbus Crew
1477.4835 – Philadelphia Union
1466.3639 – Portland Timbers
1455.9092 – Chivas USA
1452.0821 – New England Revolution
1402.8849 – Toronto FC

 MLS Single Table

Every winner of the 100m since the inaugural event in 1983 has been black, as has every finalist from the last 10 championships with the solitary exception of Matic Osovnikar of Slovenia, who finished seventh in 2007.

Assuming that this success is driven by genes rather than environment, there is a rather obvious inference to make – black people are naturally better sprinters than white people. Indeed, it is an inference that seems obligatory, barring considerations of political correctness.

But here’s the thing. This inference is not merely false – it is logically flawed. And it has big implications not merely for athletics, but for the entire issue of race relations in the 21st Century.

…

The same analysis applies to the sprints, where success is focused on Jamaicans and African-Americans. Africa, as a continent, has almost no success at all. Not even West Africans win much.

The combined forces of Mauritania, Guinea-Bissau, Sierra Leone, the Republic of Guinea, Liberia, Ivory Coast, Togo, Niger, Benin, Mali, the Gambia, Equatorial Guinea, Ghana, Gabon, Senegal, Congo and Angola have not won a single sprinting medal at the Olympics or World Championships.

The fallacy, then, is simple. Just because some black people are good at something does not imply that black people in general will be good at it.

…

Imagine a similar argument using the Central African Bambuti, a black tribe more commonly known as Pygmies. With an average height of 4ft we could assert that the Bambuti are naturally better at walking under low doors. Would it be legitimate to extrapolate that black people in general have a natural advantage at walking under low doors?

Our tendency to generalise rests on a deeper fallacy – the idea that “black” refers to a genetic type. We put people of dark skin in a box labelled black and assume that a trait shared by some is shared by all.

Is it wrong to note 100m winners are always black?

First is the 2nd 2p2 NBA Auction which you can read about here and watch in real time here. Basically I created a blind auction platform where people can spend up to $1,500 to fill out an NBA roster of 10 players. The players are released 15 per day with a mostly random selection (all the top 100 players are guaranteed to go in the first week). Last time was a ton of fun with plenty of people overbidding early or snatching great values late. I bet no one learns though, and history repeats itself.

The second is the return of the MLS Elo Ratings! I talked about the difference in the new way I do the ratings a few months ago if you’re interested, otherwise just go check ‘em out. They’re certainly not perfect this early in the season, but I like the way they’re trending a lot more than I did the old method.

The big difference is that, in my opinion, game results (win/tie/loss) were being heavily overvalued against goal differential, especially for playoff games. 10-15% of the games were having way too much effect on the overall end-of-season ratings that should be based, at least somewhat, on all 30-something games.

I’m much happier with the balance between the two now.

After some input from a few people (including one of the guys from http://www.smartfootballrankings.com/) and a bit of experimenting I’ve made some very significant changes to how the Elo Ratings are calculated. As of right now I haven’t updated the original pdf, so you can still check out the basics for the original calculations. I expect to have the new version up with all the changes spelled out by the end of the week.

Or you can just read on!

The first and biggest change is that the ratings are now iterative. What does that mean exactly? Well it means that I run through the whole season 15 times in a row, recalculating after each game, in order to get the final ratings. I got to 15 loops because that’s the number of iterations I found to be necessary for the ratings to change less than 1/100 of 1% on each additional go around.

The two big pros I see from this method:

1) It makes sure that teams are accurately rated for each match earlier in the season. If Real Salt Lake turns out to be terrible this year (they won’t) we don’t want to give opponents too much credit for beating them early in the season when we thought they were the best team in the league.

2) It gives games more meaning. Each game is calculated in there 15 times now rather than just once. In a league that will only play 34 games this year, that means I should be able to get reliable ratings a little earlier in the year than halfway through this time, and hopefully have a better sense of what the results mean rather than just where the team started the season.

The biggest con:
I could not find one single example of anyone else doing this, so I kinda just played around with stuff until I was happy with how the numbers were coming out.

So with that gigantic looming potential negative in mind, allow me to explain some of the other changes.

K-factors, the values that change based on the game’s setting, have been drastically reduced. Last time it was 15 for a regular season game, 30 for a playoff game, and 45 for the MLS Cup Final. Now those numbers are 1, 1.25, and 1.5 respectively. This was done mostly for cosmetic reasons. When I kept the numbers at 15/30/45 for 15 iterations the ratings tended to span from 1200-1700 rather than the 1350-1625 that we’re used to and seemed to drastically overvalue the MLS Cup Final as a rating point. 1/1.25/1.5 got the numbers in line with where they were previously.

What I really should do is go back through the ratings with a few different variations, and then see how the real results match up to the expected to results given each matchups rankings. For instance, a team with a 100 point advantage playing at a neutral site should win 64% of the time. That would give us the most accurate model, but it’s a lot of work I haven’t put in yet.

As for the calculations, I still set all beginning-of-season ratings by halving the difference between the previous year’s end-of-season rating and the 1500 base. So if Philadelphia Union ended at 1450, they would start the next season at 1475. I will re-run the entire season’s results off of those base values after each match day, incorporating all new results from that day at the same time. One other idea I played with was changing the home field advantage in the calculations by team, but couldn’t decide if that was a good idea so I held off.

I’ll do a follow up post either tomorrow or Wednesday with the end of year rankings under each method. Suffice to say that I really like the newer versions output a lot more. I think the previous method gave too much weight to results and not enough to goal differential.

And one final note, if you’ve actually made it this far, I caution you to remember what the MLS Elo Ratings are and what they are not. They do have a large proponent of saying which team is the best in MLS, but it is not strictly a power ranking. It looks at the sum of a team’s results on the season, and rates them by most to least impressive.

Phew, that was a lot of words for a lot of work.

The bad news: (not shockingly) I found absolutely no way of predicting fumbles with any measure of success. The most interesting thing to me was that having a higher percentage of running plays was actually slightly negatively correlated with the average number of tackles it took before a team fumbled.

The other news: Didn’t do anything on Elo because it’s going to be an incredibly time-intensive process until I get a little better at python (which I “know” now!) and can script it all.

Two leading candidates right now:

1) Fumbles in football

How easy is it to predict who will fumble a lot? How random is the distribution of which team recovers the fumble? Under what scenarios are fumbles more likely to be recovered by the offense/defense? etc

2) MLS Elo as a predictive tool

This is the MLS Elo rating system I created. I’ve been talking with a student in Paris though (gotta love the Internet) and I think we may have made a couple improvements to it. I’m going to test these improvements against another piece of data I found recently, which is the gambling odds of every regular season MLS game from 1999 to 2009. Can Elo beat the bookies? We’ll know soon but, spoiler alert, without any adjustments it put up a 15% ROI over a season in the French league.

There’s nothing cooler than getting emails like this from people. So if you’re building something, put it out there for people to see. The response you get may surprise you.

As I said, I plan to read 50 books this year and this was #1.  Technically I did start the book before January 1, but it’s 700 freakin pages and I still read 300-something this year, so I’m counting it.

Simmons does a handful of cool things in this book which centers around his quest to rank the 96 best players of all time in order to fill out his hypothetical Hall of Fame.  The writeups for each player are enjoyable and well researched, so if you can get past the fact that he’s still a Celtics homer you should enjoy it.  Other fun wrinkles: the top 10 “what-ifs” of all time (What if Dr. J had never gone to the ABA?), the top 10 teams of all time (’96 Bulls got jobbed), his “wine-cellar team” (12 best players’ single years to making an unbeatable basketball team), and a fascinating few pages covering an interview with Bill Walton that makes me wish he was my grandfather.

Yeah it’s long, but it’s one of the best books out there if you want to learn more about professional basketball in America, and want to do it with some humor.

If you like basketball and have a few months to kill, definitely check it out.

The Book of Basketball: The NBA According to The Sports Guy

How coaches from one conference vote for various conferences

I only looked at the six BCS conferences here because I have to imagine you don’t care very much about the differences in voting between the Sun Belt and Conference-USA.

There’s not a whole lot to say here, it shows up pretty much how you’d expect.  Conferences rank their members higher, rank conferences close to them hire (ACC and SEC, Big 12 and Big Ten) and rank teams who are toss-ups with their own much lower (Big East ranking Florida 6th).  I think that the Pac-10 wins “Worst voter” award, although the ACC is pretty close thanks to the breadth of bad votes (mostly due to Bobby Bowden).

I hope you’ve learned this by now, but click the pictures to get a full sized version.

ACC Teams

Big 12 Teams

Big East Teams

Big Ten Teams

Pac 10 Teams

SEC Teams

And one final little nugget, a breakdown of how the top 25 would shake out under each conferences voting patterns.  The numbers to the right of the team are the number of points they earned.

How coaches from BCS conferences and mid-majors vote

The point of this is to see whether or not coaches in BCS conferences are making an obvious attempt to keep mid-major teams lower in the rankings, and thus lowering their chances for BCS games.  One reason to do this would be that mid-major teams in lower bowls means more money for the six major conferences to individually split among their members.

Similarly, the mid-major coaches may want to see one of their own do well, and thus will try to rank them higher than they probably deserve.

This one is going to be more pictures and less talk. Click any picture for a full-size version.

First, the votes for BCS teams by BCS conference and mid-major conference coaches.

And now votes for mid-major teams by these two groups

While the numbers do show a bias for coaches to vote for teams in their same classification there isn’t a huge difference, especially in the BCS Bowl spots.

Some of the biggest differences: Nebraska, Wisconsin, Arizona, Houston and East Carolina.

The part I like the most about this is the effect on the bottom end of the coaches vote, the last teams some coaches add to their top 25.  Oklahoma, UNC and UGA only received votes from BCS conference coaches.  Troy, Nevada and Middle Tennessee State only received votes from mid-major coaches.

One more picture: the top 25 (and other receiving votes) for each classification

Not a huge difference between the two until you get into the middle rankings.

The whole purpose behind this is to find evidence of bias in the secret poll that accounts for 33% of a team’s chance at a BCS bowl. The three things I am looking at initially are as follows:

1) The effects of coaches voting for their own team
2) How coaches from BCS conferences and mid-majors vote for these two classifications
3) How coaches from one conference vote for various conferences

This is part one, and part two will hopefully be up in the next day or two.

The effects of coaches voting for their own team

It seems kinda silly, first of all, that you would ask a coach who works 16 hours a day 340 days a year to rank nearly a quarter of the 120 teams playing DI-A football. Now add that to the fact that they are allowed to vote for their own team, who they are infinitely more familiar with than anyone else out there. Do you think there’s a 1% chance that Steve Spurrier watched a Wisconsin game this year even though he ranked them 20th?

So, here is a nice little breakdown of the coaches whose teams ended up receiving at least a single vote. I used the points formula instead of ranking (25 points = 1st place, 24 = 2nd, etc) so that averages make more sense. (click the picture for higher resolution)

So, as you can see, most coaches’ for their own team are inflated (props to Mike Riley, Jeff Tedford, Steve Spurrier, Ken Niumatalolo and Mark Richt for being the most fair. Tressel fans: obviously he did well also, but it’s harder to rank yourself higher when there are 5 undefeateds and 2 obviously better teams).

Now, some people could go through these and say that other coaches did decently too. “Hey, Mike Leach didn’t vote much higher than the average of people who voted for Texas Tech!” you could say. And that’s correct, but it’s also correct that 52 out of 59 people also didn’t even have them on their ballot, so giving them their second highest ranking (behind Bo Pelini, I can’t wait for the conference analysis) is pretty bad.

On average, coaches gave their team 2.23 more points than a random voter, and 0.87 more points than a random voter who we know voted for his team.

Ultimate punk list (based strictly off the numbers, not any of my own rankings)
5 – Bo Pelini – 6 spots higher than the average Nebraska voter (55/59 voted)
4 – Jim Harbaugh – over 9 spots higher than the average Stanford voter (52/59 voted)
3 – Bob Stoops – one of only two people to vote for Oklahoma, ranked them 23rd (Bobby Bowden had them 24th. We know he’s senile because he had Cal at 19, Clemson at 22, and USC at 23)
2 – Gary Pinkel – one of only two people to vote for Missouri, ranked them 22nd (Ken Niumatalolo had them 25th)

And the #1 total punk who is part of what is ruining college football

1 – Butch Davis – the ONLY PERSON out of 59 coaches who could find a reason to rank 8-4 North Carolina, he ranked them 24th, over Wisconsin and Houston among others.

The problem is, I got waaaay too in depth trying to find relationships.  As much as I’d love to research the history of every coach and assistant at each of these schools, I couldn’t do it.  For an example of what I mean, it is pretty awesome and telling that the 3 people who voted Georgia Tech #7 (the highest we received) where Ken Niumatalolo, the Navy head coach who worked with Paul Johnson for 6 years after playing under him, Charlie Weatherbie, who has Navy connections, and Dan Mullen, who played against Georgia Tech this year.  I’d love to be able to go through 59 voters and 42 teams and find all of these relationships and see what they cause.

Unfortunately, I just don’t have time for that so I’m going to stick to how coaches ranked their own team if that was an option, how they ranked teams in their conference, and how they ranked teams in the BCS conferences versus mid-majors.

That should be plenty good enough and I can get it up and finished by tomorrow.

So, you’re telling me I can write a paper about something really cool to do with sports statistics, and then if it’s good enough go talk about it in front of people like Mark Cuban, Daryl Morey and Bill Simmons?  Sounds like a plan.

Also, unrelated, applied for Startup Riot 2010 today.  Not to present anything, but just to attend.  So far, this year in conferences has been better than last as Venture Atlanta’s companies were better on the whole I thought, there was a much less crazy lunchtime speaker, Future Media the next day had some cool stuff, and I didn’t piss anyone off in the process of registering.  All around successes, imo.

These are the end of year rankings.

Playoff seeds are written next to the 8 teams that made it.

Freakin tragic that Dallas couldn’t keep their hot streak going for one more game to qualify for the playoffs.  They had such an amazing second half, and would’ve been amazingly dangerous to whoever had to play them.

Overall happy with how things turned out.  LA gets penalized here for winning close games and going 5-3-2 in their final 10 games.  Seattle, on the other hand, gets a big boost by winning their last 3 on the way in, including victories at Columbus and Kansas City.  Yes, even when the opponent is bad, wins on the road are impressive.

In a weird twist, possibly the best matchup of the playoffs is going to be in this first round between Seattle and Houston.  Unfortunately, it won’t mean as much as if they were to meet later on, but fortunately this is the only playoff round with home and away games, so we’ll get to see them square off twice.

LA/Chivas should be a lot of fun too, just to see the derby as a playoff match.  If LA wins, that’ll make for a great Western Conference final as well.

On the East side of things, blech.  I would probably rather gouge my eyes out than watch two New England/Fire games, and the Crew is going to patiently dismantle RSL in a way that is awesome if you’re a fan of the technical strategy they can employ but kinda boring when you want to see teams open up play.

I’ll go ahead and pretty much guarantee a Crew/Fire conference final, if only because it may be the soccer equivalent of those old Heat/Knicks playoffs series and I really want to see these two fan bases against each other.