2018 Week 9 Total Conference Wins Outlook

Submitted by Ecky Pting on November 1st, 2018 at 10:10 PM


“‘I am ashamed to say it,’ I returned, ‘and yet it's no worse to say it than to think it. You call me a lucky fellow. Of course, I am. I was a blacksmith's boy but yesterday; I am—what shall I say I am—to-day?’”

- Pip
 (Great Expectations by Charles Dickens)

Spin Up

On the heels of its annual Improvement Week, Team 139 can pick up where it left off two weeks ago, and resume its Revenge Tour with renewed vim and vigor. With the prospect of settling the third of four outstanding scores, victory in this case will yield sweet validation, vindication, and vengeance.

Next on the slate of course are the Nittany Lions, a team that’s managed to snatch defeat from the jaws of victory (OSU & MSU) nearly as much as it’s benefited from its opponents doing the same (App State, Indiana & Iowa). Consistency - and Efficiency,in particular - has become a bit of an issue for this year’s Nits, which has seen its overall Offensive S&P+ rating decline from #10 at the end of last season to #22, and its Defensive S&P+ rating decline from #12 to #21 as of now, both drops predominantly due to declines in Efficiency (i.e. staying ahead of the chains, or keeping your opponents behind them). What’s more? James Franklin, of course! Amongst audible sighs and lamentations for running backs and OC’s lost, the Reprobate Tactician himself returns once again to Ann Arbor, the site of his monumentally sad call to kick a 21-yard field goal to punctuate a 3rd quarter drive with his team down 28-0, cutting UM’s lead from four scores to four scores, after calling a timeout to… avoid a delay of game which might have pushed them back from the 4 to the 9 yardline? Ice his own kicker? Signal to Harbaugh to please lighten up already? Indeed, it’s Franklin’s own contemptible inscrutability that set this wheel in motion, which brings us to last season, when Franklin took the opportunity to attempt a touchdown on 2nd & goal, up 42-13, on the last play of the game. It's perhaps plausible that his misguided logic had him thinking to hang half-a-hundred on M with a 2-point conversion, which would have required an untimed down to which a winning team is not entitled? Whatever. Turnabout is fair play in big-boy football, which brings us to the clip:

And with that, it's on to the fancy stats. S&P+ continues to hold your ever-lovin' Maize-n-Blue in the realm of CFP contender at #4 in all the land (tops in the B1G), a mere 5.3 expected points behind the #1 ranked Death Star. ESPN's Football Power Index (FPI) dropped UM one spot to #5, moving the Sooners up to the 4-spot. Even so, M is still tops in the B1G. Likewise, the Power Rank drops UM one spot to #6, so averaging the 3 together gives UM an aggregate ranking of #5, which coincidently matches the first release by the CFP committee.

What’s more, all three ratings still concur that this one-loss Michigan team, on a neutral field, is better than the undefeated Notre Dame. ND’s fancy-stats rankings remain in the #6 to #9 range. Yet, given the initial CFP placements released this week, it’s clear that the Fig-things’ head-to-head win is going to supercede any fancy-stats-based comparison. Humans...

Schedules, Spreads & Win Probabilities

As is typical for this part of the season, the focus of the analysis is on the all-important in-conference slate, so as to make a closer examination of the prospects for the teams to make it to the B1GCG in Indy the first Saturday of December.

B1G East Schedule Rundown

The tables of schedules below shows the conference schedules for all seven teams in the B1G East based on the ratings following week 9. The last table in each set 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. Those are found in a later section.

S&P+ Results


With the Boilermakers' loss to the Spartans, the bye week has led to a further erosion of the Buckeyes’ expected conference win total (just below 6.7 expected wins), and a widening of Michigan’s margin over them, at 8.2 expected wins. M remains only team in the entire B1G without a conference loss, and the only team in the B1G that is favored by S&P+ in all of its remaining games (albeit by the same narrow margin of 1.9 points when M goes to Columbus).

Meanwhile, after defeating the Hawkeyes, the Nits narrowed their deficit to 1 win behind the Buckeyes. Sparty also edge up to 5.3 expected wins, a slightly less miserable number than last week. OSU, PSU and MSU – remain underdogs in one game apiece. PSU is a two-score dog to Michigan, and is less than a one-score favorite over Wisconsin. MSU remains a one-score dog to OSU. That said, MSU is less than a one point favorite over Maryland and Nebraska!

Looking ahead, S&P+ shows Michigan remains favored by an even 9 points when it hosts the Nits in Ann Arbor in two weeks, for a win likelihood of 72%, or about 2:5 odds.

Two others – Indiana and Maryland – are seeing a shrinking bubble of bowl eligibility. The Hoosiers are most likely doomed after screwing the pooch at Minnesota last week. Meanwhile, the Terps have a couple of opportunities remaining to clinch a bowl berth being less than a point underdog to Sparty, and slightly favored against the Hoosiers.

FPI Results

FPI continues to have Ohio State favored in all of its remaining games, however, the margins have eroded significantly, and is favored by less than 1.5 points in The Game. Michigan continues to lead the B1GE with 8.0 expected conference wins with a margin of a full win next-best Ohio State. It's then another 1-win drop to third place Penn State at 6.0 wins, with Sparty close behind at 5.6 wins. MSU is a dog to only OSU, however, it is favored over both Maryland by less than one score, and over Nebraska by a TD with a 2-point conversion.

Looking ahead, FPI still has Michigan favored by 5.2 points in its next game vs. Penn State, for a win likelihood of 63%, or about 1:2 odds.

At this point, UM has solidified its position to make a play for all the marbles in the Game in Columbus. Of course, this is provided Michigan can return players like Tarik Black and Rashan Gary from injury and bring the full force of Improvement Week to bear on its next target for revenge.

Power Rank Results


The Power Rank splits the difference between S&P+ and FPI in a sense. Michigan is expected to win just over 8 games, and is not an underdog in any of its remaining games, but that’s not to say it’s favored in all of them. The margin for The Game from PR is the narrowest possible margin - 0.0 points - an exact toss-up. OSU still lags nearly 1.5 expected win behind, however, since they are looking at a second virtual toss-up with MSU in East Landfill. PR still maintains MSU in third place, now at 5.8 expected wins, 0.2 wins ahead of PSU.

Looking ahead, Michigan’s margin over PSU has shrunk to 7.0 points when the Nits visit Ann Arbor in two weeks, for a win likelihood of 65%, or about 2:1 odds.

PR also falls in line with the others regarding the Bubble and Bottom-feeder teams. The Hoosiers, now below 2.0 expected B1G wins and 2 games remaining as less than a one-score dog are not likely to garner a bowl invitation, even if they beat the Terps. As for the Terps, their best opportunity to lock down a bowl bid will be to knock off the Hoosiers.

B1G East Expected Conference Wins Distributions

The bar plots below show the expected total conference 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 right-click & open in a new window 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 multiline plot (bottom right) is just an overlay of the same data from the other seven bar plots.

S&P+ Results


The S&P+ distributions continue to show that Michigan separating at the top of the division in sole occupation of the 8-win mode, with a strong lean toward 9 wins. OSU lags behind, occupying the 7-win mode with slight lean toward 6 wins. As of now, the likelihood of the Wolverines going undefeated in the B1G stands at 37%. The likelihood of OSU running the table stands at just over 18%.

Meanwhile, PSU and MSU are battling for third place. PSU occupies the 6-win mode with a slight lean toward 5 wins, while Sparty has the 5-win mode with strong lean toward 6 wins. Maryland is maintaining its post-season potential with a strongly upward leaning 4-win mode, whereas Indiana is beginning to push up the daisies with its now downward leaning 2-win mode.

FPI Results


The FPI results show a similar separation at the top. Michigan, the #4 team in all the land, is maintaining its advantage as the sole occupant of the 8-win mode. OSU lags behind in sole possession of the #2 spot with a balanced 7-win mode.From that point, a 1-game separation to where MSU and PSU share the 6-win mode with PSU balanced and MSU leaning toward 5 wins. Michigan registers just over a 27% chance of going undefeated in the B1G, while OSU’s chances of winning out stand at just under 27%. The distributions for M and OSU have very similar characteristics, only separated by one expected win, which constitutes OSU’s loss to Purdue.

Power Rank Results


As with the others. the Power Rank shows Michigan claiming the highest mode of 8 expected B1G wins, with slight lean toward 9 wins. OSU lags behind with a 7-win mode that leans well toward 6 wins. One expected win behind the Buckeyes are the Nits and Sparty sharing the 6-win mode with leans toward 5 wins. PR show Michigan with a better than 30% chance of going undefeated in the B1G. OSU has just under a 16% chance of winning out. Maryland has sole occupancy of the 4-win mode, followed a couple wins back by the Hoosiers in a now downward-leaning 2-win mode.

B1G West Schedule Rundown

The next set of schedule tables shows the conference schedules for all seven teams in the B1G West based week 9 results and ratings for the remaining games from S&P+, FPI and the Power Rank. Again, the last table in each figure 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. Those projections can be found in the next section.

S&P+ Results


The story in B1G West has taken a stunning turn toward the bizarre, with Northwestern now sitting atop the expected wins standings at 6.50 expected wins, nearly 0.7 wins ahead of next-best Iowa. Most of this separation is due to the Cats having a half-game lead on the rest of the division by having played (and won) one more game than all the other teams. Still, the Cats have effectively eaten the Badgers’ lunch. As such, the race has reopened for Hawkeyes since their prior loss to the Badgers is rendered moot by the matchup with the now-leading Cats that remains on the schedule. Thus, there remain two teams in the B1GW that control their own destiny: Northwestern and Iowa. Win out and advance to the B1GCG.

That said, the Badgers are still within 0.9 wins of the Cats, but would obviously need to win out and have the Cats drop two of its remaining 3 games. Not a path of high probabilities with Penn State still looming on the Badgers’ schedule. Similarly, the luster has worn off the Boilers to some extent, who are now nearly 1.3 expected wins behind the Cats. Yet, all 4 of the Boilers’ remaining games have less than one-score margins, and they’re favored in 3 of them! But, like the Badgers, they’ll need Cats to drop two. To that end, as it stands now, the Cats are underdogs in two of their remaining games, so - anything could happen. Altogether, the top four teams remain within 1.3 expected wins of each other.

Meanwhile, the winter-emergent Cornhuskers are now on a two-game winning streak. After getting off the ground against FCS Bethune-Cookman, the Huskers’ Death Harvester quickly reached escape velocity last week at Minny. It may just thrash a nuclear warhead or two from its low earth orbit when transiting Columbus this weekend.


FPI Results


The FPI results for the B1G West show an even tighter race than S&P+,  and now put the Cats at the top with about 6.6 expected wins, with the Hawkeyes about 0.5 expected wins back - which is really about even since NU has half-game on everyone else having already played 6 conference games. A critically important wrinkle with the FPI ratings is that Iowa is favored in all of its remaining games, which would mean a return trip to Indy for the Hawkeyes. Meanwhile, the Badgers and Boilers are 1.0 and 1.2 expected wins back, banking on some chaos to ensue that might move them ahead.

Power Rank Results


The Power Rank results show Northwestern gaining some separation from the other contenders. At 6.9 expected wins, the Cats are a solid 1.0 wins ahead of the Hawkeyes. However, much like FPI, PR has the Hawkeyes favored in all of their remaining games, including the pivotal matchup with the Cats, where they’re favored by nearly a touchdown. So, the Iowa loss in Happy Valley may have put NU in the driver’s seat, but Iowa is still riding shotgun and giving NU the stink-eye.

B1G West Expected Conference Wins Distributions

S&P+ Results


The S&P+ distributions overlay shows the chaos of the B1G West. The Wildcats, Hawkeyes and Badgers all share the 6-win mode, with the only differences being that the Cats lean toward 7 wins and the Badgers toward 5 wins, with the Hawkeyes balancing in the middle. Of course, the caveat that now applies is that the Cats have played (and won) one more game. After their let-down in East Landfill, the Boilers have slid back to the 5-win mode, but still leaning upward toward 6 wins. The Boilers still have a relatively high variance in their distribution since all 4 of Purdue’s remaining games have margins of less than one score, with three at about three points or less!

FPI Results


FPI tells a more nuanced story than S&P+, showing Northwestern separating itself into th 7-win mode, but still with a strong lean back toward the 6-win mode that’s shared by Iowa and Wisconsin. The Boilers are hanging in there as well, with a distribution that favors the 5-win mode, but straddles nearly evenly into the 6-win mode. On the other end of the scale are the other three teams: Nebraska, Minnesota and Illinois clustered around the 2-win mode.

Power Rank Results


As previously noted, the PR results are now showing the clearest separation between Northwestern and the other contenders, with sole occupation of the 7-win mode with only a slight lean toward 6 wins. Indeed, the Cats distribution has the highest peak of the bunch, and so has the lowest variance. Iowa is quite consistent through the 3 ratings, and is locked onto the 6-win mode, but shares it with the Badgers, who are leaning more toward 5 wins now. Not to discount Purdue, however. It’s higher-variance distribution still favors the 5-win mode, but leans significantly toward 6 wins.

B1G Expected Final Standings

As mentioned above, following are projections of the divisional conference standings at the end of the season based on the combination of current records and the projected wins and losses for the remainder of the season.

Based on simple point-margin enforced wins and losses, 2 of the 3 ratings forecast Michigan to win the B1GE, with FPI holding that OSU wins out to gain a trip to Indy. The Game looks to be the determining event in the B1GE,  but the real enigma in the mix is OSU. Each of the 3 ratings forecast a different final record for the Buckeyes, ranging from 6-3 (S&P+), to PR’s 7-3 (PR), to 8-1 (FPI).

As for the B1GW, the standings illustrate the incredibly tight race as described above. The S&P+ actually shows a four-way tie at the top, while the FPI and PR both show two-way ties. However, in all three ratings, the forecast calls for the Iowa Hawkeyes to be the team to punch its ticket to Indy.

For Michigan to meet Iowa in the B1GCG would be the perfect piece de resistance to complete Michigan’s Revenge Tour, as Iowa would stand as the last team remaining in the B1G to whom a Harbaugh-coached Michigan team has an unavenged loss. And beyond that, who knows? A potential CFP rematch with the Fig-Things? More revenge! And a final with Alabama? Still more revenge, albeit not for any player on Team 139, but there’s still Greg Mattison!

"FigThing Irish"

S&P+ Forecast

FPI Forecast

Power Rank Forecast

Head-to-Head Win Differences

Now is a good opportunity to to have a look at just which teams have a legitimate chance of advancing as the B1GE representative in the B1GCG in Indy. The following charts illustrate the pairwise chances of various combinations of teams finishing ahead or behind each other in the final divisional standings. Since Michigan is now the favorite across the fancy-stats board more or less, all combinations include the Wolverines. The sidebar charts show the non-head-to-head win probability distributions for the individual team. The center chart is the probability distribution for the difference in non-head-to-head total wins. The +1, even and -1 differences are then subdivided by the head-to-head win probability between to the two teams. The percentages in the text boxes in the center chart indicate the sum-total likelihoods for each team to finish ahead of the other, included the head-to-head matchup.

S&P+ Distributions

Looking at Michigan vs. Ohio State, S&P+ ratings now give Michigan a  75% chance (or 1:3 odds) of finishing ahead of Ohio State, which in all likelihood would lead to a trip to Indy.

2018w09_SP_Michigan-Ohio State_conf_wins_diff_pdf.png

Since all three ratings forecast a tie between Iowa and Northwestern for the B1GW here’s a look at the head-to-head differential probabilities for them. As the chart shows, Northwestern, despite being a heavy underdog in the actual head-to-head game, it still has the narrowest of leads overall. This is the effect of have a half-game advantage, or one win already in the bag compared to Iowa. S&P+ ratings give the Cats a 50.1% chance (or pretty much a toss-up) of finishing ahead of Iowa.


FPI Distributions

Looking at Michigan vs. Ohio State, FPI ratings give Michigan a 64% chance (or about 1:2 odds) of finishing ahead of Ohio State, which would likely include a trip to Indy to close out the B1G season.

2018w09_FPI_Michigan-Ohio State_conf_wins_diff_pdf.png

Here, FPI gives Iowa a very slight edge, 52.5% (still pretty close to a toss-up), in the head-to-head matchup with Northwestern.


Power Rank Distributions

Looking at Michigan vs. Ohio State, PR ratings give Michigan better than a 74% chance (or about 1:3 odds) of finishing ahead of Ohio State. Noice!

2018w09_PR_Michigan-Ohio State_conf_wins_diff_pdf.png

So, in the end, when looking at theses head-to-head distributions, it becomes clear that all three ratings in the aggregate favor Michigan to win the Big Ten East, despite in at least one instance, and underdog in one of its games. Because the chances of OSU losing yet another game before The Game are not insignificant, the possibility of Michigan clinching the B1GE divisional title before the The Game are sizeable.


Overlay Redux

Just to wrap things up with a quick visual summary, here are the combined overlays of the total wins probabilities broken out by division, sized for download and quick-reference on your smarty-pants phone.

S&P+ Results


FPI Results


Power Rank Results


Yours in football, and Go Blue!



November 2nd, 2018 at 2:26 AM ^

Iowa does not control its own destiny in the Big Ten West.  In a three-way tie between Iowa, Northwestern, and Wisconsin, each at 7-2, Northwestern is heavily favored.  Each team would be 1-1 against each other and 5-1 in the division.  It would go all the way down to the fifth tiebreaker, which is the cumulative record of their opponents in the other division.  Northwestern’s opponents are Michigan, MSU, and Rutger.  Wisconsin’s are Michigan, PSU, and Rutger.  Iowa’s are PSU, Indiana, and Maryland.

Let’s use S&P+ win totals for simplicity, except that we’d be locking PSU’s game with Wisconsin in as a Wisconsin win.  NW’s opponents have 13.67 expected wins.  Wisconsin’s would have 13.41.  Iowa’s would have 11.4.

This understates Northwestern’s advantage, because they hold the head-to-head tiebreaker over Wisconsin if Iowa gets eliminated from the tie at this step.

Iowa really would benefit from an additional Wisconsin loss.

The Maizer

November 2nd, 2018 at 9:40 AM ^

This is a great supplement to the OP, thanks. I like this potential situation where playing better cross-division opponents helps a team rather than the usual hindrance. Of course if Iowa plays Rutgers instead of PSU, they're probably in control of their own destiny now.

Ecky Pting

November 2nd, 2018 at 10:47 AM ^

Not so fast, my friend!

I like your approach to scoring the inter-divisional records based on expected wins vs. simple projected wins (I think I'm gonna code it like that), but upon further review - Northwestern wins the S&P+ 4-way tie based on the first 3-or-more-way tiebreaker (I need to check my code now...)? This looks at the record of the tied teams among themselves (including Purdue), and separates out Northwestern and Wisconsin, after which the head-to-head applies:

Records among 4-way tied teams:

Northwestern : 2 - 1 (+Purdue +Wisconsin -Iowa)
Wisconsin:       2 - 1 (+Iowa +Purdue -Northwestern)
Purdue:            1 - 2 (+Iowa -Northwestern -Wisconsin)
Iowa:                1 - 2 (+Northwestern -Wisconsin -Purdue)

Correct me if I'm wrong - my brain has a cramp now.

But as for FPI and Power Rank, Iowa prevails in a simple 2-way tie with NU at the top by the projected win in the head-to-head.


The Maizer

November 2nd, 2018 at 12:07 PM ^

I'm not sure why you're saying "not so fast". I think he's right in that if Iowa and Wisconsin win out and Northwestern beats Illinois and Minnesota, Iowa would unlikely win the tie-breakers in that 3-way tie. Thus, Iowa does not control its destiny.

You're right about the 4-way tie, but if Iowa wins out, Purdue has at least 3 losses and there is no 4-way tie. Seems as though Northwestern is the only team that can claim to control its own destiny.

The Maizer

November 2nd, 2018 at 1:01 PM ^

Ahh, I see. That's different than arguing about Iowa controlling their own destiny, however. His projection included Iowa winning out and only using S&P+ ratings to determine the inter-division records. That's a good way to decide if someone controls their own destiny or not, imo. I understand the framework of your response now though, thanks.


November 2nd, 2018 at 2:04 PM ^

You’re correct.  I was specifically taking issue with your comment that Iowa controls its own destiny.  Since there is a scenario where they win every remaining game and do not win the Big Ten West, they do not control their own destiny.

Northwestern is in excellent shape to win the West; as you point out, they win the four-way tie at 6-3 and are in very good shape in all possible three-way ties at 7-2.