Week Nine Recap: Everything is AWESOME!

There are some days when a single song sums up a situation very precisely. Saturday was one of those days. The particular song that I am thinking of goes a little something like this:

"Everything is awesome. Everything is cool when you're part of a team. Everything is awesome. When you beat the Wolverines."

At least that is how remember it going. 

As we all saw, in perhaps the biggest game in the storied history of the rivalry, the Michigan State Spartans came from behind to beat the Wolverines by a score of 37-33. The Spartans were surprisingly able to run the ball while stopping the Wolverine rushing attack. Early turnovers and an equally surprisingly competent Wolverine passing attack had the Spartans on the ropes.

But the Spartans bend-but-don't break defense once again forced the Wolverines to mostly kick field goals while the Spartans scored touchdowns. Once the turnovers evened out, it was just enough to secure the victory for the Green and White.

For the Wolverines, this loss must feel a bit like stepping on a Lego. For what feels like the 70th year in a row, the Michigan faithful had convinced themselves that this was the year when "the special" things were going to start happening for their program. This was supposed to be the year when the recruiting rankings, the over-paid coaches, and the ugly helmet were finally going to translate into wins over the good teams, and not just all the mediocre ones on the schedule. 

This was supposed to be the year when the older brother put the younger brother into his place and sent him back down to the basement to play with Duplos. It was all foretold back in the Spring like some ancient prophesy and repeated in the Ann Arbor echo chamber until they all started to believe it.

Instead, for the tenth time in 14 tries, the Spartans gave the Wolverines the business.

Going forward, it will be interesting to see what happens with Wolverine Coach Jim Harbaugh. Back in the summer, everyone was talking that Harbaugh might soon lose his job and have to look for a new opportunity. (Perhaps that would give him more time for his awesome community). That talk died down as the wins accumulated. I expect it to return a roar again soon.

As for Coach Mel Tucker and the Spartans, side by side it feels like they are going to win forever and party forever (or at least until next weekend). It's awesome to win, and it's awesome when the Wolverines lose. It is even more awesome when it happens together.

Week Nine Results

As is my weekly tradition, Figure 1 below shows the results of all 53 games involving two FBS teams in Week Nine.

Figure 1: Results of Week Nine showing the actual point differentials relative to the opening spread.

The teams that were awesome and overachieved this week include Oklahoma State, Louisiana, UCF, Wake Forest, Minnesota, Wisconsin, Utah, Marshall, BYU, and Kansas State. There were not again games that crossed the dotted line over into underachievement, but USC came very close to not being awesome.

Table 1 below summarizes the 13 games that ended in an upset relative to the opening Vegas line and in comparison to the computer picks.

Table 1: Upsets based on the opening Vegas line compared to the upset projections from Week Nine

Week Nine featured a total of six upsets where the line was a touchdown or greater. These include Washington State (+16) over Arizona State, Miami (+10) over Pitt, Purdue (+7) over Nebraska, and West Virginia (+7) over Iowa State. Naturally, the Spartans' (+2) awesome win over the Wolverines is also on the list. 

As for the upset picks this week, they were a bit more awesome than usual. My metric got two-thirds of the picks correct and the FPI was over 40 percent. Year-to-date, the both my algorithm (36 percent) and the FPI (34 percent) are lagging historical performance, but this weekend helped.

Now if you really want to see awesome performance, check out Table 2 below which gives the results of the computers picks against the opening spread (ATS).

Table 3: Results of the highlighted picks versus the spread in Week Nine

Things were not so awesome for the FPI which swung and missed on the only pick of the week. However, my algorithm was correct on five of the total six picks. This brings my computer's performance up 63 percent for the year. The FPI has sunk a bit more below .500 through nine week, and the overall performance where I combine the picks from both models sits at 54 percent year-to-date.

For Week Nine as a whole, everything was once again pretty awesome from my algorithm. It went 35-18 (66 percent) against the opening line, even though it got the Michigan / Michigan State game wrong. For the year, my computer is still a less-then-awesome 226-248 (48 percent) but it is making progress. 

Much like the Wolverines, thing were significantly less awesome for ESPN's FPI. That computer was only 22-31 (42 percent) and overall is now down to 223-251 (47 percent). I have heard rumor that FPI software engineers think that the officials are simply conspiring against them.

Updated Big Ten Odds and Expected Wins

Following the results of Week Nine, I have re-run the full-season Monte Carlo simulation using the updated power rankings, including the current uncertainty in those rankings, in order to update the season odds for each team. Table 3 below gives the update for the Big Ten conference.

Table 3: Updated Big Ten rankings, expected wins, strengths of schedule and season odds.

Even outside of the state of Michigan, it was a pretty big weekend in the B1G. Let's start in the West Division where Wisconsin, Minnesota, and Purdue all scored key victories. The Golden Gophers of Minnesota are now alone in first place with the Badgers, Boilermakers, and the fading Hawkeyes a game back in the loss column.  

My computer has a lot of respect for Wisconsin and as a result now gives the Badgers the slight edge in the division race with 42 percent odds. The Badgers have wins in hand against Purdue and Iowa and a reasonably manageable schedule down the stretch (with games at Rutgers, versus Northwestern, and versus Nebraska). There is a good chance that Wisconsin will be traveling to Minneapolis on Thanksgiving weekend for a winner-take-all showdown with the Gophers.

Minnesota's odds are just a tick lower at 40 percent, despite being a game ahead in the standings. If Wisconsin were to stumble (for example against Nebraska) and if the Gophers can run the table (by winning versus Illinois, at Iowa, and at Indiana) the final weekend won't matter and Minnesota will make their first trip to Indianapolis. But, all signs are pointing a Thanksgiving showdown where my computer projects that the Badgers (-3) will be a slight favorite.

With wins by both Michigan State and Ohio State this weekend, the battle for the East is basically down to those two teams and Michigan, and the odds are surprisingly close. My computer has finally come around to the idea that the Spartans are legitimate contenders, and I actually project MSU with slightly better overall odds than Ohio State by two tenths of a percentage point (36.7 percent versus 36.5 percent).

While our eyes are telling us that Ohio State is still likely the best team of the three, my computer still projects that the Wolverines should be favored on a neutral field against the other two (despite the result in East Lansing). But, the Spartans are simply refusing to lose. 

Penn State is essentially out of the race, but they can still play the role of spoiler for the Spartans and/or the Wolverines. Ohio State and Michigan State also need to avoid stepping on a Lego in one of the upcoming sneaky road games including next week when MSU travels to Purdue and Ohio State head to Nebraska. There is still plenty of opportunity for drama in the Big Ten East race.

The updated Big Ten win distribution matrix is shown below in Table 4.

Table 4: Updated Big Ten win distribution matrix after Week Nine

As expected, the big win over the Wolverines game the Spartan's expected win total a nice boost up to 10.31. More precisely, Table 4 gives MSU a 78 percent chance to get to at least double digit wins, a 43 percent change to get to 11-1 and a 12 percent chance to run the table in the regular season.

A closer look at MSU's remaining schedule

Figure 2 below gives my updated projections for the Vegas lines and win probabilities for all of the games remaining on MSU's schedule. These numbers are generated based on the results of the full season Monte Carlo simulation.

Figure 2: Updates projections for the Vegas line and win probabilities for MSU's remaining schedule.

With the season now in full swing, there is a lot more accumulated data and the uncertainty in that data is decreasing. As a result, these projections have not changed much over the past few weeks. As of today, MSU is expected to be a big favorite over Maryland, a narrow favorite at Purdue and versus Penn State, and an underdog at Columbus.

Next week's contest in West Lafayette is the very definition of a trap game. The Spartans opened as a 2.5-point favorite, and my computer projects a virtual toss up. So far this year, Coach Tucker has been very good at keeping his team focused on the task at hand. The Spartans have both awesome momentum and an awesome opportunity at hand. In order to make last weekend's win really count, the Green and White need to take care of business over the next two week.

Either way, the Spartans are well positioned for a very nice bowl game, but a pair of wins in the next two weeks would bring the record to 10-0 and place the Spartans almost certainly in the playoff picture. In fact, I would be mildly surprised if the Spartans are not in the top four when the first playoff poll is released on Tuesday evening.

Getting a win in Columbus is going to be tough, if MSU were to win that game and make it to 11-0, they would clinch the Big Ten East title before the Thanksgiving Turkey was even on the table. In the more likely scenario where the Spartans lose in Columbus, the finale against the Nittany Lions would more likely be for a spot in the Rose Bowl. 

Based on the strength of MSU's schedule, a 10-2 record would very likely secure the Spartans a spot in the New Year's Six in either the Fiesta Bowl or the Peach Bowl. The Rose Bowl would also still be in play in this scenario, depending on who the rest of the conference shakes out.

National Overview

Finally, let's take a quick look around the country at the rest of the action. Once again, the updated conference odds can be viewed by clicking on the hyperlink attached to each conference name. 

In the ACC this weekend, Wake Forest rolled Duke to run their record to 8-0 and their Atlantic Division title odds to 75 percent. However, the Demon Deacon close the season with three of the final games on the road, including a trip to Clemson and to North Carolina. Furthermore, the lone home game left on the schedule is against NC State (23 percent division odds) who still control their own destiny.

In the Coastal Division, Pitt was upset by Miami this week and as a result, the Panthers' division odds have dipped to 67 percent. Pitt is now just a game ahead in the loss column of Virginia (21 percent odds) and Miami (eight percent), with a home game against the Cavaliers still on the schedule.

The Big 12 is starting to look more like a three-team race for two spots in the Big 12 Championship game. Baylor (51 percent odds) is one of those teams and Texas is not following the Bears win over the Longhorns this weekend. Iowa State (29 percent) also seems to be fading following an upset loss at West Virginia. Joining Baylor at the top are the two schools from Oklahoma. 

The Sooners (47 percent) are at the top of the standing, but are third in odds, due in large part to a brutal end to the schedule. After next week's bye, Oklahoma plays at Baylor, versus Iowa State, and at Oklahoma State. In contrast, the Cowboys (71 percent, best in the conference) close at West Virginia, versus TCU, at Texas Tech, and versus Oklahoma.

In the Pac-12 North, Oregon (38 percent) is a game ahead of the pack in the standings, and a new potential challenger has risen up my odds table. Washington State (24 percent) upset Arizona State and now claim the second best odds in the division, just ahead of Washington (21 percent). Oregon will face the two schools from the state of Washington over the next two weeks.

In the Pac-12 South, Utah scored a big win over UCLA this weekend and now has a commanding lead with a 95 percent chance to advance to the Pac-12 Championship Game. The Utes are a game ahead in the standing and already have head-to-head wins over the next three teams in the standings: Arizona State, UCLA, and USC.

In the SEC East, the race is now officially over. Georgia drank Florida under the table in the World's Largest Outdoor Cocktail Party and Kentucky got upset at Mississippi State. The Bulldogs cannot be caught. 

In the SEC West, Auburn's win over Ole Miss effectively makes the division a three-team race with Alabama (72 percent) still heavily favored. The winner of next week's contest between Auburn (20 percent division odds) and Texas A&M (six percent division odds) can steal the crown from Bama if Auburn can also upset the Tide in the Iron Bowl.

As for the Playoff race, my algorithm still projects Georgia as a virtual shoe-in with 96 percent odds, but the rest of the top four include Wake Forest (47 percent), Notre Dame (37 percent), and oddly Michigan also at 37 percent). The Spartans' odds are a tick below at 29 percent, which is still good enough for fifth place, nationally.

That said, I ran a few new numbers this week which give perhaps a more realistic projection of what the initial playoff rankings might look like. I constructed a new, simple metric to measure actual wins relative to strength of schedule. It is a performance-based and not a prediction-based metric that counts wins either above or below the expected value of wins that an average Power Five team should amass with the same schedule as the team in question. Those results are shown below in Table 5

Figure 5: Rankings based on actual wins relative to the each teams strength of schedule (defined as the expected number of wins an average P5 team who have with the same schedule, year-to-date) 

This metric places undefeated Georgia, Michigan State, and Oklahoma in the top three spots, with Notre Dame in fourth place.

In notable Group of Five action, the Bearcats of Cincinnati continued to roll with a win over Tulane and continue to reside at the top of my Group of Five New Years' Six leaderboard with almost a 60 percent chance to claim that slot. In order to claim that spot, however, the Bearcats will need to win the ACC Championship game and they will almost certainly face Houston in that contest, thanks to the Cougars win over SMU this weekend.

The Group of Five team with the second best NY6 odds is Texas San Antonio (8-0), who was idle this week, but who still have the best conference odds in Conference USA at 36 percent. 

In the Mountain West, Fresno State knocked San Diego State from the ranks of the unbeaten and now appear to be the new conference favorites with 48 percent odds.

In Sun Belt action, Louisiana (99.5 percent division odds) appears to still be on a collision course with Appalachian State (72 percent division odds). In the MAC, Miami of Ohio (69 percent) and Northern Illinois (59 percent) are still the projected division winners.

Against all odds, I have reached the end for this week. I hope that this week bring you nothing but awesomeness. Until next time, enjoy, and Go State, beat the Boilers!