B1G Expectations: 2016 Week 9 Total Conference Wins Update

Submitted by Ecky Pting on

Big_Ten_Conference_logo.png Expectations

Week 9 Conference Wins Update

Preamble

“Break their hearts my pride and hope, break their hearts and have no mercy.”

- Miss Havisham (Charles Dickens’ Great Expectations)

Nine weeks into the season and Michigan has passed a major milestone. A Big Ten record of 5-0 marks the best start to a Big Ten season since the Lloyd Carr era, and provides further validation of Michigan’s deliverance from the clutches of incompetence. In so doing, the Wolverines at long last methodically exorcised another demon that had been causing a degree of agitation for some years now and had grown particularly acute over the preceding 378 days. That exorcism included a small token of atonement in the form of Jabrill Peppers’ 85+ yard defensive 2-point conversion return of a botched option pitch to close out the scoring and the game. Not quite a reciprocally calamitous outcome as last year, but it does make a nice bookend. This is not to say that anything in the fourth quarter really mattered much at all, except in the chronically delusional consciousness that is Sparty in the collective sense. Just to illustrate in the form of probabilities and fancy stats that are the basis of this diary, a couple of points are worth noting that one should consider in putting this game into some sort of perspective. They are as follows:

  • The lowest in-game win probability for Michigan was 88.4%, based on an initial win probability of 96%. Regardless of the starting value, however, the minimum win probability occurred with 7:46 remaining in the first quarter. This was immediately after Wilton Speight threw the long incompletion to Drake Harris on M’s first offensive play.

     
  • The game entered the fancy-stats definition of garbage time at beginning of the fourth quarter with Michigan up 17 points. Shortly thereafter Kenny Allen converted his third field goal of the game with 14:42 remaining to put M ahead by 20. All subsequent scores or yardage accumulations are inconsequential when viewed through the virtual reality goggles of fancy-stats.

     
  • It remains that the only team that has taken Michigan to the no-garbage-time wire this season is Wisconsin. With any luck, they might be the only team.

Nonetheless, with a renewed commitment the focus for Team 137 turns now toward the work that remains to be done to achieve the lofty goals that are coming within reach, and to mitigating the few risks that threaten those objectives. The same might be said in regard to its perennial nemesis the Buckeyes, who after being taken to the wire for the third straight game by Northwestern in Columbus (with Indiana 3:51 from being a fourth), continued to exhibit weaknesses on both sides of the line, manifest in an inability to sustain drives or get its defense off the field.

Background

The impetus of this diary is the desire to characterize the competitive landscape of the Big Ten Conference through the synthesis of total win probability distributions for each of the teams. The distributions are derived from the relative expected points ratings from Bill Connelly (S&P+), ESPN (FPI), and occasionally Ed Feng (The Power Rank). The key is that the ratings are based on expected points, which are in turn translated into win probabilities. Each of these three ratings are generated from their respective advanced statistical analyses and metrics. In doing so, they achieve varied results ... some more pleasing than others depending on your point-of-view.

Anyway, here you will find further ruminations on said statistics into still more statistics as a means for enabling further discussion, jumping to conclusions, flying off of the handle or goading your rival. Also included is a fresh look at the all-important head-to-head win-differential probability distribution for the matchup between a select pair of contenders in the B1G East.

Schedules, Margins, Probabilities & Distributions

B1G East Schedules & Margins Rundown

The table of schedules below shows the overall schedules for all seven teams in the B1G East based on the Bill Connelly’s S&P+ weekly ratings. The last table simply shows a rank-ordering of the B1GE teams based on their  expected in-conference win totals, it’s not a projection of divisional standings based on projected wins, losses, and tie-breakers.

Michigan, by virtue of a first-half throttling of Sparty, continues as the #1 ranked team in all the land as per S&P+. U-M has expanded its lead now to five spots over of OSU. Penn State, meanwhile, moves up the S&P+ ranks another spot and now stands at #15, holding at  4th best in the B1G, and 3rd in the East.

S&P+

Looking at the S&P+ probabilities, the Wolverines lead the B1GE with about 8.6 expected B1G wins, ahead of the now 2nd place Nittany Lions by 1.3 expected wins. The Buckeyes now trail the Nits by 0.3 wins. Michigan is the only team in the B1G at this point expected to exceed 8 wins; OSU and PSU the only teams expected to exceed 7 conference wins. U-M and PSU are both still favored in all of their remaining games. As such, OSU is an underdog in The Game, but even before that, the OSU matchup versus the once-beaten Huskers in Columbus on Saturday may make or break the Buckeyes’ prospects for a B1GE title.

Indiana, after dispatching Maryland, remains in the fourth spot. The Hoosiers, with nearly 4.2 expected wins and being favored in 2 of their remaining games, are on track for bowl eligibility. Meanwhile, Maryland is still on the bowl-eligibility bubble looking for its sixth win, but with slightly more than 3 expected wins and LOLRutgerz still on the schedule, the Terps have an ace in the hole.

FPI

As expected, the FPI results differ slightly. Here Michigan retains the #2 spot, while OSU slipped one spot to #5. In turn, M tops all teams in the B1GE with just over 8.3 expected wins, maintain its 1.3 win lead over OSU. As was the case last week, FPI results show U-M to be favored in all of its remaining games; the only game in which OSU is not favored is The Game. The margin, however, has narrowed to only 1.6 points. Penn State followed up its stunning upset of OSU with a demolition of Purdue, and has actually creeped ahead of OSU by 0.3 wins to claim the #2 spot in the B1GE at nearly 7.1 expected wins. PSU has now become bowl-eligible, is favored in all of its remaining games, and is on track for a 10-2 season. After beating Maryland, Indiana hold the #4 spot firmly at about 3.8 wins and remain in the edge of bowl-eligibility. Likewise, Maryland, now with only 3.2 expected wins and as a favorite in only one more game is on the bowl-eligibility bubble. MSU, meanwhile, is now favored in 2 more game this season: Illinois and LOLRutgerz in what should be a tremendous battle for The Participation Trophy.

Here’s a link to a chart showing the results from the Power Rank-ings.

B1G East Expected Conference Wins Distributions

The bar plots below show the expected total overall wins distributions for teams in the B1G East, in alphabetical order. Noted above each bar is the probability for that number of wins (you may need to click & embiggen to read it). The bar with the highest value is the most likely outcome (the mode). Also flagged on each plot is the expected overall win total (the mean). The last line plot is just an overlay of the same data from the other seven bar plots.

S&P+

Once again Michigan stands apart from all other with the highest mode, and that mode, dear reader, is that of an undefeated 9 wins! What’s more, PSU now shows the next highest mode at 8 wins, with a skew toward 7 wins. Following the Nits are the Buckeyes with a strong and balanced mode … of ... 7 wins, which suggests that the most likely outcome is that OSU will lose one more game. The complementary nature of OSU losing one more and Michigan winning out gives statistically-minded folks a warm-fuzzy when pondering the covariance of the M and OSU distributions.

Still, a bit of a pile-up remains at the 8-win mode. A quick computation shows that the much ballyhooed three-way tie among Michigan, Ohio State and Penn State now has a probability of about 11%, largely because as the overlay shows, it’s almost 5 times as likely that OSU or PSU at least more game. Nonetheless, it’s worth repeating that in the new B1G tie-breaker rules, overall winning percentage eliminates PSU first from the three-way tie since they lost an OOC game to Pitt. OSU can only advance by winning out, including The Game. PSU could only advance if M drops two more games than PSU from here on out. Clearly, with the Buckeyes’ backs to the wall, the B1GE divisional championship at this point is Michigan’s to lose. The likelihood of UM having an undefeated season at stands at 64.1% (up from the 18.8% before obliterating Rutgers) or nearly 2:1 odds in favor.

Indiana still sits at 4 wins, but now leaning toward 5 wins, whereas Maryland is balanced 3 wins. MSU has improved to a mode of 2-wins, but with a strong lean toward a single win, while Rutgers is still looking to go winless in the B1G.

FPI

What’s remarkable about these FPI results are the nearly indistinguishable distributions of OSU and PSU. Both show the same 7-win mode leaning slightly toward 8 wins. Michigan leads the division and the conference with its 8-win mode, leaning strongly toward an undefeated 9 wins. Also, UM registers a 43.7% chance to win out.  Once again, a bit of a logjam appears at the 8-win mode. FPI works out to a similar likelihood of the three-way tie as S&P+: 10%, largely because as the overlay suggests, it’s almost 5 times as likely that PSU and OSU lose at least one more game.

From the 3 contenders at the high end, a 3 win gap separates the remaining teams in the B1GE. In order, they are Indiana, Maryland, MSU and LOLRutgerz at 4, 3, 2, 0.5, respectively. at the 4-win mode. Maryland and Indiana, with modes of 4 wins and 3 wins, respectively. LOLRutgerz at MSU - the Slobber-Knocker of Self-Loathing - is shaping up to be an instant BTN Classic.

Here’s a link to a chart showing the results from the Power Rank-ings.

B1G West Schedules & Margins Rundown

The next table of schedules shows the overall schedules for the B1G West based on the Bill Connelly’s S&P+ weekly ratings. Again, the last table simply shows a rank-ordering of the B1GW teams based on their  expected win totals - it’s not a projection of divisional conference standings per se.

S&P+

The B1GW S&P+ results have the principal contenders, Wisconsin and Nebraska, at 6.6 and 6.2 expected wins, an almost exact reversal from last week. For Wisconsin, the defeat of Nebraska has put them in control of their destiny. Having traversed most of the gauntlet that was their early season schedule and being favored in all their remaining games, the Badgers are on the inside track to the B1GCG. Unless the Huskers can knock off the Buckeyes in Columbus on Saturday, they will at best finish in a tie with the Badgers, but will of course lose the tie-breaker.

Northwestern, Minnesota and Iowa remain congealed in the second tier of likely bowl-eligibles. Northwestern, now expecting nearly 5.3 B1G wins, has an edge of about 0.2 wins over Minnesota. From there, Iowa lags another 0.6 wins. The resurgent Wildcats, after a gritty loss to the Buckeyes in Columbus, now face the Badgers. Still, the Cats are favored in 2 of their remaining games. Iowa still appears to have reached its high-water mark, being a favorite in only 1 of its 4 remaining games (at Illinois). That one win however, may suffice to make the Hawkeyes bowl-eligible. Alas, Illinois is no longer favored in any of its remaining games, as is the case for Purdue as well.

FPI

FPI results now have Wisconsin leading the B1GW with nearly 6.3 expected wins, with Nebraska trailing close behind at just under 6.2 wins. Like S&P+, FPI has the Badgers favored in all its remaining games, with the Huskers underdogs in only its next game against OSU. Northwestern follows at just under 5.5 wins (up from 5.3 last week despite losing to OSU), with Minnesota in the 4-spot another 0.6 wins back. Wrapping up the likely bowl invitees is Iowa at about 4.7 expected wins. Northwestern and Minnesota are both favored in 2 of their remaining games, but Iowa is favored in only one of its four remaining.

Here’s a link to a chart showing the results from the Power Rank-ings.

B1G West Expected Conference Wins Distributions

The bar plots below show the expected overall win distributions for the B1G West teams, in alphabetical order.

S&P+

The story in the B1GW continues to be how close the race to Indy remains, although with only 4 games left the opportunities to move ahead are fading. Five teams still have modes in the 4 to 7 win range. Wisconsin is currently the only team with a mode of 7 wins, and it’s a very strong mode indeed. Nebraska now sits at the 6 win mode, leaning toward 7 wins, but would need the Badgers to lose again to advance. Northwestern and Minnesota are balanced and nearly indistinguishable at the 5 win mode. Iowa remains at the 4 win mode, skewed toward 5 wins.

FPI

The FPI results tell a slightly different story. The race here is even tighter than the one predicted by S&P+, showing 5 teams in either the 5 or 6 win modes. Both Wisconsin and Nebraska register 6 win modes; both Iowa and Minnesota register with 5 win modes; and Northwestern is evenly balanced between the 5 and 6 win modes. Wisconsin has a stronger lean toward the 7 win mode and also has the advantage of having beaten Nebraska and Iowa already, but has yet to face Northwestern or Minnesota. Altogether, those five teams should attain bowl-eligibility (Northwestern and Iowa have yet to clinch it). Meanwhile Purdue and Illinois both look a game better than LOLRutgerz and Sparty at this point.

Here’s a link to a chart showing the results from the Power Rank-ings.

Michigan vs. Ohio State Big Ten Wins Differential

Background

The win-differential distribution simply shows the likelihood of one team (say, Michigan) finishing with a conference record that is some number of games better or worse than another team (say, Ohio State). Keeping in mind that in the event of a tie, the winner of the head-to-head match up determines the tiebreaker … the probability of the teams having identical conference records (i.e. a win differential of zero) heading into the final head-to-head meeting is then pro-rated in proportion to the win probability of the head-to-head game. The same principle also applies to the probabilities of either team having a one-game lead going into (our outside of) the head-to-head (i.e. win differentials of +1 and -1). This is because a team trailing by one game would still clinch the tie-breaker by winning the final head-to-head game. Thus, the total likelihood of Michigan finishing ahead of Ohio State is the sum of all the maize-and-blue shaded bars (i.e. U-M wins two or more games than OSU), plus a proportional split of the -1, 0 and +1-differential bars. It’s worth noting that this total likelihood does not indicate the likelihood of making it to the B1G Championship, as it says nothing about how other teams in the B1G East do, or even how Michigan or Ohio State do in the absolute sense. For example, if both teams were to finish tied in the B1G at 7-2, which means that UM and OSU would be losing 2 games each, at that point another team (Penn State) may have the lead.

S&P+ Results

Beginning as usual with the results of the S&P+ analysis, this week’s chart still shows that the most likely outcome (70.9% likelihood) is that U-M is one game up heading into Columbus. So, it stand that The Game will likely decide who will play for the B1G Championship. What’s more, looking at the head-to-head matchup, the win probability for Michigan has expanded again to 69.0% (that’s a cool 7.7 point margin!), so UM collects a 48.9 point share of the 70.9 points for the likelihood of winning when coming in up one (and finishing ahead two games). OSU collects the remaining 22.0 points.

The second most likely scenario, now with a 21.8% likelihood, is that UM comes into Columbus two games ahead of OSU. Of this possible outcome, UM collects the entire 21.8 points, of course, because UM would still be assured of finishing one game ahead of OSU regardless of the outcome.

The third most likely scenarios that UM comes with the same record as OSU. This scenario has a 5.2% likelihood, of which UM collects a 3.6 point share for its likelihood of winning and finishing in one game ahead. OSU collects the remaining 1.6 points.

The fourth most likely scenario, with a likelihood of 2.0%, is that UM comes into Columbus three games ahead of OSU. All 2.0 points for UM in this case.

One other vanishingly small scenario registers on the chart. Noodling through it is left as an exercise for the reader.

In total according to the S&P+ ratings, Michigan now has a 76.4% likelihood (down slightly from 76.8%) of finishing the season ahead of OSU - better than 3:1 in favor!

FPI Results

Painting a slightly less rosy picture, here is the same chart based on the FPI ratings following the week 9 results. This shows a slightly narrower margin for UM in the race against OSU to the B1GCG. However, as with S&P+, the most likely outcome is that UM heads into Columbus up one game on the Buckeyes. In the head-to-head matchup, UM is rated high enough to overcome OSU’s home-field advantage, giving Michigan a 54.1% likelihood (a 1.6 point margin) to win the game. To sum it all up, UM has a 67.6% likelihood of beating out OSU at season’s end, or a shade better than 2:1 chance.

Here’s a link to the chart showing the results from the Power Rank-ings.

Bonus Section: Nebraska vs. Wisconsin Big Ten Wins Differential

It seems like a good time to take a closer look at the primary contenders in the B1GW since they just met on the field this past weekend, those contenders being Wisconsin and Nebraska.

S&P+ Results

So here are the S&P+ results after running the differential analysis rubbing the Nebraska distribution against Wisconsin. The results are different from the UM-OSU rub. In this case, the most likely outcome (with a 36.0% likelihood) is that Nebraska is ahead of Wisconsin for all games outside their head-to-head matchup. However, since Wisconsin already won the head-to-head game, it’s Wisconsin that collects the entire 36.0 point share of that outcome. In all, Wisconsin now has a 79.3% likelihood of finishing ahead of Nebraska at the end of the regular season.

FPI Results

And here are the FPI results for the Nebraska-Wisconsin differential distribution. Same story as with the S&P+ results, just not so significantly in Wisconsin’s favor.

Here’s a link to the chart showing the results from the Power Rank-ings.

Wrap-up

So there you have it. Evidence continues to mount that Michigan will actually be in a favorable position rolling into Columbus the last Saturday of the month. In the meantime, Michigan will need to remain focused on the intervening games, the first of which is against the Terps. Beyond that, the night game at Kinnick Stadium looms large as the most significant risk between now and Columbus. Kinnick will be an inhospitably dark, cold and hostile environment that promises to be a more challenging test than what the Wolverines encountered this past weekend in East Lansing.

The prospects for Michigan football to play in the Big Ten Championship Game may have reached a plateau this week. Nonetheless, a very competitive game awaits OSU (vs Nebraska) that will be a barometer by which to gauge the true threat posed by the Buckeyes. In all, it bodes well for Team 137 as it continues to work toward its goals to compete for the Big Ten Championship, and beyond.

Yours in football, and Go Blue!

Comments

Bo Glue

November 2nd, 2016 at 11:12 PM ^

Love seeing the UM-MSU win probabilities in there too. It felt about that close to me live. Garbage time vanity yards were meaningless from a win probability perspective.

blueblue

November 3rd, 2016 at 10:42 AM ^

This is my favorite feature, but for the time-constrained:

- Michigan is still ranked #1 in the country according to S&P+.

- Michigan has a 69.0% chance of beating Ohio State and is favored by 7.7 points.

- Michigan has a 76.4% chance of winning the Big Ten East.

- Wisconsin has a 79.3% chance of winning the West.

- At no point the last Saturday's game did Michigan have less than an 88.4% chance of beating the Spartans.

- MSU now has an 83.1% chance of defeating Rutgers on Nov 12 to take the Participation Trophy.

bronxblue

November 3rd, 2016 at 2:15 PM ^

Great stuff.  I usually read it but don't comment, but now that we are in the meat of the season it feels like the numbers really do tell a good story about where the conference stands.  OSU remains the one major hurdle, and like you said, Nebraska is a tough out so who knows.

tasnyder01

November 4th, 2016 at 8:31 PM ^

Takes me back to exam P. Best sentence ever

The complementary nature of OSU losing one more and Michigan winning out gives statistically-minded folks a warm-fuzzy when pondering the covariance of the M and OSU distributions