Posts Tagged ‘Ty Ratings’

Predictions for tonight’s most consequential college basketball games

February 16, 2012

I’m test driving a prediction system I may use for the upcoming NCAA men’s basketball tournament, depending on how successful it proves to be.  For tonight’s two “impactful” games, I predict the following scores, with each team’s win probabilities next to them:

Predictions  16-Feb
Wisconsin 58 21%
Mich St 63 79%
NC State 72 27%
Duke 83 73%

System Explained

To make my prediction for each game, I assumed two scenarios for the particular game.  First I assumed that the home team in the given contest would do to the opponent team’s overall statistical averages what it has done to the rest of its home opponents statistical averages.  I then calculated the score produced under that scenario.  Next I assumed the opposite, namely, that the opponent would do to the Home team’s home statistical averages what it has done to the rest of its opponent’s averages.  I then calculated the score produced under that scenario.   Then for each scenario I used Win Score to estimate each team’s probability of victory under each scenario.

So for instance in the Duke-NC State game, if Duke does to NC State’s averages what it has done to the averages of the rest of the teams that have visited Cameron Indoor Stadium this season, then NC State will get killed, 86-70.  NC State would finish the game with a box score that would produce a WS of 27.0 and a DWS of 48.5.  Under such conditions, NC State would have a probability of winning only 13.7% of the time.  If, however, NC State does to Duke‘s home averages what it normally does to other opponent’s overall averages, then Duke should still prevail, 80-74.  Under that scenario, Duke would finish with a box score that would equate into a WS of 37.5 and a DWS of 32.0.  Under those conditions, Duke would win 59.6% of the time.  Thus, the average score I produced was: Duke 83 NC State 72.  Under the first scenario, Duke’s win probability would be 86.3% and under the second scenario Duke’s win probability would be 59.6%, for a combined rough win probability average of 73%.

For those who gamble, the line on the Badger-Spartan game closed at Spartans -6.0 (the opening line was -5.5), which is right about where I’m at, and the final line on the Duke game was Duke -10.5 (the opening line was -10), which is, again, right where I’m at.

Let’s see how this works.  I hope I am way off on the first prediction.


Gracias, Los Lobos

February 16, 2012

Normally when I stick my neck out, I end up getting a Robespierre treatment.  Last night I did my first NCAA Tournament post, and in it I rated the New Mexico Lobos the 4th strongest team in the likely tournament field.  I pointed out in the post that the ranking was out-of-step with other analytical power rankings.  However, I did not realize New Mexico, with a 20-4 record, was not even ranked at all by the Associated Press.  That’s stunning.

Anyway, last night los Lobos took on #13 San Diego State on the road.  It was the perfect set-up for a severe undermining of my Ty Rating system.  Didn’t happen.

Gracias, Los Lobos.

And thanks to the CBS Sports Network, a cable channel, last night I was able to actually watch the team I rated the 4th strongest in the country.  They are certainly tough.  They get on the boards and they play strong defense.  Whether they are the fourth best team in the nation, I don’t know.  But they are very, very good, and they are certainly a team to keep your eyes on.

Power Ranking the likely 2012 NCAA Basketball Tournament Field by “Ty Rating”

February 15, 2012

I’m getting a head start on handicapping the likely 2012 NCAA Basketball Tournament.  I have taken all of the teams mentioned in the various “bracketology” sites, minus the low seed automatics, and I have power ranked the top 60 teams using something I call the “Ty Rating”.

Ty Rating is simply each team’s expected Winning Percentage (derived from the difference between the team’s Win Score average and its Defensive Win Score) subtracted from the expected Winning Percentage the rest of the country would have against the very same schedule.  In other words, it first evaluates each team’s performance, and then adjusts it for the strength of the schedule the team faced.  All of the calculations are based upon numbers I found at this nifty gambling site called “StatFox Sports” (Sidenote:  While I love the site, if they are not affiliated with Fox Sports, or with the old site StatFox, they are creeping very close to two trademark violations).

StatFox Sports makes the Ty Rating possible because it not only lists each team’s “Team” and “Opponent” statistics, it also lists the averages yielded and produced by those opponents.  By doing so, it allows me to precisely adjust each team’s success according to the strength of its schedule.  SOS adjustment is an absolute must when it comes to college sports analysis because of the widely different competition faced by the different schools.  Before now, I would have had to calculate each school’s opponent strength manually.  That’s way too much work.  With StatFox Sports its all done for me.  That’s why I’ve been looking for a site like StatFox Sports for quite a while.  I basically stumbled on this beauty, and now I’m back in the college basketball business, big time.

“Ty Rating” Calculation Example using #23 Virginia Cavaliers

Tony Bennett’s Virginia Cavaliers have a team Win Score Average of 35.1 — not that great, just above the BCS average (based on my opponent strength calculations, I peg the upper Division I Win Score average at 31.98, and the Defensive Win Score average at 28.01, the difference is borne by the 200 or so lower Division I schools the team’s feast on).  However the Cavaliers Defensive Win Score average is a phenomenal 12.6, way way above average.  You subtract the difference and divide by ten and you get a Team Marginal Win Score of +2.25, which translates into an expected Winning Percentage of about 0.884, or about 20.3 wins in their 23 games played.  Their actual record is 19-4, so MWS estimates extremely well.  But, that does not give a necessarily accurate portrait of Virginia’s relative strength as a basketball team, because they could have been playing the Washington Generals every night for all we know.

So to adjust my power rating of Virginia to account for the strength of the opponent’s Virginia has faced, I take the collective Win Score average produced by Virginia’s opponents’ opponents, and that happens to be 28.6, pretty high.  Then I calculate the collective Defensive Win Score average yielded by Virginia’s opponents’ opponents and that happens to be 27.9.  If you put those two numbers together, you get an Opponent’s Opponent MWS of 0.07, which means Virginia has played a relatively weak schedule, because the rest of the country would be expected to play 0.514% basketball — or winning basketball — against the same schedule.  For comparison, the NCAA Tournament field Opponent’s Opponent expected winning percentage average is 0.435%.

So, while Virginia has an impressive raw Marginal Win Score and winning percentage of 0.884%, when you adjust for their weak schedule, by subtracting the generic opponent expected Winning Percentage of 0.514%, you get a more modest “Ty Rating” for Virginia of 0.369, which is just above the field average “Ty Rating” of 0.354.  Thus the Ty Rating levels the field and provides an opponent neutral evaluation of each team’s relative strength as we enter “Bracket Season”.

How to read the Chart

The Chart below features a ranking of the 60 most likely qualifiers and bubble teams for this season’s NCAA Tournament as presented by ESPN’s Bracketologists.  The ranking is based on each team’s Ty Rating, as explained above.  The first column marked “WS” is the team’s Win Score average.  Win Score is an efficiency score based on a weighting of box score statistics based according to how each statistic correlates with winning.  The column marked “DWS” is each team’s Opponents Win Score average.  The third column is the expected Winning Percentage for a team with a Win Score/Defensive Win Score differential equal to the one posted by the given team.  The fourth column, marked “SOS” for strength of schedule, is the very same evaluation, except done on the Opponent’s opponents.  In other words, it is the expected winning percentage the rest of the country would post against the very same schedule of opponents.  Finally, there is the “Ty Rating” which is an expression of each team’s relative strength by comparing the difference between the expected winning percentage each team has achieved against the expected winning percentage the rest of the country has achieved.

I have analysis of the field below that.

TEAM WS DWS exW% SOS Ty Rating
1 Kentucky 48.2 12.7 1.105 0.423 0.682
2 Ohio St 41.2 12.6 1.005 0.375 0.631
3 Mich St 42.2 14.1 0.979 0.392 0.587
4 New Mex 44.1 13.2 1.027 0.455 0.572
5 Syracuse 46.1 19.8 0.949 0.383 0.565
6 Missouri 46.5 18.3 0.981 0.443 0.538
7 Kansas 40.6 18.5 0.877 0.345 0.532
8 UNC 46.1 20.5 0.937 0.406 0.531
9 Wisconsin 36.6 11.8 0.923 0.404 0.519
10 UNLV 47.5 18.8 0.989 0.482 0.507
11 Indiana 41.5 21.7 0.838 0.349 0.444
12 Duke 39.5 23.2 0.779 0.343 0.436
13 Baylor 41.8 20.3 0.867 0.436 0.431
14 Texas 35.4 19.2 0.777 0.361 0.416
15 Uconn 38.2 22.6 0.767 0.355 0.412
16 Witch St 43.9 18.4 0.935 0.524 0.411
17 Gonzaga 39.7 20.3 0.832 0.429 0.402
18 Florida 44.2 26.9 0.796 0.407 0.388
19 Louisville  36.5 19.3 0.794 0.407 0.386
20 California 36.9 17.9 0.825 0.441 0.383
21 Flor St 34.5 17.9 0.784 0.409 0.375
22 St Marys 43.8 19.9 0.908 0.535 0.373
23 Virginia 35.1 12.6 0.884 0.514 0.369
24 Creighton 45.1 25.8 0.829 0.465 0.365
25 Memphis 39.1 24.1 0.757 0.392 0.365
26 Arizona 35.2 19.7 0.765 0.407 0.358
27 Kan St 30.9 15.7 0.759 0.406 0.353
28 Marquette 39.4 23.1 0.779 0.429 0.349
29 St Louis 34.3 16.5 0.804 0.455 0.349
30 Iowa St 38.8 25.1 0.735 0.387 0.348
31 NC State 40.4 26.3 0.742 0.391 0.351
32 BYU 43.7 21.9 0.872 0.529 0.342
33 Miss State 39.4 26.6 0.719 0.384 0.335
34 W Virg 36.1 23.8 0.711 0.384 0.327
35 Gtown 37.1 16.8 0.762 0.438 0.324
36 Vanderbilt 37.3 25.7 0.699 0.377 0.322
37 Alabama 32.6 19.8 0.719 0.404 0.315
38 Wyoming 31.1 13.3 0.804 0.501 0.303
39 Lng Be St 37.5 21.9 0.767 0.465 0.302
40 Wash 35.1 24.3 0.686 0.392 0.294
41 Midd Tenn 35.8 17.2 0.818 0.526 0.292
42 Akron 35.8 20.8 0.757 0.479 0.278
43 San D St 35.5 19.2 0.779 0.507 0.271
44 Minnesota 35.9 23.1 0.719 0.449 0.269
45 Xavier 34.3 24.2 0.674 0.409 0.265
46 Michigan 31.8 23.4 0.635 0.367 0.268
47 Ntr Dame 34.7 25.7 0.655 0.391 0.264
48 Harvard 34.8 15.5 0.829 0.569 0.261
49 Murray St 38.4 18.2 0.845 0.611 0.234
50 Oregon 31.3 24.3 0.621 0.389 0.232
51 Miami 35.1 26.9 0.641 0.409 0.232
52 Purdue 32.5 26.1 0.613 0.383 0.229
53 Belmont 44.1 25.76 0.813 0.585 0.228
54 Oral Rbts 34.3 23.4 0.687 0.472 0.215
55 Nthwstern 36.5 32.4 0.572 0.363 0.209
56 Cinn 35.1 24.1 0.689 0.485 0.203
57 Seton Hall 32.7 25.7 0.621 0.424 0.196
58 Temple 38.1 27.2 0.595 0.414 0.181
59 South Miss 32.6 22.1 0.681 0.513 0.168
60 Drexel 30.2 17.9 0.711 0.545 0.166
AVERAGE 38.1 20.9 0.791 0.435 0.354

Kentucky and Ohio State are this season’s War Machines

If you are looking for the favorites in this year’s NCAA field, it has to be Kentucky and Ohio State.  First of all, Defensive Win Score, combined with a decent +40 Win Score, is usually the mark of a champion.  Both Kentucky and OSU have those qualities, and they are the only two teams in the entire expected field that have Ty Ratings above 0.600.  They have to be the prohibitive favorites.  Look at Kentucky’s expected Winning Percentage — the team should not have lost a single game!  (An expected winning percentage above 1.000% is a function of the uneven distribution of statistics).

Last Two Champions “Ty Ratings”

With all of that said about how strong Kentucky and OSU are, last season’s champion, UConn, had a Ty Rating of only 0.303, which would have been good for #38 in this season’s initial poll.  Two years ago, the champion, Duke, had a Ty Rating of 0.509, which would be good for #9 in this season’s initial poll.

Underrated and Overrated

No matter what kind of analytical Power System you use, you will always have a head scratcher.  This season’s is New Mexico, a team that is #4 in the initial Ty Ratings, ahead of UNC, Kansas, Missouri and Duke.  New Mexico is not as highly rated by others as they are by me, but those are the breaks.  I have to maintain the integrity of the system.  New Mexico is my early sleeper.  Another two teams who may be underrated are the battling Wisconsin Badgers, and the UNLV Running Rebels (who lost last night).  Others in the list of underrated would be the Big Ten’s Michigan State and Indiana, each of whom grade out better than the respect they are currently being afforded by national polls.

The overrated seem to live in the ACC.  Virginia, as I mentioned, has not played a strong schedule, but they certainly play winning defense.  UNC is not as strong as reputed.  Neither is Duke.

One team that is vastly overrated is Murray State.  Murray State has a gaudy record, but they have been very lucky, and they have not played a very strong schedule at all.  They could be an overseed that you would look to eliminate early in your bracket.

My ratings do not like the Georgetown Hoyas, either.  But alot of of others, including Ken Pom, have them much higher rated


Let’s look at Joe Lunardi’s  “Last 4 In” (Minnesota, NC State Cincinnati, and Miami) compared to some of his “Last Outs” (Xavier, Washington, Belmont, Wyoming, Oregon, and Northwestern).  Of those ten teams, which do the Ty Ratings favor?

The Ty Ratings favor in reverse order: Minnesota (0.269); Washington (0.294); Wyoming (0.303); and NC State (o.351).  Obviously, the Ty Ratings disagree with Lunardi heavily on the worthiness of Cincinnati (0.203) and Miami (0.232).  It also sees NC State as more of a lock, and Wyoming as a deserving of much more respect (Lunardi has them in his “Second Four Out” — Ty Ratings have them all the way in).

More to Come

I will be keeping up the Ty Ratings on a separate page of this blog, and commenting on them all the way up to bracket picking time.  Stay tuned.  I will also be analyzing, retroactively, how the Ty Ratings would have fared in past tournaments.  Stay tuned.