29 December 2014

NU's 2015 B1G Projections: Post Non-Conference Edition

NU is now on the eve of beginning its 2015 B1G campaign.  What do the kenpom data have to say about how things will play out?

The events of the past 10 days -- which includes not only a 20 rung improvement in NU's kenpom rankings thanks to three decent wins but also the further degradation of the B1G's overall strength -- have marginally improved NU's outlook for B1G play.  The bad news is that the mode of the Monte Carlo simulation still has NU winning four games.  But the good news is that on a weighted average basis NU is expected to win 4.7 games which rounds up to five B1G wins.  Beggars can't be choosers.  In the spirit of the glass half full take on things here are three takeaways from the above bar chart:

Good: NU's prospects of a winless B1G season are now only 0.31%.

Better:  NU has a 30% chance of matching or besting its 6 B1G mark from last season.

Best: The prospects of a winning overall season have improved from a depressing 3.1% when we last looked into the kenpom crystal ball to a downright plausible 14.5%!

In terms of the B1G race yesterday we posted the big picture (resubmitted above).  The news is also marginally better here.  In short Rutgers has nosed out NU as the favorite for B1G basement.  However ground yourselves in the fact that there is a greater than a 50/50 chance that NU will finish in a tie for 13th or worse.  New NU futility records loom ominously on the horizon.

Lastly we have come to respect CCC's forthright and at times insightful comments to the media.  On the other hand there is this quote according to an insideNU article posted earlier today "[NU has] by far the most difficult [conference] schedule in the league.  It's like double the second toughest."


No.  No it doesn't.  Nebraska does.**  For an explanation of how this chart is derived see our recent blog post.

** Note: this analysis does not attempt to account for the Home/Away effects of the B1G imbalanced schedule.  This impact is not significant enough to justify tracking down the necessary data and quantifying the effects.  However qualitatively speaking the kenpom win probabilities are constructed in such a manner that is most beneficial for teams to play their single games at home against similarly ranked opponent (i.e., NU would be better off playing its one game against similarly ranked Rutgers at The Welsh rather than at the RAC).  This effect is amplified somewhat as the the quality of the team diminishes (i.e., it is a more important issue to Northwestern than to Wisconsin).

GAMBLE Report 2.02

"You can't really use youth as an excuse anymore.  The youth on this the team that plays has all played a significant amount over the six games.  I think we should know the ups and downs of college now.  We have played enough basketball to know how the college game is and how the feel of the game goes"

-- Vic Law [Daily Northwestern article, December 3, 2014]

The 2015 edition of Northwestern basketball has completed the non-conference portion of its schedule. Has the infusion of CCC-recruited players begun to make its mark?  Do we see any inklings of progress towards the NCAAs glass ceiling?  If you believe Vic Law -- the guy who has been tagged as the "Purple Moses" and will lead NU to the NCAAs promised land -- one might expect so much from his proclamations earlier this month to the Daily NU folks.

Let's see what the GAMBLE Report has to say about the matter.....

Since our last run of the GAMBLE Report NU has played six games and won four of 'em.  NU was favored in five of those games (all but Butler), but the Central Michiganders had something to say about that when they visited The Welsh back on December 17.

NU's collective performance over this six game span has resulted in a modest 4 rung drop in the overall kenpom rankings to #135.  On offense the team has improved by 25 rungs thanks to much improved marks in three of the four factors (all but FTR).  On defense the team has regressed by 23 rungs principally because this team continues to struggle in creating opponent turnovers.  

Of our internal benchmarks on the whole this 2015 team is most comparable (least different?) on offense and defense to the ill-fated 2013 team.  Sure the team is rebounding much better than in 2013, but that difference has been offset by turnovers on both ends.  Overall this 2015 team is 31 rungs worse on offense than the 2013 team and 25 rungs better on defense.  

Things were looking bleaker after the C Michigan game when NU sunk down to #155.  So the optimists can run with the bounce back over the last few games as a sign that this team is starting to gel if they wish.  Although it is unrealistic to expect a return to the 2014 team's defensive prowess we think  that more tangibly it would help matters if NU could shore-up the defensive side of things a bit.  

Regardless the fact remains that this team is still stuck in neutral from where it was at the end of the ill-fated 2013 season and far behind the NIT years.  Since Dr Jim has stated in no uncertain terms that this program will be held to external benchmarks (read: NCAAs) it is crystal clear that much works needs to be done by CCC over the next few years.

28 December 2014

2015 B1G Projections: End of Non-Conference Play Edition

Over the past six weeks B1G teams have worked through their ways through 12 or 13 non-conference games. These non-conference slates were designed by coaches to be of varying strengths with the intent of preparing B1G teams for what is now to come.  There are no more Mississippi Valley States left for young teams like Northwestern to steady themselves nor are there any more North Carolinas lurking for teams to bolster their strength of schedules and earn the fancy of the NCAA selection committee.  It is time for the annual slug fest otherwise known as the B1G conference schedule.

This year B1G teams have at least put on boxing gloves to give hope to even the least ballyhooed of B1G squads.  For this year will mark the end to impressive 4-year run atop the kenpom conference standings.  Enough B1G blood has been drawn by non-conference foes that the B1G now stands fourth in the conference standing -- behind not only a very impressive B12 conference and the expected ACC powerhouse but also the decimated Big East.  If only NU could've eked out a win over Butler...

Before the B1G mayhem begins we thought this to be a good time to take stock of where the B1G teams stand.  Behold our updated 2015 B1G Projections:

These projections are presented in largely the same format as our prior two editions but with some additional information added at the bottom to add some color.  The top section of the chart shows the probabilities of where a B1G team will finish within the 2015 B1G standings.  In case all those probabilities are information overload we've added a last row to this top section with the weighted average expected finishes.

We've also had some requests for projections of teams expected records.  While we won't go so far as to indulge our B1G friends with bar charts like the ones we produce for NU we have tossed them a bone with some key summary statistics at the bottom of this table.  First is the expected conference W-L record.  This is the weighted average win-loss record based on a 500,000 trial Monte Carlo simulation using the individual game win probabilities per kenpom as of the end of the non-conference slate.  Wisconsin is expected to win somewhere between 15 and 16 games on the year while Iowa is expected to be 9-9 and Rutgers is expected to win between 4 and 5 games.  

The next line ranks these expected win-loss records which might be one way to do a power ranking if one wanted to do something more fancy like taking into account the imbalanced B1G schedules than simply going with the kenpom rankings themselves.  

Next we've added a few common statistics on the Monte Carlo distribution including the minimum** wins recorded under a 500,000 trial simulation -- probably of most interest to Badger fans who sadly can expect a minimum number of wins (7) which is greater than the expected win totals for four B1G teams.  We've also shown the mode which is the most likely outcome for the season based on how the teams have played to date.  For the self-loathing types we've also shown the maximum** wins -- NU and Rutgers fans we are sorry to report that neither team did not win 14 or more B1G games even once in 500,000 simulations.  

Lastly because it is a Monte Carlo simulation we wanted to show the probabilities of (1) a team's most likely outcome (roughly about the same but for the least variable teams at the ends of the spectrum -- Wisconsin and Rutgers), (2) a team's chances of going winless (pretty slim for all teams), and (3) a team's chances of running the table (really only relevant for Wisconsin which has about a  1/20 chance of it; while Wisconsin is expected to dominate the B1G this year, they are nothing like Kentucky which has about a 1/4 chance of going undefeated in the SEC).

If all goes as expected -- which it won't -- here's what the betting man says will happen.

1.    Wisconsin
2.    Ohio State
T3.  Michigan State and Maryland
5.    Minnesota
6.    Illinois
T7.  Indiana and Iowa
T9.  Michigan and Purdue
T11. Penn State and Nebraska
T13. Northwestern and Rutgers

The top 6 teams have greater than a 3 in 4 chance of finishing in the top 7 which in our mind is where the bubble will form for this year's 14 team B1G that is ranked fourth in the country.   That means Indiana and Iowa fans could have a roller coaster of a ride this year at least in respect to whether they will make the NCAAs.  

** Of course 0 and 18 wins are the minimum and maximum possibilities for ALL B1G teams, but we think it's more interesting to see what comes out of a 500,000 trial simulation than state that Wisconsin has a 1 in 108 quadrillion chance of going winless in the B1G while Rutgers has a 1 in 3.2 trillion chance of going 18-0 in the B1G.

18 December 2014

On the Ramifications of the Imbalanced 2015 B1G Schedule

We've heard the talking heads, bloggers, and fans lament that NU drew the short stick when the B1G 2015 conference schedules were determined.  Is it Fact or Fiction that NU was a victim of the B1G schedule makers?

Astute sport fans are right to point out that because a team can not play itself (or at least it can't in an official game kinda way) that it is a matter of consequence for the weaker teams to have more difficult schedules and stronger teams to have easier schedules.  This is true.  However it is also true that until the time comes when the powers that be decide to shelve the current 18 game B1G schedule in favor of a true 26 B1G round robin schedule, imbalances in the schedules can and will exist.  And it is because of these imbalanced schedules that there will be beneficiaries and victims from the schedule makers.  Which teams fall into each bucket is not self evident and requires further analysis.  The kenpom.com data will show us the way out of these woods.

The first column of data shows the average ranking of a team's B1G opponents as the schedule makers have set it up for 2015.  The second column of data shows the average ranking under the hypothetical round robin schedule.  The differences between these two average rankings are one way -- and a good way in our opinion -- for determining which teams have benefited from the 2015 schedule makers and which teams have not.

This differences have been computed in the third column of data and color coded in green to denote the beneficiaries and in red for the victims of the schedule makers.  For example let's talk about Maryland.  As the schedule makers have set it up Maryland's B1G opponents have an average ranking of 65.3.  Under a true round robin schedule Maryland's average B1G opponent rank would be a more difficult 59.5.  Thus Maryland on average will be playing an opponent that is about 6 rungs worse during the B1G slate than it would have under a round robin schedule.

As you can see Northwestern is a decisive victim of the schedule makers.  This winter NU will be playing an B1G opponents with an average ranking of 43.8 while under the round robin hypothetical the average B1G opponent ranking would have been 6 rungs worse at 49.8.  If NU fans want to lament this fact then look no further than Rutgers -- the worst B1G team according to kenpom.com -- only appearing once on the schedule.

But while there is reason to make some hay about the B1G schedule NU fans ought not go too with the bellyaching.  Or at the very least please don't use it as an excuse if and/or when NU finishes with a worse B1G conference record than last year (a 6 win benchmark).  Why not?  Because the average 2015 B1G opponent this year is ranked 43.8 which is roughly the same as it was in 2014 (42.2).  That means should NU fail to achieve 6 wins again then NU ought to look in the mirror for the reason why and not the schedule makers.

One last anecdote about this analysis.  Despite the fact that Rutgers only plays NU once the Scarlet Knights can lay claim to being the 5th biggest beneficiaries of the B1G schedule makers. Playing each of the three highest ranked teams in the conference only once really helps to offset their misfortune of getting to play NU only once.

2015 Forecasts: Post CMU Edition

NU had a rough game last night.  Although NU was a 7 point favorite the CMUers left the Welsh with an 80-67 victory.  The loss put a dent in NU's kenpom ranking with NU sagging 25 rungs from 131 down to 156.

This downgrade is reflected in diminished expectations for NU's 2015 season.
NU is now expected to go 12-19 overall for the regular season which is down from 13.5-16.5.  The prospects of a winning season are now just 3.1%.

Within B1G play the "good news" is that NU is still forecasted to win 4 games.  However if NU continues to sag in the kenpom rankings then you can expect a downgrade to just 3 wins.

The most depressing part of this update is the overall B1G picture.

Worse: With a 500,000 trial simulation NU never finished any higher than third.  Even lowly Rutgers had a few observations where it finished 2nd.

Worser: There is a 95% probability that NU finishes no higher than 11th in the Big Ten.

Worst: NU now is the most likely team to finish 14th.  NU is also the most likely team to finish in either 13th (tied or alone) or 14th.


14 December 2014

Forecasting the 2015 B1G Race (as of Dec 14 2014)

Yesterday we looked at NU's overall and B1G prospects for the 2015 season.  Today we've taken it to the next level by forecasting the 2015 B1G race.  Why?  Well for one thing this type of analysis might be of interest beyond our little purple patch of people following NU basketball.  If we are successful in drawing in a wider audience of basketball nerds across the conference then maybe with a boost in advertising revenue we can afford to buy our mamas some new shoes for Christmas!  Hahaha.  We sleigh ourselves.  Rim shot.

More realistically this provides self-loathing NU fans with some objective insight into that burning question: "Will NU make history and be the first 14th place B1G team?"

To focus on more positive things Wisconsin has a 84% chance of either winning or having a share of the B1G crown.  They are the clear favorite in the B1G race which makes their loss to Duke at home all the more unfortunate because it gives the ESPN talking heads fodder for why the ACC is a better conference.  Sparty and Buckeye fans have good reason for hoping their teams will at least be in the thick of the B1G title race.  Newcomer Maryland and Minny have a punchers chance at.  Of those two we found it interesting that Maryland has a slightly higher chance of getting a B1G title than Minnesota even though Minnesota has a higher kenpom ranking; must be a function of the imbalanced B1G conference schedules.  

It should come as no surprise to NU fans, but it is still a slap in the face to see that NOT EVEN ONCE OUT OF 100,000 SIMULATIONS is NU forecasted to have at least a share of the B1G crown.  Same goes for you Rutgers.  Oh where would NU fans be without the Scarlet Knights?  Misery loves company ya know. 

If you set your eyes to the last row then you can see which teams are at the greatest risk of becoming the first 14th team in B1G history.  NU has roughly a 25% chance of that happening which is second only to the 43% chance that our good friends from Piscataway have of earning that dubious distinction.  Welcome to the B1G friends.  Fellow traditional basketball weaklings -- Penn State and Nebraska -- also have some basis for fearing the dreaded 14th place but only with much lower 4.4% and 3.0% probabilities, respectively.

Note that 25% is NOT the estimated probability that NU will finish in the conference basement.  The probability is higher and perhaps significantly so.  That is because embedded in the 13th place probabilities are likely a significant number of outcomes in which two teams tie for the basement.  Similarly the 12th place probabilities have some three-way ties for the basement and so forth.  

Of course NU can add to its list of futility statistics by becoming either the first 13th place team or one of two teams that finished 13th in B1G history.  The chances of that happening are in the second to last row.  NU has a 34% chance of that happening while Rutgers has a 31% chance.  Nebraska and Penn State also have reason to fear that outcome at a 12.5% and 8.5% probabilities, respectively.

If you sum up the last two row then NU has a 6 in 10 chance of breaking new ground in B1G history while Rutgers has a 3 in 4 chance.  NU kicks off its B1G campaign in Piscataway on Tues Dec 30.  That epic battle may go a long way in determining which team sets a new benchmark in B1G futility.  

13 December 2014

2015 Forecast: Pre-Holiday Cupcakes Edition

Here are some bar charts that forecast NUs results using the current win probabilities per kenpom.com.  First is an overall chart:

NU has 31 games on its regular season schedule.  Thus 16 wins or more would mean a winning overall record.  We'll help save you the trouble of breaking out a calculator: NU has just over a 15% chance of a winning regular season record.  If Vegas were to care about NU basketball then they'd likely set the over/under at 13.5 wins.  CCC's 2nd season -- one with favorable cream-puff non-conf schedule and a lightened B1G load -- is expected to finish at (13.5-17.5).  Yikes!

Next up we show NU's forecasted B1G wins:
For those of you who are holding out hope that CCC will improve on last year's 6 win campaign ..... don't.  Kenpom data predict NU has an 11% probability of at least 7 wins.  More depressing: kenpom data gives NU just a 4% chance of at least 8 wins, a 1% chance of at least 9 wins, and less than a 0.25% chance of a winning conference record.  

Conversely NU is much more likely to regress this season. NU has a 75% chance of not achieving 6 B1G wins.  Worse yet is that there is slightly better than a coin flip's chance that NU doesn't even make the 5 win mark.  That is to say a betting man would jump at even odds for a bet that NU finishes 2015 no better than the ill-fated 2013 season.  

That's just sad.

07 December 2014

GAMBLE Report 2.01

December 7.  A date that will live in infamy.  Many will remember the 2014 anniversary as the date the first college foootball playoff bracket was announced, but that circus side show obscures the main event: the unveiling of the first GAMBLE Report for the 2015 season.

More than a year has passed since we published the first GAMBLE Report.  By now NU has made its way through 2/3rds of its cream-puff non-conference season and 1/4 of the entire 2015 schedule.  It's early still, but this seems to be as good a time as any to dig into our assessment of the progress being made under CCC.

For orientation, the first line of data are for the current 2015 season.  Reading from left to right, the kenpom rankings are shown for the team overall, the team's adjusted ORtg, the primary 4 factors for that ORtg, followed by the DRtg and finally by the four factors for DRtg.

This year, we will be using three benchmark comparisons: (1) last year, the first season of results for CCC as a head coach; (2) the star-crossed 2013 season -- which we frankly hope to drop before too long and will do so if/when CCC can put that anomalously low ranking well in his rear view; and (3) the NIT years.

The first thing to notice is that NU's team ranking is now #131.  Not only is this 54 rungs lower than where NU began when it tipped off against HBU, but it also is where NU finished in rankings the last two seasons.  Of course the #131 is also below the average #66 ranking that CBC achieved during the NIT years.  Those clamoring for progress this year must be either frustrated, in denial, or perhaps are patiently projecting a team that is going to be improving during the season.

The #207 AdjO is more than 100 rungs better than last year's woeful offense thanks to nearly across the board improvement in the four factors (all except TO%). While a 100 rung improvement may sound impressive it really isn't because 207 ain't good by any measure.  Furthermore it is still 56 rungs worse than the ill-fated 2013 offense and isn't even in the same zip code as the NIT offenses. In short CCC hasn't given us much reason to forget the days of the efficient CBC-coached Princeton Offense.  On the glass half-full side we do think it's a good sign that things are moving in the right direction on offense as CCC gets his guys onto the roster.

On the flipside the AdjD has come back to earth and regressed 72 rungs to #86 overall.  Looking at the four factors we see a mixed bag.  NU has had some improvement to its poor propensity in turning its opponents over, and NU's defensive rebounding is actually looking like a strength at the moment.  However the negatives are that NU's weak non-conf opponents have done better in terms of shooting percentages and getting to the charity stripe.  With this year's youth movement we had expected some regression to the defense, but our hope was to hold the line at #50 or so.  Currently the 2015 defense is about 50 rungs better than the 2013 defense and 85 rungs better than the NIT years.  Overall we're still seeing a big improvement over the CBC years.