College Basketball Preseason Ratings For 2013-14

posted in NCAA Basketball

NCAA hoops starts in 4 days, which means we’re in the final stages of setting up our college basketball section for the upcoming season. Rosters have been loaded on our test server, and we’re making some final checks and tweaks before releasing everything to the wild. So, we won’t have official record projections and conference standings predictions posted until tonight or tomorrow, but we’ve got our preseason ratings prepped and ready to share.

These ratings are completely data-driven, with no manual fudges (except for the four new Division I teams, for whom we estimated ratings based on research into past results and current rosters). As with last year, the main inputs to our system are past team ratings, current rosters, player stats from the past few seasons, and recruiting info.

How We Create The Preseason Ratings

The basic idea is that we establish a baseline prediction for a team, given their power ratings from recent years, and assuming an average amount of roster turnover. Then we make some adjustments based on how much value each team is returning on offense and defense (see our post on returning production from two years ago for a brief recap of how we calculate that), as well as the strength of their recruiting classes from the past few years, and the value of any transfers they’ve added this season.

We also use NBA draft info, as losing a high pick seems to have a negative effect beyond what we can capture in our returning value stat.

Below is our full ranking from 1 to 351, plus the key factors driving each team’s rating. Here’s a guide to what each column shows:

  • Rank — This year’s projected end-of-season ranking. Teams with a lot of roster turnover may start out slower than those returning a lot of players, but this is where we are projecting teams to end in our final ratings.
  • Team — Self explanatory.
  • Conf — Conference affiliation.
  • Rating — Projected end-of-season rating for this year. This is expressed in “points above average”, so it’s the expected margin of victory against an average team on a neutral court. Note that the actual final rankings will probably have a wider distribution than our preseason ratings. That’s to be expected, as these are the *average* projections for each team. Some good teams will do better than expected, and some bad teams will do worse, which will lead to a wider actual spread in the final ratings.
  • Last Yr Rank — This is a team’s rank in a special version of our new predictive ratings that doesn’t take last year’s preseason ratings into account. We use this version so that our preseason ratings from previous years don’t influence the ratings for this year.
  • Baseline — This is simple predicted rating for this season that is based only on the team’s ratings from the past few seasons, and doesn’t take the actual current roster into account. We show this to help you get an idea of how important the returning value numbers and the recruiting info is.
  • Ret. Off% — A single number that tries to capture how much of a team’s offensive value is returning. Every player’s value is calculated based on their offensive rating, usage rate, and minutes played. Then we use this year’s roster to calculate what percentage of last year’s value is returning.
  • Ret. Def% — Same as above, but using a defensive rating rather than offensive.
  • Rec + Trans — A single number that summarizes the projected value of a last few recruiting classes and this year’s Division I transfers. Juco transfers are ignored. Most of the value here comes from this season’s recruiting class, but there is still a bit of value in having good classes two or three years ago, presumably because those highly-ranked players are more likely to improve this season than another non-elite recruit is. [Eagle-eyed readers may note that this year’s Kentucky team has a lower bonus than last season. That’s not because this year’s class is worse; it’s because our analysis this year suggests our recruiting bonuses last season were too large.]

The main purpose of these ratings is to drive our college basketball projection pages [link to come once this season’s projections are up]. Before the season starts, those will be 100% driven by these ratings. As the year progresses, the preseason ratings will have less and less of an impact, until they are nearly irrelevant at the end of the season.

Of course, they’re also fun to talk about. If you’ve got questions or comments on these ratings, fire away in the comments below the post!

TeamRankings 2013 NCAA Basketball Preseason Ratings

UPDATE (11/5/2013 1:20 PM ET): Per comments from @RamBBallStatMan@Hoops_Nerd, @ShockerHoops, and others, we discovered a bug in the way we were handling transfer data. Correcting that bug in the historical data and retraining the model led to changes in the importance of certain variables. In particular, the impact of transfers and recruits has increased, presumably because the data is cleaner than before. In addition, the transfer lists for various teams have changed, leading them to rise or fall in the rankings. This final version has some significant changes from the initial rankings we posted, so we apologize for any confusion or inconvenience that may have caused.

There will be no more major updates to these rankings, as we are loading them into the database this morning in order to create the conference and full season record projections, which will be posted later today.

RankTeamConfRatingLast Yr RankBaselineRet. OffRet. DefRec + Trans
4KansasBig 1219.7716.414%25%9.7
6Michigan StBig Ten18.01113.881%84%0.3
7Ohio StateBig Ten17.4916.063%76%0.6
8WisconsinBig Ten17.41214.065%54%3.5
10Oklahoma StBig 1215.8229.896%85%0.8
12N CarolinaACC15.12311.658%73%1.7
13BaylorBig 1215.02410.865%61%4.6
16CreightonBig East14.71611.085%76%0.0
18IndianaBig Ten14.7216.222%30%4.9
19MinnesotaBig Ten14.21711.459%47%3.2
22GeorgetownBig East13.51911.661%80%0.6
23MichiganBig Ten13.5614.155%63%1.2
24IowaBig Ten13.1348.288%88%0.2
26VillanovaBig East12.8577.873%73%2.3
28Notre DameACC12.7399.769%72%0.8
29Saint LouisA-1012.6299.373%67%1.6
30IllinoisBig Ten12.5409.232%41%6.6
34New MexicoMWC11.73210.274%66%0.1
35MarquetteBig East11.73010.663%59%1.2
37St MarysWCC11.42110.759%61%1.0
38ProvidenceBig East11.0695.392%86%1.8
40Boise StateMWC10.7545.992%89%0.0
42Wichita StMVC10.51810.951%39%1.6
44XavierBig East10.3796.450%59%3.4
45PurdueBig Ten10.3777.561%57%1.9
47St JohnsBig East10.2993.991%88%2.1
51North Dakota StateSummit9.7734.098%96%0.5
52Boston ColACC9.11082.898%96%1.2
53Iowa StateBig 128.9269.735%38%2.4
55GA TechACC8.8933.982%74%2.4
57Kansas StBig 128.72811.045%51%0.0
58La SalleA-108.7486.579%79%0.0
64Arizona StPac-127.9764.755%51%3.5
65ButlerBig East7.9497.954%69%0.0
66S MississippiCUSA7.8586.461%53%2.0
67Wright StateHorizon7.81251.6100%92%1.3
68LA TechCUSA7.2813.090%85%0.2
69Utah StateMWC7.01134.088%75%0.0
70San Diego StMWC6.8389.729%44%0.5
73Seton HallBig East6.51074.357%77%0.3
74NC StateACC6.4339.025%19%2.9
75Geo MasonA-106.31371.891%92%0.2
76U MassA-106.3862.965%75%2.1
77Indiana StMVC6.3942.877%87%0.0
78Miami (FL)ACC6.21411.511%15%1.7
79Florida StACC6.11115.060%69%0.5
81Weber StateBig Sky5.9744.463%71%0.0
82S MethodistAmerican5.8177-1.195%86%2.9
84WI-Grn BayHorizon5.41330.771%72%2.7
85Stony BrookAm. East5.3673.963%65%0.0
86St JosephsA-105.3754.460%69%0.1
87W VirginiaBig 125.31174.963%45%1.6
89TX El PasoCUSA5.21123.463%67%0.6
91N Mex StateWAC5.1893.475%72%0.0
92Middle TennCUSA5.0417.539%45%0.0
93N IowaMVC4.9715.753%52%0.1
94Colorado StMWC4.9278.716%26%2.1
96Wake ForestACC4.61360.853%84%2.1
98Oregon StPac-124.51052.869%73%0.0
99VermontAm. East4.31390.791%88%0.1
100Central FLAmerican4.01261.777%74%0.0
101San FranscoWCC4.01380.584%85%0.5
102Fla Gulf CstA-Sun3.91100.963%68%1.9
103Texas A&MSEC3.9904.846%59%0.2
104Arkansas StSun Belt3.9144-0.350%58%4.0
105S FloridaAmerican3.81242.565%62%0.7
106TexasBig 123.71046.116%54%0.3
108Geo WshgtnA-103.61092.061%74%0.4
109Wash StatePac-123.5914.056%67%0.0
110Georgia StSun Belt3.4189-1.689%82%1.6
113Kent StateMAC3.11271.950%51%2.9
114NorthwesternBig Ten3.11203.441%48%0.4
115Boston UPatriot3.11450.083%81%0.7
116OklahomaBig 123.0476.715%43%0.4
117Missouri StMVC2.9206-0.853%70%2.7
119E KentuckyOVC2.91210.081%85%0.1
125DePaulBig East2.4155-0.675%61%1.2
128Penn StateBig Ten2.21491.337%50%1.4
130VA TechACC2.11572.044%73%1.5
131Rhode IslandA-102.0187-0.854%55%2.4
134NebraskaBig Ten1.71232.439%57%0.7
135S Dakota StSummit1.6972.948%62%0.1
136UCSBBig West1.5217-2.292%86%0.0
140Fresno StMWC1.31181.374%44%0.5
142Col CharlestnCAA1.21530.464%71%0.1
144Miss StateSEC1.1245-2.475%91%0.0
145Cal PolyBig West1.0160-1.267%69%0.6
147Cleveland StHorizon1.0262-3.482%77%1.9
148Santa ClaraWCC1.0822.242%46%1.4
149W KentuckySun Belt0.9175-1.367%74%0.5
152South Carolina UpstateA-Sun0.7183-2.788%87%0.0
153Oral RobertsSouthland0.61500.630%47%2.7
156Wm & MaryCAA0.5208-3.395%81%0.0
158LA LafayetteSun Belt0.4225-4.183%87%0.9
160MontanaBig Sky0.31560.759%60%0.0
163Ste F AustinSouthland-0.1852.442%43%0.0
165NW StateSouthland-0.1135-0.956%67%0.4
166Loyola MymtWCC-0.2212-2.799%75%0.7
167Rob MorrisNEC-0.21420.363%56%0.0
168St BonaventA-10-0.21162.331%51%0.0
169UC DavisBig West-0.2205-4.491%75%1.1
171W MichiganMAC-0.3146-0.150%61%0.0
172U PennIvy-0.4250-5.2100%100%0.0
174Cal St NrdgeBig West-0.5209-4.189%73%1.1
175W CarolinaSouthern-0.6238-4.387%89%0.0
180S CarolinaSEC-0.9200-1.159%48%0.7
181High PointBig South-0.9219-5.283%77%1.2
182Charl SouthBig South-0.9185-2.669%66%0.0
185HawaiiBig West-1.1202-2.870%54%2.2
186UC IrvineBig West-1.2141-0.958%55%0.3
187Murray StOVC-1.31322.416%27%1.7
189Youngs StHorizon-1.5190-2.569%65%0.0
191San DiegoWCC-1.5179-2.664%58%0.9
192Morgan StMEAC-1.5207-3.568%76%0.0
193Illinois StMVC-1.6565.55%11%0.0
194N ColoradoBig Sky-1.6253-4.381%79%0.6
195Morehead StOVC-1.6223-2.654%44%2.5
196NC CentralMEAC-1.7158-2.039%49%1.8
199TX-ArlingtonSun Belt-1.8148-0.746%46%0.4
200Sam Hous StSouthland-1.8234-4.084%72%0.4
201Texas TechBig 12-1.9237-3.2100%68%0.0
202HartfordAm. East-2.0228-5.986%88%0.0
205S IllinoisMVC-2.1171-1.238%44%0.0
206Norfolk StMEAC-2.2216-4.688%72%0.2
208Lg Beach StBig West-2.31700.423%38%0.0
209S AlabamaSun Belt-2.7204-3.678%56%0.1
210AR Lit RockSun Belt-2.7224-4.172%74%0.0
211E CarolinaCUSA-2.71280.627%27%0.9
212Central ConnNEC-2.8257-5.774%83%0.0
213NC-AshevilleBig South-2.8211-3.128%59%1.9
215Air ForceMWC-2.91011.815%31%0.0
218North DakotaBig Sky-2.9264-7.396%73%1.7
219Miami (OH)MAC-3.0249-4.235%69%1.0
221Florida IntlCUSA-3.1182-3.257%45%1.7
222Holy CrossPatriot-3.2229-4.664%62%0.0
223LibertyBig South-3.3279-7.488%81%1.9
224Old DominionCUSA-3.4274-3.364%63%0.0
225AlbanyAm. East-3.4152-1.641%51%0.0
226Mt St MarysNEC-3.5222-4.667%67%0.0
227Gard-WebbBig South-3.5201-3.942%62%0.9
228CS FullertonBig West-3.7184-2.126%41%2.2
229App StateSouthern-3.8268-5.566%65%0.5
233St Fran (NY)NEC-4.0221-4.664%69%0.0
234TX SouthernSWAC-4.1164-2.540%38%0.8
235TX ChristianBig 12-4.2251-4.257%44%1.2
236CS BakersfldWAC-4.2239-5.939%43%3.7
237E MichiganMAC-4.2254-5.737%69%0.7
238SE MissouriOVC-4.2226-5.551%65%1.1
241TN StateOVC-4.6176-2.234%43%0.0
244Jksnville StOVC-4.7210-4.238%43%1.1
246South DakotaSummit-4.7244-5.865%75%0.2
247James MadCAA-4.8186-1.936%42%0.0
248Coastal CarBig South-4.9243-3.827%62%0.5
249Texas StateSun Belt-4.9261-6.168%62%0.2
250WinthropBig South-4.9270-6.372%67%0.0
255GA SouthernSouthern-5.1278-8.041%62%3.1
256Ball StateMAC-5.1258-5.259%72%0.0
257Bowling GrnMAC-5.1213-3.250%55%0.0
258Austin PeayOVC-5.2307-7.183%76%0.6
261North TexasCUSA-5.4260-4.748%33%1.1
262W IllinoisSummit-5.4167-2.428%36%-0.1
264RadfordBig South-5.4298-9.098%88%0.0
265Central MichMAC-5.5246-5.355%73%0.0
266E Tenn StA-Sun-5.6317-6.770%70%0.0
269TroySun Belt-5.7280-6.848%60%0.9
271N FloridaA-Sun-5.9242-5.152%39%1.0
276Savannah StMEAC-6.4218-4.50%36%1.4
277Montana StBig Sky-6.4300-7.864%77%0.0
278McNeese StSouthland-6.5305-8.063%78%0.0
280St PetersMAAC-6.7273-6.264%45%1.2
282Fla AtlanticCUSA-6.8241-4.211%48%0.3
283TX-San AntCUSA-6.9247-4.634%54%-0.2
284TX A&M-CCSouthland-7.0308-8.675%66%0.8
285Sacred HrtNEC-7.2275-7.154%50%1.2
286Prairie ViewSWAC-7.2324-11.4100%100%0.0
287Portland StBig Sky-7.3297-7.337%67%0.4
288Nicholls StSouthland-7.3285-8.143%70%0.0
289CampbellBig South-7.6295-7.244%67%0.2
290N HampshireAm. East-7.6256-6.371%49%0.5
291Northern KentuckyA-Sun-7.6248-6.028%46%0.0
292TN TechOVC-7.7281-6.856%62%0.1
293E WashingtnBig Sky-7.8293-7.770%70%0.0
294E IllinoisOVC-7.8277-7.461%51%0.2
295UC RiversideBig West-7.9312-8.3100%62%0.1
296Nebraska OmahaSummit-7.9326-12.287%100%0.7
297LA MonroeSun Belt-8.1328-11.138%84%0.0
298NC A&TMEAC-8.1231-6.154%34%0.0
300Grand CanyonWAC-8.5n/an/an/an/an/a
301Chicago StWAC-8.5302-11.4100%82%0.7
303Utah Val StWAC-8.7289-8.265%46%0.9
304MaineAm. East-8.7266-6.025%35%0.0
305SE LouisianaSouthland-9.0283-7.337%44%0.1
306Coppin StateMEAC-9.0309-9.1100%71%0.0
307N ArizonaBig Sky-9.1314-8.460%48%0.8
308Sac StateBig Sky-9.1276-7.856%46%-0.1
310VA MilitaryBig South-9.2301-8.866%47%0.7
311St Fran (PA)NEC-9.3329-11.376%88%0.0
312Ark Pine BlSWAC-9.5296-9.357%59%0.0
313Delaware StMEAC-9.6287-7.952%45%0.0
316Florida A&MMEAC-9.8327-11.980%80%0.0
317San Jose StMWC-9.8288-6.954%36%0.0
318TN MartinOVC-9.8333-12.686%81%0.1
320Idaho StateBig Sky-10.0318-9.693%63%0.0
322Kennesaw StA-Sun-10.0330-11.215%61%1.0
323N IllinoisMAC-10.3323-10.1100%69%0.5
324Maryland BCAm. East-10.6315-11.07%62%0.0
325New Jersey TechInd.-10.6267-8.029%46%0.0
327TX-Pan AmWAC-11.2316-10.526%48%1.4
328Central ArkSouthland-11.3310-10.427%49%0.3
329Incarnate WordSouthland-11.5n/an/an/an/an/a
331SIU EdwardOVC-11.7306-10.762%49%0.1
332S UtahBig Sky-11.8313-9.018%46%0.0
333S Car StateMEAC-11.8338-13.376%80%0.0
334Alcorn StateSWAC-11.9321-12.04%45%1.1
336Massachusetts LowellAm. East-12.5n/an/an/an/an/a
337Jackson StSWAC-12.9322-10.412%21%1.8
338BinghamtonAm. East-13.1341-13.350%51%0.0
339Houston BapSouthland-13.4331-12.766%54%0.2
340Miss Val StSWAC-13.8342-12.90%47%0.0
341Alab A&MSWAC-13.8340-13.857%59%0.0
342PresbyterianBig South-13.9334-11.632%44%0.0
344Abilene ChristianSouthland-14.5n/an/an/an/an/a
345New OrleansSouthland-14.7346-14.626%49%0.8
346LongwoodBig South-14.8344-14.382%39%0.2
347Maryland ESMEAC-15.2343-14.3100%57%0.3
348Alabama StSWAC-15.4335-12.413%35%0.0
350F DickinsonNEC-15.8339-13.226%39%0.1
351Grambling StSWAC-21.9347-22.670%70%0.0


  • RJT

    You gave VCU 0.0 value for their recruits + sophomore upgrades + Florida State 5th year transfer/monster PF Terrance Shannon. Sorry, can’t buy into the formula.

  • David Hess

    Well, the recruit thing is simply because I go by the RSCI consensus rankings, and they have VCU with no high ranking recruits.

    But it looks like you are right about Shannon. He shouldn’t make a HUGE difference, due to to his poor offensive numbers. But it does look like I had a bug in the transfer calculations. I’me re-running the numbers now, and will re-post shortly.

  • RJT

    David, thanks for the response. You are also not taking into account that Jordan Burgess, a consensus 4 star recruit from a year ago, is starting his freshman year after redshirting last year. I’m not trying to pull the “we’re so awesome” stuff, just simply pointing out that the newcomer/transfer formula might have a kink in it, because VCU has six newcomers, three of whom are going to make an immediate impact.

  • David Hess

    Yeah, this model definitely wouldn’t be able to take into account Burgess. But based on the historical data, even RSCI top 100 recruits from 2 years ago don’t make much *consistent* difference, so a player outside the top 100 wouldn’t end up affecting the projected ratng much, even if we could take him into account. However, the program baseline depends on past ratings, and if VCU has a history of recruiting similarly talented players, that should show up a bit via the past ratings.

    If you haven’t checked out Dan Hanner’s preseason ratings, I think you might like them. He does a player-based simulation, so his system would account for stuff like VCU’s six newcomers. They’re on ESPN Insider:

  • RJT

    Thanks for the explanation. It’s difficult if not impossible to incorporate qualitative analysis into a quantitative based formula. They both have their pluses and minuses, but overall I don’t really have much of a problem with these pre-season rankings. Cheers!

  • Guest

    Great stuff guys – thanks. Any chance we see projected win numbers too?

  • David Hess

    Those should be out tomorrow evening if everything goes according to plan (or Wednesday if not).

  • JB

    With the new rule changes this year aimed towards benefitting more scorig, a lot of physical teams like Pitt, Big Ten etc. may have to alter their style of play. It will also be interesting to see how it affects totals. Any thoughts TR on how your betting picks models will adjust?

  • David Hess

    It’s impossible to say for certain until we actually get some games under the models’ belts. However, when we took a look this preseason at what variables were most important, a lot of them concerned statistics related to luck, and un-reproducible events. Essentially, a lot of the value in the models, in the past, has been via them figuring out reasons for a team to be over/under-rated by the public. For the most part, I don’t think those should change as a result of the new rules.

    We’ll definitely be monitoring the totals picks very closely. If we see a larger fraction of Under picks than usual, and they’re doing particularly poorly, we’ll reevaluate the situation.

  • Owen

    Really interesting results. As a Michigan fan, I might be a little worried! Also, it’s impressive how high up Harvard is on the rankings. A few metrics I might think might be interesting to see are: Times scored (field goal or free throw) per possession. Also, as I’m watching the Virginia-VCU game, I can tell that Virginia clearly controls the tempo of the game, considering they are a slower-paced, defensive-minded team, whereas VCU is fast-paced and plays very aggressive, etc. Is there a statistic on TR (maybe I haven’t found it) that predicts how much one team controls the tempo or type of game they play? I have no idea how it would work, but I think it’s be really cool to see the results. I’m guessing possessions and scoring chances might be factors in this.

  • David Hess

    “Times scored (field goal or free throw) per possession. ”

    Floor% is an *estimate* of this. It’s not directly taken from play by play info, but it’s a guess based on the box scores, taking into account the number of field goals, free throws, and missed free throws for each team:

    “Is there a statistic on TR (maybe I haven’t found it) that predicts how much one team controls the tempo or type of game they play”

    Pomeroy’s adjusted tempo tries to measure that. A team that always plays to whatever tempo their opponent dictates should be exactly average in Adjusted Tempo, while a team that dictates a fast or slow tempo will have a low or high Adjusted Tempo. It’s still a bit early for the stat to mean anything, but later in the year it will be interesting:

    We have raw tempo on TR, but it’s not adjusted for opponent pace:

  • Jack

    do you update these ranking thru the season?

  • David Hess

    The preseason rankings? No. However, these are the starting points for our priored predictive ratings, which update daily:

  • Wayne McGlothlin

    Help me out. My team wins, by more than the spread. They drop .02 Kansas wins tonight, by less than the spread but they rise by .03 How does that happen? Few teams played tonight, so it can’t be that effect.

  • David Hess

    I’m not sure I understand your question — the teams are rising 0.02 and 0.03 in what number? Rating? Projected wins? (Neither of those are specified to 2 decimal places, so doesn’t seem like that could be it.) Something else entirely? If you link to the page where you’re seeing the changes, I should be able to better explain them.

    At any rate, here are some general ideas:

    1) The spread is not a factor in the ratings or the season projections, so performance vs. the spread is not very relevant here. Performance relative to the power ratings prediction is, and sometimes those will agree with the spread, but not always.

    2) The more games a team plays, the less weight their preseason rating gets, so if a team has a high preseason rating, they could drop even with a pretty good performance.

    3) A difference of 0.05 in team ratings (if that’s what you’re asking about) is pretty close to irrelevant. That would mean one team is favored by an extra 0.05 points on a neutral court according to our ratings. So 1 extra free throw every 20 games.

    4) In our New Ratings (the ones that use these preseason ratings), Kansas was favored by about 13 last night. They won by 8. Their rating dropped by 0.1 points: … That seems perfectly reasonable. I don’t know what your team is, but you ought to find something similar if you look them up.

    Hope this helps a bit. If you’re still confused, please leave a link to the page in question, and I can give you a more detailed answer.

  • Wayne McGlothlin

    To be more specific: Saturday Wichita State is favored by 14 at Drake. They win by 18, but Sunday’s rating moved from .651 to .649. KU was favored over OU by double digits. They win by eight. Their rating goes from .688 to .691. What determines the “spread”. Is it the betting line? Or, is the Team Rankings ratings between teams the determinant in the line that you report? Thx

  • Wayne McGlothlin

    Link that I’m looking at:

  • David Hess

    That’s the RPI, which is the formula the NCAA uses to group/judge teams during the NCAA Tournament selection process. Margin of victory isn’t used, just wins and losses. The formula is:

    25% your team’s winning%
    50% your team’s opponents’ collective winning%
    25% your team’s opponents’ opponents’ collective winning percentage.
    [there are also some bonuses based on whether your wins and losses come at home or on the road]

    So, beating a terrible team is bad for your RPI.

    We agree, it’s a terrible system. We only publish it because the NCAA uses it.