The All-Knowing Derbytron is now on derbytron.com

This site will no longer be updated. Go to derbytron.com for all the fresh Derbytron content.

6.01.2009

Rankings Update - June 1, 2009

 Team Rating2009 RecordVacuum RatingSOSTotal Games2009 Games
1Gotham 0.8502-00.9500.66182
2Windy City 0.7382-00.8050.63082
3Texas 0.7073-00.8760.494103
4Denver 0.6834-10.8040.52995
5Philly 0.6802-00.7570.61582
6North Star 0.6783-00.8950.45033
7Madison 0.6771-00.5810.56851
8Charm City+10.6707-10.7750.556118
9Steel City-10.6687-00.9050.39887
10Oly 0.6444-00.7480.53644
11Rat City 0.6082-10.6170.61973
12Bay Area 0.6001-10.5740.56672
13Dallas 0.5984-00.8220.38174
14Rose City 0.5931-10.5140.62862
15Kansas City+20.5892-00.7870.43162
16Detroit-10.5832-10.6210.56063
17Houston-10.5802-00.7360.44462
18Minnesota 0.5471-00.6010.50561
19Boston 0.5402-30.4920.57795
20Cincinnati+70.5314-10.6180.47595
21Atlanta 0.5203-40.4580.52187
22Harrisburg Area 0.5182-20.4840.51074
23Sioux Falls 0.5152-20.5810.46544
24Fort Wayne-40.5154-20.5520.44086
25Sacred City-10.5143-00.7640.35563
26Duke City 0.5081-20.4120.59783
27Providence-20.5072-10.4470.55673
28Nashville 0.5033-10.6960.37764
29Ohio 0.4923-20.2810.62995
30Angel City 0.4792-10.5810.41663
31Naptown+50.4713-20.5440.41385
32West Texas-10.4701-20.3650.52463
33Pikes Peak-10.4653-30.4510.468106
34Arch Rival+10.4562-30.3620.54175
35Carolina-20.4540-20.2250.66972
36Tucson-20.4521-10.4040.49262
37Burning River 0.4360-20.3420.52762
38Tampa Bay 0.4300-20.2530.51662
39Memphis 0.4161-10.5020.38232
40Maine+20.4002-30.3890.49175
41Northwest Arkansas-10.3931-10.3990.43362
42Grand Raggidy-10.3871-40.2410.538115
43DC+10.3772-10.4790.35863
44Assassination City+20.3712-20.3800.37564
45Omaha 0.3691-30.2870.47354
46Rocky Mountain+10.3570-30.2310.52673
47Salt City+10.3550-00.2910.36050
48Brewcity-50.3490-10.2560.50261
49Long Island 0.3320-10.0000.59731
50Dutchland 0.3130-30.1470.43133
51Slaughter County 0.3120-20.0170.55722
52Connecticut 0.3020-20.0860.48052
53No Coast 0.2940-40.0740.51574
54Big Easy 0.2850-40.1310.42374
55Alamo City 0.2790-20.2710.36262
56Green Country+30.2661-40.1800.37985
57Oklahoma+30.2501-20.0910.43853
58Dixie-10.2470-20.0140.44952
59Arizona-30.2361-30.1800.33464
60Dominion-20.2330-20.0730.43072
61Sin City 0.1550-10.2000.19731
         
 *Unofficial Team       

22 comments:

  1. who makes you the GOD of these? You need to get a life man! Denver #4 really? You blowing them?

    ReplyDelete
  2. I'll reword your comment to say this: "Why is Denver deserving of #4?"

    Well, this is a computer ranking so pretty much the only variables are scores and where the games are played. So, let's do some comparisons:

    Denver beat West Texas by 265.
    #3 Texas beat West Texas by 203.

    Denver beat Duke City by 47.
    #11 Rat City beat Duke City by 7.

    Denver beat Pikes Peak by 118.
    #10 Oly beat Pikes Peak by 143.

    As you can see, Denver's scores stack up pretty well against other top teams. Ignore what happened in April of last year or 3 years ago. It's no longer relevant. The teams playing today are not the teams of 2 years ago so maybe you saw Denver play a couple years ago and they didn't look very good, well, you should probably put that out of your mind.

    ReplyDelete
  3. How tight is your home track advantage? Is this why despite beating Harrisburg by 20 fairly recently at home, Providence is ranked 5 places below them?

    ReplyDelete
  4. Plus Carolina at 35?? Methinks your ranking system isn't quite ready for primetime...

    <3 anon 8:41

    ReplyDelete
  5. The home track advantage adds 1.4% to the away point total and subtracts 1.4% from the home point total. At the end of the season, I'm going to look a that number and see if it should be more or less. But, as for your question, no. Providence got a .713 rating for that game and Harrisburg got a .295. The reason Harrisburg is below Providence is mostly their performances in 2008. With the weight structure, Providence's 2008 games add up to a really bad 2009 loss.

    I will openly admit that Carolina is too low but I'll also say that every other ranking system out there has them too high. Carolina is only playing top 10 teams this year and while I appreciate it from a competitive standpoint and from a fan perspective, it makes rating them especially difficult as you don't get a comparison between them and a mid-ranked team that would better place them (or, if they are acurately placed in the DTR, make other people realize that they are).

    Let's look at the games that are being counted for Carolina:

    -27 Charm City HOME
    -39 Philly HOME

    2008
    +45 Atlanta HOME
    -64 Texas
    -64 Philly
    -40 Windy City
    +2 Boston

    IMO, they clearly don't deserve to be top 10 (unless they beat Gotham this weekend) as they've gotten creamed by every top 10 team they've played but every other rankings system has them top 10 (are those rankings not ready for primetime? ;) ). From my human perspective, I would put them 15-24, but for a mathematical equation, it's hard to truly rank them when they get beat by 25+ in every game.

    ReplyDelete
  6. and Carolina SLAUGHTERED Naptown (191 to 34)...and they are ranked ABOVE Carolina??

    ReplyDelete
  7. Ah, arguing over rankings, one of the great joys of the sports fan. I guess it depends on the purpose of your "all-knowing" derbytron. If it's to have fun with math, cool. If it's to predict the strength of teams in relation to one another, I would argue that there's much tweaking to be done. I will admit that I haven't seen Denver play since last year's ECDX, but I have seen Philly, and I find it hard to believe that the former would beat the latter head to head. In addition, I've seen both Madison and Charm City play more than once in the last year, and I would be surprised to see Charm City lose that game.

    This is as much fun as the BCS, and I look forward to seeing what happens as you have additional data to work with.

    ReplyDelete
  8. And Carolina took down Madison 131-54 during that same period.

    2009
    -27 Charm City
    -39 Philly

    2008
    +157 Naptown
    +77 Madison
    +45 Atlanta
    -64 Texas
    -64 Philly
    -40 Windy City
    +2 Boston

    I think teams like Carolina that are hard to rank make an interesting test case for your rankings.

    ReplyDelete
  9. Well, the difference between arguing this and the BCS is that the BCS actually decides who plays in the national championship which justifiably makes people angry. This doesn't decide anything so it should just be for fun.

    I saw Denver play last year as well and there can't be any doubt that they have improved exponentially in the past 12 months. There is no way they would have beat Pikes Peak by 118 last year or Duke City by 47 or West Texas or anyone by 265. That's why these rankings are only using recent games because teams can greatly improve or get worse over the course of a year.

    Carolina and Naptown unfortunately wasn't a WFTDA Sanctioned game otherwise I would count it and that would greatly affect the rankings. Unfortunately, I can't just pick which non-WFTDA games I want to use because that would be too inconsistent and I can't use all non-WFTDA games without including all non-WFTDA teams which is just not something I'm prepared to do at this time (ass-loads upon ass-loads of data entry).

    As for the Madison/Carolina game, is that WFTDA Sanctioned? If it is, I'd like to add it, but Flat Track Stats doesn't have it in their system either so I'm assuming it's not.

    It's still early in the season so when looking at these, you have to realize that most teams have less than 10 data points. That is a very limited amount of data. I believe the rankings stand on their own as they are but I think after the ECE is when you'll really start seeing the quality of the equation I've put together.

    ReplyDelete
  10. To Anon 4:32, check the lastest DNN rankings.

    ReplyDelete
  11. re:"the get beat by 25+ in every game".

    25 points is like 2 jams, no?

    ReplyDelete
  12. The Carolina-Madison game was a WFTDA-sanctioned bout. The Naptown bout was not. So in their last four bouts, Carolina is -27, -39, +77, +45, for a scoring margin of +14. Over the period that you've tracked, the margin swings to -14, hardly a 25-point differential. By that logic, it looks like Carolina's star should be on the rise.

    Of course, I don't necessarily agree that scoring margins are a good predictor of head-to-head, or even tournament, success. I think basketball is a good example, and a fair comparison to derby. If you look at the NCAA Men's Division I champions, they team with the widest scoring margin has won the tournament three of the last eight years. The first- and second-ranked teams by scoring margin have only met once in the last eight years.

    Here's the breakdown:
    2008-2009 North Carolina (1) defeats Michigan State (32)
    2007-2008 Kansas (1) defeats Memphis (2)
    2006-2007 Florida (3) defeats Ohio State (11)
    2005-2006 Florida (5) defeats UCLA (49)
    2004-2005 North Carolina (1) defeats Illinois (3)
    2003-2004 Connecticut (3) defeats Georgia Tech (12)
    2002-2003 Syracuse (23) defeats Kansas (1)
    2001-2002 Maryland (11) defeats Indiana*

    I know that scoring margin isn't the only factor used to determine the Derbytron rankings, but I just delicately suggest not relying too much on how wide the margin of victory is. There are a number of factors that contribute to that margin (e.g., the decision to use the second half of a blow-out game as an opportunity to play less-experienced jammers). In addition, some teams just aren't big scorers. We see this in every sport; teams that win consistently but don't tend to blow out opponents. It seems to me that the relative ranking of the opponent beaten would be more important than the margin of victory. If Carolina (35) were to upset Gotham (1) by 1 point on Saturday, it seems that (if I understand the Derbytron methodology properly), that victory would not be as "valuable" as Denver (4) defeating West Texas (32) by 265. That doesn't make sense to me, but then I've already expressed my thoughts on the predictive value of scoring margins.

    Regardless, I think this is an interesting and fun project, and as I said, I look forward to seeing the rankings after ECDX and to future iterations of the Derbytron.

    *Not in the Top 30, and information was only available for the top thirty teams.

    ReplyDelete
  13. @Anon6:43

    This is a point that I've seen mentioned numerous times elsewhere that I disagree with. How often have you seen a 25 point differential over 2 jams in close games? Probably once in 3 games? How many times have you seen the 25 point differential over 2 jams by the losing team? Once in 5 games? How many times have you seen the 25 point differential over 2 jams by the losing team in the last 2 jams of the game? Ever? I haven't seen it. The likelyhood of that happening is probably less than 1%, so I don't really consider it as a real possiblity. I'd compare it to a team being down 14 points with 2 minutes left in football. Sure, they can score a touchdown, get the onside kick, and then score another touchdown but how often do you see that happen? Once every five years?

    I like to look at games seperated by scoring passes rather than jams.

    Under 3 scoring passes (15 pts) differential = close game
    3-5 scoring passes (15-25 pts) = good win
    5-10 scoring passes (25-50 pts) = solid win
    over 10 scoring passes (50+ pts) = blowout

    ReplyDelete
  14. @Anon10:55

    Okay, do you have a source for the Carolina/Madison bout? I'd really like to get it in the rankings if it is sanctioned but neither derbymatic or FTS have it listed as sanctioned.

    You're right that scoring margin is a major factor but I'm not sure you're understanding how it is implemented. A blowout from around 50 and over will get the winning team a number between .9-1.0, so basically anything above 50 is seen by the system as only a slight variation. Like, Texas's 400-point blowout got them a 1 but a 120 point blowout would probably be something like .992 (sorry, I don't have the spreadsheet on the computer I'm on otherwise I could tell you the exact numbers). There is also a "bonus" for winning even by 1 point as it is using an exponent so the numbers rise quickly and then level off. So, a 2 point win would be like a .570 for the winning team and a .430 for the losing team before all of the other adjustments that come with it.

    In your Gotham/Carolina Denver/West Texas scenario, Denver and Carolina would probably get close to the same value for each of their wins, but it would actually have a far greater effect on Carolina's rating because their number is much further from Denver's. Now, if Caroline were to win by 20, they'd get a huge number.

    ReplyDelete
  15. My sources for the Carolina-Madison bout are the head referee for the bout and Carolina's WFTDA representative.

    I was started to respond further, but then I clicked the link in your methodology and noticed that you're relying on Bill James and sabermetrics to develop your methodology and realized that you're probably an actuary or a statistian (who may or may not play Strat-o-matic). Now that I get where you're coming from, I'll just say this. Derby is not baseball, and I hope that you will continue to evaluate and adjust your predictive model based on actual results in order to improve its effectiveness. I look forward to watching that progression.

    ReplyDelete
  16. I'll add that game to the next update and it should have a pretty decent effect. Thank you for pointing out the error.

    Derby is definitely not baseball and the exponent I'm using isn't even close to a square and I'm definitely going to look at it at the end of the season to see if I need to change it. Ken Pomeroy uses that equation for his college basketball rankings (kenpom.com) which I find to be one of the most accurate systems out there.

    ReplyDelete
  17. Arch Rival has played has played 10 sanctioned bouts since joined WFTDA. Not 7. We played 5 sanctioned bouts in 2008.

    WCR L
    Ft. Wayne L
    No Coast W
    NW Arkansas W
    Grand Raggidy W5

    Also 5 so far in 2009.

    ReplyDelete
  18. "So, only games from the last six months of the previous year are counted. Not all games are used if teams played more than five games in the those six months. In that case, only the last five are counted (unless multiple games were played on the same weekend (nationals and regionals being the prime example))."

    I'm going to change the title of that column as it seems to cause confusion.

    ReplyDelete
  19. You may have already answered this but - how often do you update your rankings?

    ReplyDelete
  20. Monday of every week. You can always see the latest rankings on the homepage at http://allknowingderbytron.blogspot.com/

    ReplyDelete
  21. Carolina vs. Madison was apparently not sanctioned. If you wanted a definitive source, you could have a WFTDA rep ask WFTDA Sanctioning about it.

    The folks at Flat Track Stats have a very definitive source of sanctioning information available to them (I know because I questioned their omission of a bout once). I'd be highly shocked if they got something like that wrong. It could happen, but I wouldn't expect it to.

    ReplyDelete
  22. DNN lists that bout as "WFTDA Regulation." Which means, "It's two teams that are from two WFTDA leagues, using WFTDA's rules, but isn't a sanctioned bout."

    Since the teams listed are their travel teams (and not home or B teams), this suggests to me that one team (probably Madison?) had someone skating who wasn't on their 20 skater WFTDA roster.

    Because of the crappy economy, skaters getting laid off, leagues finding it harder to fund travel expenses, sometimes you are faced with the choice of skating short two or three (or more!) skaters and making it an unsanctioned bout and subbing in some non-rostered skaters.

    By and large that decision really depends on negotiations between the interleague reps and the contract.

    ReplyDelete