Archive for the ‘stats analysis’ Category

A couple months ago, KenPom added something called “defensive fingerprint” on each team page.  Mr. Pomeroy’s explanation:

Defensive Fingerprint attempts to objectively identify the style of a team’s defense. Inputs into the system are the departure from the D-1 norm of the following defensive characteristics…

– assist percentage (triple weight, higher means a more likely zone team)
– 3-point attempt percentage (triple weight, higher means a more likely zone team)
– free throw attempt percentage (double weight, higher means a more likely man team)
– turnover percentage (single weight, higher means a more likely man team)
– defensive rebounding percentage (variable weight depending on offensive rebounding percentage, higher means a more likely man team)

All those factors go into a super-secret formula that calculates whether, based on its stats, a given team is likely a zone team or a man-to-man team.

What’s interesting is that, despite the fact that MSU played man-to-man defense for all but a handful of possessions this past season, the formula spits out an “inconclusive” on our team page.  To investigate this phenomenon, I’ve put together a table showing MSU’s rankings for the five stats used in the formula.  The first set of numbers are the full season; these are the numbers that KenPom is using.

All Games/National Rank
MSU Value MSU Rank Indicates
Assist % 52.4 131 Neutral
3PA/FGA 35.8 271 Zone
FTA/FGA 36.5 178 Neutral
TO% 19.9 190 Neutral
Opp OReb% 27.3 11 Man-to-Man

You can see why the formula can’t identify us as a man-to-man team.  Our opponents shot a lot of 3-pointers, which makes us look like a zone defense.  But we ranked 11th nationally in defensive rebounding percentage, which implies that we play man-to-man defense.  For the remaining three factors, we’re very near the national averages, so the formula has nothing to go on.

One reading of these numbers is that Tom Izzo’s defensive scheme got the best of both worlds this season:

  • By placing in emphasis on preventing dribble penetration by hedging off shooters, the team forced a lot of perimeter shots from its opponents (the primary benefit of a zone defense).
  • But the fact the team was fundamentally playing man-to-man defense, particularly on the interior, meant the team didn’t sacrifice anything in terms of defensive rebounding (generally the main weakness of a zone defense).

Before declaring victory in the age-old quest to find the perfect defensive scheme, though, I think we meed to get a little more definition on those three middle-of-the-road formula factors.  To do so, I pulled the same numbers for conference games only (with ranks within the Big Ten).

Conference Games/Rank
MSU Value MSU Rank Indicates
Assist % 54.2 2 Man-to-Man
3PA/FGA 32.4 9 Zone
FTA/FGA 34.7 7 Man-to-Man
TO% 20.7 8 Zone
Opp OReb% 24.7 1 Man-to-Man

For assist percentage and free throw rate, we look more like a man-to-man team.  For turnovers, we look more like a zone team.

I’m not sure the low assist percentage is necessarily a major asset or weakness.  But the other two numbers are unfavorable.  Playing physical man-to-man defense resulted in a relatively higher number of fouls that created additional free throw opportunities for our opponents.  At the same time, the fact that our perimeter defenders were more focused on preventing penetration than with disrupting our opponents’ offensive rhythm meant we didn’t create a lot of turnovers.

Overall, then, we had one strength (defensive rebounding) and one weakness (fouling quite a bit) generally associated with man-to-man defense and one strength (forcing perimeter shots) and one weakness (not creating turnovers) generally associated with zone defense.

On net, the way this team played defense obviously worked pretty well, as they finished the season ranked 10th in the country in adjusted defensive efficiency.  Most of the numbers above are pretty consistent with the team’s numbers over the past several seasons, indicating that Tom Izzo’s approach to defense hasn’t changed much in recent years.  The biggest change from 2007-08 to 2008-09 was an increase in defensive rebounding percentage of roughly 4 percentage points.  As one might expect with a Tom Izzo-coached team, the key to success was rebounding.

P.S. You can probably sense I’m stalling for time by throwing a lot of numbers at you.  There’s been a bit of a delay in getting the new site launched.  It should be ready to go late this week or first thing next week.


Read Full Post »

From a four factors perspective, rebounding was clearly the dominant strength of this year’s MSU basketball team.  The team ranked 6th nationally in offensive rebounding percentage and 11th nationally in defensive rebounding percentage.  In none of the other six offensive/defensive four factor components did MSU rank higher than 50th nationally.

It seems clear that, given the fact MSU ranked in the top 20 nationally in adjusted efficiency on both sides of the ball, the combined level of rebounding the team sustained over the course of the season must have been among the nation’s best.  But just exactly how good was it?

Here’s a list of the top ten BCS conference teams in the country in rebounding percentage margin (offensive rebounding percentage minus defensive rebounding percentage; figures include all games):

Michigan St. 13.4
Pittsburgh 12.3
Washington 10.8
Kansas St. 10.2
Connecticut 9.8
West Virginia 9.2
Kansas 7.8
North Carolina 7.2
Tennessee 6.8
Louisiana St. 6.8

Only nine other major conference teams in the country had a margin equal to even half of MSU’s.  (And MSU’s margin of 13.4% still ranks #1, even if you include all Division 1 teams in the rankings.)

One thing that jumped out at me in looking at the rebounding data is that offensive and defensive rebounding percentages are not as highly correlated as I might have thought:

  • Of the top 25 teams in the country in defensive rebounding percentage, only 5 also ranked in the top 25 in offensive rebounding percentage.
  • Looking at just the top 10 defensive rebounding teams , only one team ranked higher than 150th in the nation in offensive rebounding percentage (Albany).
  • Looking at the top 10 offensive rebounding teams, only three teams ranked higher than 100th in the nation in offensive rebounding percentage (MSU, Pittsburgh, Washington).
  • The statistical correlation between team offensive and rebounding percentages is a relatively modest 12.1%.

It would appear that, for the majority of college basketball teams, excellence on the boards is a skill that can only be maximized on one end of the court or the other.  To some extent, that may be a function of strategy: teams with conservative approaches to the game secure defensive rebounds before they send players down court and eschew crashing the offensive glass so they can get back on defense.  It may also have something to do with personnel: defensive rebounding is more about size and position, while offensive rebounding is more about quickness and aggressiveness.

As a visual aid, here’s where MSU falls in a scatterplot of rebounding percentages for all 344 Division 1 basketball teams this season:


Because of the emphasis Tom Izzo has placed on rebounding during his tenure as head coach, I’d come to assume that top-notch offensive and defensive rebounding generally went hand in hand.  Last season, I was puzzled as to why MSU only ranked 104th nationally in defensive rebounding percentage when they ranked 8th in offensive rebounding percentage.

This season, the team put it all back together in terms of crashing the boards with abandon on both ends of the floor.  And our appreciation for that statistical combination should be even more heightened than perhaps it has been.

Read Full Post »

The scariest piece of UConn’s statistical profile is the number of shots they block.  The Huskies swatted away roughly one out of every six (17.2%) shots their opponents attempted from two-point range this season.

MSU, meanwhile, improved somewhat this season in avoiding getting their shots blocked, reducing their offensive block% from 9.8% to 8.4%.  But it’s still something of a team weakness.  Raymar Morgan, in particular, always seems to struggle getting his shot off close to the rim against taller defenders.

(Related: At this point, I really view Morgan as a role player.  He’s there to match-up with Stanley Robinson for at leat 15-20 minutes and hopefully grab some rebounds.  But any scoring he provides will be gravy.)

The table below shows our offensive block% (i.e., number of shots blocked divided by 2-point FG attempts) for each game this season, sorted from highest to lowest (worst to best).

at Purdue L 17.0
vs Maryland L 16.7
vs Robert Morris W 14.6
vs North Carolina L 14.3
vs Louisville W 13.9
vs Southern Cal W 13.2
vs Ohio State L 12.0
vs Oakland University W 11.9
at Minnesota W 10.9
at Iowa Hawkeyes W 9.3
vs Kansas W 9.1
at IPFW W 8.9
at Northwestern W 8.7
at Ohio State W 8.3
vs Minnesota W 6.7
vs Texas W 6.5
at Penn State W 4.7
at Illinois W 3.8
at Michigan W 3.4
at Indiana W 3.1
vs Wichita State W 2.6
vs Oklahoma State W 2.0

The bad news: Five opponents blocked more than 14% of our 2-point attempts.  We lost 3 of those games.

The good news:

  • For two of those losses, we were missing Goran Suton, who ususally finds a way to get his shot off near the basket without getting it blocked.
  • Minnesota–which ranks first in the country in defensive block % at 19.0%–never managed to block more than 11% of our 2-point attempts in the three games we played against them.
  • We managed to beat both USC and Louisville, despite offensive block percentages above 13%.

What does this mean for tomorrow night?  I’m not quite sure.  The Minnesota thing is encouraging–except that their high block% is more a function of multiple guys than of a single dominant shot blocker like Thabeet.

Maybe Izzo will have a fabulous gameplan that gets Thabeet out of position and creates some easy looks near the basket.  Absent that outcome, knocking down 3-pointers and mid-range shots doesn’t look to be optional in this game,  given that UConn’s very unlikely to put us at the free throw line with any frequency.  (Shooting a combined 14-29 from 3-point range certainly helped our cause in the USC and Louisville games.)

Read Full Post »

It’s now been 30 hours since MSU punched its ticket to the Final Four.  And the buzz still hasn’t worn off.

Just over 115 hours remain until the team takes the floor in Detroit against the Huskies.  So we’ve got plenty of time to break down that match-up.

For now, let’s step back and savor the four-game run our Spartans have made to get to this point.  For some context, here are the per-game stats for the eight players averaging double-digit minutes in NCAA Tournament play:

Lucas 31.3 12.8 40.7 40.0 94.4 1.0 0.5 5.5 2.8 1.8
Suton 29.0 14.3 40.0 50.0 87.5 3.5 8.0 2.3 2.3 2.3
Walton 26.5 7.0 41.4 100.0 0.3 1.8 2.5 1.3 1.3
Summers 23.5 8.8 71.4 58.3 66.7 1.8 3.0 1.0 1.0 0.8
Green 21.0 9.0 63.6 66.7 1.5 4.3 1.3 1.3 1.3
Allen 20.5 6.8 30.0 33.3 100.0 1.3 1.3 2.3 0.5 0.0
Morgan 15.0 5.8 31.8 25.0 60.0 1.3 1.3 0.5 1.3 0.3
Roe 14.3 4.0 60.0 57.1 0.8 1.3 1.3 0.3 0.0

The amazing thing is the balance in statistical contributions.  Goran Suton has obviously stepped up his play–14.3 points and 11.5 rebounds per game are well above his season averages–but beyond that it’s been a team effort.  Kalin Lucas’ scoring average is actually below his season average, although he’s scored efficiently and posted a respectable 2.0 assist-to-turnover ratio.

Other notes:

  • Three guys shooting 40%+ from 3-point range.
  • The free throw shooting has been very good.
  • Six guys averaging at least one offensive rebound per game.
  • Seven guys averaging at least one assist per game.
  • Only one of the eight players (Raymar Morgan) has more turnovers than assists.  (This is true of only 3 of the 8 players for the full season: Kalin Lucas, Travis Walton, and–you guessed it–Draymond Green.  Speaking of Green, did you know that, going into the weekend’s games, he led the team in charges drawn on the season with six?  No other MSU player has recorded more than three.)
  • That’s not enough, even more on Green: Fifth in minutes, third in scoring, second in rebounding, tied for third in steals.  I can’t come up with a good historical comparison for an MSU player who’s stepped up more from his regular season level performance to become such a major contributor in postseason play.  He’s basically offset Raymar Morgan’s lack of offensive production.
  • What a great time for Durrell Summers to find his shooting stroke: 80.0 eFG% on 19 FG attempts.
  • Look for Delvon Roe to be a factor on Sunday.  He’s played limited minutes, largely due to some bad match-ups in terms of smaller and more athletic opposing lineups.  But the numbers don’t indicate he’s playing poorly.

OK, with all that, let’s talk about what individual plays you’ve enjoyed the most over the past four games.  The Kalin Lucas and-one basket to break the tie against Kansas in the final minute has to be at the top of the list.  What other plays go on the highlight reel so far?

Read Full Post »

Pop Quiz

Among the players in MSU’s regular ten-man rotation:

  • Which player ranks 3rd on the team in offensive rating?
  • Which player ranks 2nd (tie) in defensive rebounding percentage?
  • Which player ranks 4th in assist rate?
  • Which player ranks 3rd in block percentage?
  • Which player ranks 2nd in steal percentage?
  • Which player ranks 2nd in free throw rate?

Hint: The same player is the answer to all six questions.

Got an answer yet?

No cheating.


Are you sure?


The answer is . . .

Draymond Green

Not bad for a guy that Tom Izzo was planning to redshirt up until preseason practices.  Green has obviously posted these tempo-free stats in fairly limited minutes*, but the stats indicate he’s been productive across the board when he’s been on the floor.

The simplest complimentary descriptor one cay apply to Green is “basketball player.”  He makes plays.

Green is the odds-on favorite to start next to Delvon Roe in the frontcourt next season.  For now, though, he’s gone from preseason afterthought to key postseason contributor in a span of four months.

*I find that individual tempo-free stats are actually more useful for bench players than for starters.  For a guy playing 30 minutes/game, we have a pretty good idea what 15 points, 8 rebounds, and 1.5 blocks per game means.  For a guy playing 10 minutes a game, it’s harder to know what 3 points, 2.5 rebounds, and 0.5 blocks per game means.  (Numbers are made up.)  The tempo-free numbers are, threrefore, more illuminating for the guy getting limited minutes per game–once you have a full season of data, at least.

Read Full Post »

Going into NCAA Tournament play, our starting lineup looks like it may finally be completely healthy.  Kalin Lucas, Travis Walton, Raymar Morgan, Delvon Roe, and Goran Suton should all be able to play 25-30 minutes per game at near 100%.  (Roe is perhaps the exception to this statement, but he’s certainly as ready to contribute effective minutes as he has been all season.)

That’s reason for optimism, as there have been only a few stretches of games when all five of those players have been fully available.  And we’ve been pretty good in those stretches.

The concern is what the guys coming off the bench will be able to contribute in any given game.  In particular, Chris Allen and/or Durrell Summers will need to play 20+ minutes per game as the first reserve options when the starters rotate out.  Allen has been erratic all year, mixing one or two solid games at a time with stretches of games in which he seems to be completely off kilter offensively.

Summers, meanwhile, became a major contributor when Morgan had to sit out with pneumonia, scoring more than 20 points in 3 games of a 4-game stretch at one point.  More recently, though, Summer’s scoring touch has disappeared completely.

Here are Allen’s and Summer’s key stats over the last 10 games:

Allen Summers
Mins/G 18.4 20.9
Pts/G 8.2 4.7
2pt% 48.3 35.7
3pt% 33.3 16.7
FT% 81.8 62.5
Reb/G 1.9 2.9
Ast/G 1.2 0.5
TO/G 1.2 1.5

Summers’ numbers are abysmal.  His 3-point stroke has gone missing, and there haven’t been many opportunities for him to score in transition.  The only thing he’s continued to bring to the table is the ability to rebound from the perimeter.

Allen’s stats actually look pretty good on a per-game-average basis–although they’re boosted somewhat by a 16-point performance in the home game against Indiana that began the 10-game stretch.  His shooting numbers are acceptable, his passing ability has been a factor in flashes, and–in my subjective view–he’s become a pretty solid man-to-man defender.

So which player do you prefer to see as the first player off the bench?  Do you continue to give Summers major minutes, hoping he can regain mid-conference-season form, giving the team a third 20-point scoring possibility to go with Lucas and Morgan?  Or do you give Allen the minutes, recognizing that he’s likely to give you a brief flash of scoring in each game, but probably isn’t going to be a game-changer?

The numbers point toward Allen, but his 2-11 shooting performance (0-6 from beyond the arc) against Ohio State in Indy remains fresh in our minds.  The third option, for which we don’t have much data, is Korie Lucious.  He was great in the Ohio State game, but he makes us really small on the perimeter and that game was the first time he’s played more than 15 minutes in a game as a collegiate player.  Assuming, he’s going to play 5-10 minutes at point guard when Lucas is out of the game, it’s hard to see him playing another 10-15 minutes on the wing.

Let’s hear it: Who should Tom Izzo point to when he makes his first substitution tomorrow night?  (And no fair saying Draymond Green.)

Read Full Post »

Wednesday Night Links

That’s Not Enough: Even MORE Evidence of How Smart I Am!

Patrick at It’s Just Sports posted some very kind words about this blog today.  This reminded me that I had made one other preseason prediction in an interview I did with Patrick last November:

And, for this occasion, I’ll got out on a limb and say that at least one banner gets hung in the Breslin Center at the start of the 2009-10 season–be it for the Big Ten regular season championship, the Big Ten Tournament championship, or a Final Four appearance.



Can We Put It All Together?

A graph of game-by-game offensive (blue) and defensive (red) efficiency figures for MSU in conference play this season, based on last year’s methodology:


The graph tells a pretty simple story: We got much better on defense as the conference season progressed, but we also regressed some on offense. There are three factors I can point to in terms of our offensive efficiency going down:

  1. Raymar Morgan’s illness (which contributed to our offensive rebounding percentage falling off somewhat).
  2. The late-season 3-point shooting slump.
  3. Playing Illinois/Purdue–and their ball-hawking, turnover-inducing man-to-man defenses–three times in the final six games.

On defense, the team really came together playing Izzo’s man-to-man scheme–which requires a lot of switching and hedging–as they avoided mental lapses resulting in good 3-point looks.  Of our final 9 conference opponents, only two shot better than 30% from 3-point range.

If we can regain the offensive rhythm (and rebounding dominance) from early in the conference season and combine it with the much-improved defensive cohesiveness the team showed down the stretch, we’re going to be a legitimate Final Four contender.  Of course, if the offense doesn’t regain it’s early form and our defense doesn’t hold up against unfamiliar conference opponents, the forecast is much less rosy.

I’m not going to post the full group of graphs for all 11 teams this year (I don’t see as many interesting stories this year; lots of .500 teams that played at a .500 level for most of the season).  But I will say one thing: Beware the Badgers.  Per usual, they appear to be peaking at the right time.


Read Full Post »

Older Posts »