28 January 2015

Is Northwestern Improving?

In our last post we found there was scant evidence that Northwestern has been improving in B1G play.  Since that time NU played valiantly on the road at Maryland -- a game in which NU led nearly wire-to-wire but ultimately succumbed to the all-too-familiar black hole that invariably gnaws at fans, players, and coaches alike.  NU lost in heart breaking fashion to #13 Maryland 68-67.

The end result notwithstanding this game was undoubtedly felt like the best NU performance to date.  Time will tell whether this performance is a sign of enduring improvement in this team.  That said we were still curious what the data say about how this single game performance portends for the future.  Our CCR analysis sheds some light on the topic.




The solid purple line shows NU's game-to-game performance overall.  Notice that the Maryland game is the peak B1G performance to date.  This was helped by the second straight strong performance by the NU offense, but more importantly the NU defense held Maryland's offense to its average B1G performance.  Lo and behold our eyes telling us this was a good NU performance do not deceive us!

More germane to the subject of this post the dotted line shows the trend based on this overall performance.  Whereas the trend was slightly down prior to the Maryland game (chart is in our last post) the trend is now moderately upward with a game-to-game performance of 0.02 PPP.  

For now the data show that NU is improving in B1G play.  Of course this is subject to change as new data become available, but at least this is a nice silver lining to that gut-punch of a loss on Sunday.

24 January 2015

GAMBLE Report 2.04

NU (10-9, 1-5) is now 1/3rd of the way through its B1G schedule.  How does the first trimester report card look?

Since our last update NU has narrowly lost games at Michigan and against Ohio State.  NU fans and bloggers have widely praised these games as NU moral victories that bode well for the near- and medium-term future.  We're sympathetic to the notion of looking past the distraction of binary W-L results for underlying trends.  But does this narrative ring true?  If it did then we would expect to see a jump in the efficiency statistics.



Compared to GAMBLE Report 2.03 NU has improved its kenpom ranking by a statistically insignificant 2 rungs.  That's hardly consistent with a team improving game-by-game (CCC's words which have been parroted by fans and bloggers).  Furthermore this so-claimed improvement trend is not apparent even when you look back to the beginning of B1G play when NU was ranked #137 going into the game at Rutgers.

Decomposing the efficiency statistics we see that the offense continued to make strides.  The offense improved by 32 rungs over the past two games -- propelled by improved OR%.  Perhaps the rising offense is the source of the optimism?

Unfortunately the gains on offense have been offset by a defense that continues to regress this year -- dropping a further 16 rungs since our last report.  Of the four factors eFG% dropped 15 rungs, and an already pathetic TO% slipped another 24 rungs to #334 out of 351 teams.

Remember that preseason narrative put forth by insidenu that CCC is a defense-first coach?  We questioned it at the time as we couldn't recall any such claims made by CCC at the time of his hire.  We still would welcome a refreshing of our memories with such claims as this season has done nothing to demonstrate that CCC truly is a defense-first coach.

And looking at the big picture this team's AdjO and AdjD continue to converge with CBC's 2013 team -- the one that got him fired.  This year's offense is 3 rungs better than the 2013 team while the defense is 6 rungs worse.  It's difficult to get a closer alignment than that!

NU's lack of progress in the kenpom rankings comes as no surprise when you consider how NU's opponents performed against NU relative to their overall B1G efficiency statistics (as depicted in the CCR results below).  At Ann Arbor NU held the Wolverines to slightly below their average B1G results (-0.012 PPP) to date.  Meanwhile NU's inability to slow down frosh-phenom D'Angelo Russell on offense meant the Buckeyes performed above their overall average B1G performance (0.048 PPP).  Neither of these performances were wildly out of line from expectations which is why the overall kenpom ranking remains essentially unchanged.




Trendwise during B1G play the offense has been improving at a rate of 0.044 PPP per game while the defense has been regressing at a rate of -0.048 PPP.  Overall the net trend is downward (see the dotted line).  At this point it's hard to come to any other conclusion other than this improvement narrative is nothing more than wishful thinking.  

19 January 2015

2015 B1G Projections - First Trimester Edition


"Free at lastFree at last, Thank God almighty we are free at last"

 Martin Luther King Jr.I Have a Dream

Happy MLK Jr Day!  In the spirit of the day here's to hoping that someday the NCAA selection committee will be free of the scourge of such popular and misleading evaluation statistics as the RPI and Rebounding Margin.  Until that day we endeavor ever more to bring efficiency statistics to bear in evaluating B1G basketball.

We are now approximately one-third of the way through the B1G regular season.  While there is still much work to do B1G teams have already completed the early roadwork toward B1G and postseason glory.

At the top of the standings Wisconsin has held its ground since the last update and remains the favorite for winning the B1G crown.  But the Badgers are in for a battle from Maryland which has upped its odds of winning or sharing the title from 33% a week ago to 48% as things stand today.  Much of that improvement has come at the cost of the Spartans whom the Terps vanquished in style this past weekend.  Iowa has leapfrogged Ohio State as the dark horse in the B1G title race with about a 9% shot at achieving this feat.

The bottom of the standings have also seen some movement over the past week with Penn State (0-5) solidifying its status as front runner for the B1G basement.  Rutgers' surge has propelled the Scarlet Knights to dark horse status for the B1G basement (9.5% chance at finishing 14th) and left Northwestern in its dust as the second most likely 14th place finisher.

Over the weekend the Gophers managed to win their first game over Rutgers, but that expected likely outcome has done little to improve the slippage in the projected standings that occurred over their five game slide to begin B1G play.

Illinois saw its fortunes drop over the past week thanks to a bad loss to Indiana at home over the weekend.

 Iowa had the biggest upward move in its projected win total -- an increase from 10.0 wins to 11.7 wins.  Penn State had the biggest downward move in its projected win total -- the Nittany Lions are now projected to win 3.8 games which is down 1.0 win from our projections a week ago; losing a winnable home game in OT to Purdue left a mark.

In comparison to the pre-B1G play projections Iowa is the biggest winner.  The Hawkeyes are now projected to win 11.7 games vs a projection of 9.0 wins back in late December.  Maryland has also been a big winner upping their projected win total to 13.7 from 11.2.    Minnesota has been the biggest loser in B1G play with their projected fortunes decreasing from a win total of 10.4 games back in December to 7.3 wins as of today.  Penn State has endured nearly as miserable a start to B1G play with their projected win total decreasing from 6.8 to 3.8.

Northwestern and Nebraska have performed as projected in B1G play.  Other teams that haven't seen much movement in their projections are Michigan State, Rutgers.

Last but not least here again are the cumulative win probabilities...




18 January 2015

Weekend CCR awards

It's time to dust off the Carmody Court Ratio analyses.  These analyses are intended to take into account the quality of an opponent when assessing how a team has performed on an efficiency basis.  We will have more full blown analyses in the future, but for now we wanted to dip out toes back into water with some highlights of this past weekend's action.

Best overall performance:  Maryland.  The Terps easily had the best performance in their 75-59 beat down of Sparty in College Park.  Maryland played nearly 0.3 PPP better than B1G teams have managed to date against MSU.  Wow!

Worst overall performance.  Illinois.  The Illini performed 0.15 PPP worse against the Hoosiers in Champaign than Indiana's B1G opponents have managed to date.  This was principally a defensive breakdown for the Illini.

Best defense.  In a weekend generally marred by subpar defenses the Terps consistently strong defense once again topped the charts at 0.07 PPP better than B1G opponents have performed against Sparty's offense.  The only two other outperforming defenses locked horns in Ann Arbor where the Wolverines narrowly avoided going to OT against Northwestern.

Best Offense:  Wisconsin.  The Badgers turned on the jets against Nebraska in Madison to the tune of scoring 0.26 PPP more than the Huskers' B1G opponents have managed on average.

Worst defense:  Michigan State.  Sparty allowed the Terps to score 0.2 PPP more than they have in B1G games.

Worst offense:  Michigan.  The Wolverines scored 0.135 PPP less than NU has allowed in B1G play.

16 January 2015

GAMBLE Report 2.03

Northwestern now has four B1G games under its belt.  The B1G isn't exactly a monster this year, but it's still a large step up in competition from one of the weakest non-conference schedules -- 330th out of 351 at current look.  Have the freshmen been overwhelmed?  Have they started to live up to the hype?  Or is it somewhere in between?

Our last GAMBLE Report update was run on the eve of 2015 B1G play.  An update may shed some light.  Here goes...



NU is now ranked #136 which is essentially the same as #135 at our last update.   On the whole one might conclude on this basis that the freshmen have not been overwhelmed by B1G play.  Or more accurately not any more overwhelmed by B1G play than by collegiate ball.

On offense the AdjO is now ranked 180 which is up 2 from pre-B1G play.  Even a small increase is somewhat surprising because the OR% has dropped an eye popping 70 rungs.  This large drop suggests a strategic shift in B1G play to back off the offensive glass in order to avoid giving up easy fast break points.  That would make sense given the better rebounding abilities of B1G teams.  Of the other three factors only eFG% improved by 10 rungs which speaks to just how important it is to make hoops (duh!).

The defense continues to regress from last year's lofty standard.  The current ranking of #124 is down 15 rungs from prior to B1G play.  Of the four factors eFG% has slipped 10 spots while TO% has slipped 60 spots to an overall ranking of #310.  TO% is comprised in part by steals % which is also tracked by kenpom, and NU is ranked 350 (of 351) on that measure.  Yeah that zero live ball turnover game against the Illini on Wednesday didn't help.

If offense and defense trends continue then this 2015 team will come even closer to what is an already pretty comparable 2013 team.  The strategies that CCC and CBC employed for these teams differ to the naked eye, but on an efficiency basis the two teams are remarkably similar at both ends of the court.

1.5 years into the CCC era, and statistically speaking there has been no progress.   We're ok with that.  Next year?  Not so much.

12 January 2015

2015 B1G Projections - Post Rutgers Upsetting Wisconsin Edition

Tonight Ohio State battles Oregon for the first ever "FBS" college football playoff championship.   That's all well and good, but the other big news is that on Sunday "The Upset" of the 2015 B1G basketball season may have occurred in Piscataway, NJ when Rutgers took down a Kaminsky-less Wisconsin team that also lost the services of PG T Jackson during the game.  

If these Wisconsin personnel woes persist then it's quite possible that the Badgers won't run away with the B1G title as had previously been assumed to be a foregone conclusion.  At the very least we are glad that The Upset put to rest the silly Seth Davis ruminations that Wisconsin was going to sweep the B1G regular season despite the fact that the Badgers stood only a 6 percent chance of making that happen.

So how has that upset as well as a weekend's worth of other action affected the B1G projections?  Behold...

Despite the upset loss at Rutgers, Wisconsin is still the favorite to win the B1G.  However the door to the B1G crown has opened much wider for the likes of Maryland and Michigan State, and even Ohio State has seen its odds marginally improve despite the Buckeyes' loss at Indiana on Saturday.  

The effects of "The Upset" were more profound at the bottom of the projected standings with Rutgers handing over to Penn State the dubious distinction of becoming the first team to finish 14th in the B1G.  It's not all bad news for the Nittany Lions as Rutgers and Northwestern -- despite a near upset in East Lansing on Sunday -- also are still very much vying for that little piece of B1G history.

The updated win projections show just how far Maryland has narrowed the gap on the Badgers.  The Badgers are now expected to win 13.8 games his B1G season while Maryland is expected to win 13 games.  That smells like a B1G title race to us.  The three leading contenders for the B1G cellar are all within a half game of each other on an expected win basis so that race to the bottom is heating up.  
While Wisconsin's projections have suffered a serious dent the biggest loser since B1G play has to be the Gophers which lost their fourth straight game when they couldn't complete their bid for a road victory in Ann Arbor.  Minnesota is now expected to go 8-10 in B1G play whereas they were expected to win 10.4 games prior on the eve of conference play.

On a brighter note aside Maryland and Michigan have seen the biggest upward movement in the projections.  Maryland's projected win total of 13 is up from a pre-conference projection of 11.2 wins, and the Wolverines have improved by a game and a half from an expected 7.4 wins to 8.9 wins.

Last but not least is our update to the cumulative win probabilities below.  Enjoy tonight's football game everybody!




09 January 2015

2015 B1G Projections - Jan 9 2015 Edition

We plan to update these B1G projections every few games or so as our schedules allow.  All but two teams have three B1G games under their belts by now (Northwestern and Indiana have two), and all B1G teams are off tonight.  This is as good a time as any for our first mid-conference-play update to our 2015 B1G projections.

First up is an update to the projected standings.  We've highlighted and bolded the most likely outcome prior to conference play so that you can track how teams have improved or regressed relative to pre-B1G expectations.  For example Iowa fans have reason to feel better about their team than 10 days ago since the Hawkeyes are now most likely to finish in 5th (including ties) relative to prior expectations of a 7th place finish.  Conversely Minnesota fans are likely to feel down about their winless Gophers as their most likely finish has dropped from 4th to 8th.

Close observers will note that this standings update no longer has summary win statistics at the bottom.  That is because we have added the following two tables that provide more detail on the win forecasts.  Voila...

The win probabilities that comprise the bulk of this new table are self-explanatory.  In the bottom part of the table first we show the weighted average (W-L) record derived from the detailed win probabilities above (i.e., the current "Forecast").  These current forecasts are then ranked. Next for comparative purposes we have copied and pasted the "Pre-B1G" play expected (W-L) records.  The final row of data shows the current records which could be subtracted from the current projections to determine how many more wins and losses are in store for teams over their remaining regular season schedules.

Last but not least is a table that shows the cumulative win probabilities for various win totals.  It is simply derived from the win probabilities summarized in the above table.




Enjoy.