Following Michigan State's 24-17 win over Purdue on Friday night, head Coach Jonathan Smith referred to the game as "a tale of two halves." The fans who brewed an extra cup of coffee and stayed up late to watch the end of the game can certainly confirm that this was far from a scalding hot take.

In the first half against Purdue, Michigan State scored on all four drives and amassed 24 points on 220 yards of offense. Meanwhile, the defense gave up just three points and 134 yards.

But in the second half, the defense surrendered 14 points and 204 yards while the offense was forced to punt on five consecutive drive, the longest of which gained just 18 yards.

In many ways, the Purdue game was a microcosm of the season as a whole. Half of the time, the Spartans look pretty good. In fact, they look like a competitive Big Ten team with significantly improved coaching and schemes.

Examples include the wins against Iowa, at Maryland, and large portions of the games against Boston College, Michigan, the first half against Purdue, and even the early parts of the games against Ohio State and Indiana. The Spartans' performance in these parts of the season give fans hope of a better football future.

But the other half of the time, Michigan State looks suspiciously mediocre. The performance is reminiscent of the late Mark Dantonio era and the non-Kenneth-Walker portion of the Mel Tucker era. The performances in games against Florida Atlantic, Oregon, Illinois, and the later stages of the Purdue, Ohio State, and Indiana contests ring a bell here.

Now, despite already objectively overachieving in year one, half the fanbase seems to be freaking out. OK, that is an exaggeration. But, it is safe to say that a lot of Spartan fans are frustrated by the inconsistency, even if it is is understandable based on the unusual number of injuries, especially in the secondary and on the offensive line.

In any event, the regular season now comes down to just one game. If the Spartans can string together two good halves, there is a very good chance that Michigan State will send the seniors out with a win and in the process earn a valuable bowl bid. A victory next week would taste like sweet cream.

But if the inconsistency continues, Saturday could be the final game of the season for the Green and White. Despite the meager expectations coming into the season, a loss would certainly leave a sour taste in everyone's mouth and leave some fans crying over spilt milk.

Now let's check in on the picks that I made in last week's Bad Betting Advice article, starting with the overview summary shown below in Figure 1. More information about how to read this figure can be found in Week One's edition of Against All Odds.

Eight teams overachieved in Week 13 by creaming the spread by more than 14 points including Michigan, Notre Dame, Louisville, Southern Methodist, Kansas State, and Nebraska. Boise State was the only school that underachieved, yet still won.

There were a total of 22 upsets in Week 13, which was even more than the value of 17.1 my computer predicted. Table 1 below summarizes those upsets and compares them to the picks made last week.

This week's biggest upset based on the opening spread was Georgia State's win over Texas State (-19.5) in the Sun Belt Conference. But, there were several other notable Power Four conference upsets including Oklahoma over Alabama (-13), Florida over Mississippi (-9), Auburn over Texas A&M (-2.5), Kansas over Colorado (-2), and Illinois over Rutgers (-1.5).

My computer went 5-2 (71%) on upset picks which brings the year-to-date performance to 42-47 (47%). The FPI also had a strong performance, going 4-3 (57%) which brings the year-to-date total up to 28-28 (50%).

Table 2 below gives the results of the computers' picks against the opening spread.

There was only one pick on the board this week, based on my analysis of the FPI data, and it was correct. Overall, my algorithm struggled at 26-36 (42%). This brings the year-to-date performances to 30-31 (49%) for suggested bets and 342-331 (51%) overall.

My FPI-based suggested bets are now 23-16 (59%) for the year. The full set of FPI picks went 33-29 (53%). This brings the overall year-to-date performance for the FPI to 351-322 (52%).

Table 3 below gives the results of the point total bets for Week 13.

My point total bets had a good week. The lock picks were the cream of the crop, going a perfect 5-0. The other three picks were all incorrect, however, which resulted in a total performance of 5-3 (63%). This brings the year-to-date totals to 24-22 (52%) for the locks and 96-69 (58%) for the suggested bets.

Following the results of Week 13, I have re-run the full season Monte Carlo simulation using the updated power rankings, including the current uncertainty in those rankings, which provides the opportunity to update the season odds for each team. Table 4 below gives the update for the Big Ten conference and Table 5 shows the updated Big Ten win distribution matrix.

Note that all the rankings listed next to each team refer to my computer's power rankings and not the national polls.