The start of the 2023 Michigan State football season was a bit soft. The Spartan offense was a bit flat. New starting quarterback Noah Kim struggled to move the ball. On defense, Central Michigan drove the ball into the red zone on the Chippewas' opening possession and very nearly scored a touchdown. This caused the opening night crowd in Spartan Stadium to be rather quiet as well.

But slowly, the intensity began to increase. As halftime drew near, the Spartan ensemble seemed more finely tuned. The plays from both the offense and defense grew louder. By the middle of the third quarter, the Spartans were rocking. The game reached its crescendo when sophomore Tyrell Henry made a one-handed grab in the north end zone to extend the Michigan State lead to 24-7.

By the time the game had reached its finale, Michigan State extended the lead to 31-7. A night that could have turned into an elegy instead ended with the Spartans joyously singing "Victory for MSU" in the southeast corner of the end zone.

It was certainly not a flawless performance from the Spartans. There were some sour notes here and there. I personally would have liked to see a more dominant performance by the offensive line. Central Michigan also did not possess a passing attack that could adequately test the Michigan State secondary.

That said, in the later stanzas of the game, we did see some harmony between Noah Kim and the young receiving corps. The defensive front made some loud plays, and special teams was a pleasant surprise. The 47-yard field goal make by Jonathan Kim was music to the ears of the Spartan faithful.

The question now is whether the Spartans can build on the results on the performances in Week One? Will the season continue to build to an exciting climax, or will it peak early and descend into a quiet and mournful dirge? Time will tell if Michigan State has the breath to sustain the current positive note.

For those who are unfamiliar with the way I structure this series, here is a quick primer. Every week during the college football season, I will present my computer's recommended wages in my Bad Betting Advice column. Following the week's action, I will review those picks here and provide some comments on the action across the country.

One figure that I prepare is a comparison of the outcome of each game relative to the opening point spread. That result for Week One of 2023 is shown below in Figure 1.

Each data point represents one game. The position of the data point either above or below the solid diagonal line reveals whether the favored team beat the opening spread or not. The farther a data point is from the center diagonal, the bigger the deviation from the spread.

The two diagonal lines represent one standard deviation from the mean (the opening spread), which is equal to just over 14 points. In other words, in roughly one-third of all football games, the final point differential is two touchdowns away from the opening spread. Favored teams whose data point falls outside of the dotted line either overachieved or underachieved by a significant margin.

Some favored teams underachieve so much that the game ends in an upset. These games are located below the red horizontal line.

This week Figure 1 tells us that California, Miami (FL.), Washington, Central Florida, USC and Oklahoma (off scale and not shown) all played the sweet tune of overachievement. In contrast, there were eight favorites who underachieved so badly they were singing the upset blues. Those teams include Baylor, Clemson, TCU, Texas Tech and Purdue.

In my Bad Betting Advice weekly preview, I highlighted several potentially interesting bets against the money line (upsets), against the spread (ATS) and against the total points (over/under). Now it is time to see how those predictions turned out, starting with the upsets in Table 1.

Both computers (i.e. my computer and ESPN's FPI model) kept a bit quiet in Week One and each only made one upset prediction. Both turned out to be wrong. There were a total of eight upsets in Week One, which was just above the prediction of 6.8 upsets from my weekly simulation.

What is notable is that five of the upsets this week were in games where the point spread was over 10 points (including losses by Texas Tech, Clemson and Boston College), and two other upsets had point spreads over 20 points (Texas State over Baylor and Colorado over TCU). History tells us that there are typically only one-or-two 25-point or worse upsets per season. It is possible that Texas State has already claimed the biggest upset of the year.

Table 2 below summarizes the results of the recommended picks against the spread.

My computer only had one recommended pick for the week, which did not occur (thanks for nothing, Hawkeyes) but the FPI was bolder with eight total picks. Unfortunately for the ESPN model, the FPI only had one pick correct to go 1-7 (12.5%) for the week.

When the against-the-spread picks for all 45 games are considered, my model had a strong 25-20 record (56%), while the FPI struggled at 21-24 (47%).

Finally, Table 3 shows the results of my recommended point-total bets.

In my weekly preview, I was cautious about the number of picks my computer made in Week One, especially with the changes in the running clock rule, which could impact my calculations. However, my point-total picks had a very strong performance. My standard recommended picks went 15-9 (63%) while my "lock" picks were an impressive 4-1 (80%).

Only time will tell if the accuracy of these picks will get louder or softer as the weeks go by.

After a single week of play, we know only slightly more about each team than we did back in the summer. I fed the results of Week One into my computer to generate updated expected win totals, win distribution, and odds for various full season outcomes, as summarized in Tables 4 and 5 for the Big Ten conference.