2023 Week 1 Preview: Believe

The past few years have been interesting for the Michigan State Spartans. The end of the 2021 season was marked by the celebration of a championship at the Peach Bowl in Atlanta.

But the 2022 season was tough. At times it felt like the Spartans had been transported to a whole new continent and were playing a sport that was called football, except that it wasn’t actually football. (I mean seriously, what was that in the Minnesota game?). The end of the season saw the Spartans relegated to the ranks of the teams with no place to go for the holidays.

Now, the year is 2023 and a new season is upon us. This is a big year for head coach Mel Tucker and his staff in East Lansing. If Spartan fans are generous enough to grant Tucker a mulligan for the COVID-impacted 2020 season, that still leaves one very good season and one mediocre to bad season on the resume. The 2023 campaign will give a strong indicator as to which season was the fluke.

This fork in the road presents Spartan fans with a bit of a dilemma. Michigan State fans are passionate. They care, not unlike folks at a hip-hop concert whose hands are not in the air. But that passion can also go two different ways.

Option one is to be a pessimist and to expect the worst. Some say that it is the hope that kills you. It is a better strategy to simply not get your hopes up.

As a eternal optimist, I offer to those Spartan fans one simple word of advice:

Believe.

For some MSU fans, this will be a challenge, and taking on a challenge is a lot like riding a horse. If you're comfortable while you're doing it, you're probably doing it wrong.

I think it's the lack of hope that comes and gets you. I believe in hope. I believe in belief. Sometimes I even believe in miracles, and so should Michigan State fans. After all, we have witnessed a few.

We may find that as the season goes on that my belief is misplaced. But for now, I believe that that the Spartans can get better, or perhaps that they will get better. I know that they believe in themselves and that they believe in one another. I believe that locker room smells like potential.

If the Spartans can hold on to that belief. If they can truly do that, then maybe, just maybe, this season will turn out OK. Somehow, I believe that it will. That all starts now. Yeah?

Thank you for attending my Ted (Lasso) Talk.

MSU Prediction

The Michigan State Spartans open the season on Friday night with the home opener against the Central Michigan Chippewas. At the time of press, the Spartans are favored by 14.5 points with a point total over/under of 45.5. Based on this spread, there is a 85% chance that the Spartans will win straight up, based on the historical accuracy of the Vegas line. 

While that generally sounds good, is also means that there is about a 1-in-6 chance for the season to begin with a catastrophic loss to a MAC team. While I would like to believe that this will not happen, history says that it is possible. Central Michigan owns three wins total over the Spartans, including the first two meeting in 1991 and 1992. That total is equal to the tally from the rest of the Mid-American Conference combined against the Spartans.

The Chippewas also bested the Spartans in 2009 thanks to an onside-kick, a very badly timed off-sides penalty, and a game-winner field goal as time expired. Going back to 2001, Michigan State is only 2-5 against the spread against Central Michigan. Perhaps this was not the best year to invite the Chippewas back to East Lansing.

But, let's put aside all that "stinkin' thinkin'"

The computer models are generally more optimistic about the Spartans' chances. ESPN's Football Power Index predicts a 19-point win for Michigan State. My computer model likes a 20-point win, and Bill Connelly's SP+ has the Spartans by 23 points Those are all encouraging numbers, I believe.

Specifically, my computer gives me a final score of Michigan State 38, Central Michigan 18. As such, my analysis suggests that the Spartans will cover the spread and hit the over on points.

Big Ten Overview

Table 1 below gives a summary of the action in the Big Ten in Week One. This table summarizes my predicted score for each Big Ten game, the opening Vegas line, the line projected by both my algorithm and ESPN's FPI, my computer's project point total (which for Week One is simply the historically average number of points scored per game in college football), and my recommended picks against the spread and straight-up (more on this later).

Table 1: Summary of Big Ten action in Week One, including my algorithms' projected scores. The teams shaded in green are projected to cover by my algorithm. The teams shaded red are projected to cover by the FPI.

All 14 Big Ten teams are in action this week. Six of the conference teams are opening the season with a conference game. Maryland is facing an FCS team (Towson), Penn State is facing a Power Five team (West Virginia), while the remaining six teams are squaring off against Group of Five opponents.

Ohio State opens the season at Indiana (+27.5) in a game that should be a blowout. The other two conference games should be much closer. 

The marquee conference match-up is an early season Big Ten West showdown as Nebraska visits Minnesota (-7.5) to open conference play and the Matt Rhule era in Lincoln on Thursday night. The Golden Gophers are a solid favorite, but there is a 30% chance of an upset in Minneapolis.

The other conference game features two teams near the bottom of their respective divisions as Northwestern travels to Rutgers (-4). The FPI has issued an upset alert for the Scarlet Knights, but that seems like a stretch based on the situation in Evanston this offseason.

In the three Big Ten conference games this week, my computer and the FPI disagree completely on which teams will cover. My machine likes Rutgers, Minnesota, and Indiana, while the FPI is picking Northwestern, Nebraska, and Ohio State.

As for the other seven games, five of them are expected to be double digit games, including West Virgina at Penn State (-17.5). The computers general like the Big Ten teams to cover, with the exception of Buffalo at Wisconsin (-23) (where the FPI picks Buffalo to cover) and East Carolina at Michigan (-36) where both computers feel that spread is just a little too high.

The two remaining games could get dicey for the Big Ten team. Illinois is only favored by nine points at home against the Toledo Rockets. If the visitors from the MAC can steal a victory, it could lead to a 12-0 season and a berth in the New Year's Six. The computer like the Illini to cover.

Meanwhile, in West Lafayette, Purdue is a mere six-point favorite over the visiting Fresno State Bulldogs, and my computer thinks the game will be closer than that. A lose to a Mountain West team would not be the start that new coach Ryan Walters is hoping for.

Other Notable National Action

With the exception of the Naval Academy, every FBS team is in action in Week One. A total of 42 teams will open the season with an FCS opponent. As additional 27 games have opening spreads over 14 points. As a result, there are only a handful of games that are likely to have an impact on the greater college football landscape.

Table 2 below gives a summary of the key action taking place across the country to open the college football season.

Table 2: Summary of other notable action in Week One, including my algorithms' projected scores.

The most notable games are the four at the top of the table. The biggest game of the weekend is clearly No. 5 LSU (-2) versus No. 8 Florida State on Sunday evening in Orlando. Both teams believe that they can be playoff contenders. The winner gets a boost, while the loser will be behind the eight ball. Vegas and the computers give the Tigers a slight edge, but my computer picks the Seminoles on their semi-home turf in my preseason analysis.

No. 21 North Carolina and South Carolina is a toss-up according to Vegas, but the computers like the Tar Heels. North Carolina is a potential ACC dark horse, and they would benefit greatly from a win over the Gamecocks. The Florida Gators' visit to No. 14 Utah (-9) will also provide an opportunity for one of those teams to score an early season resume-builder. The computers like the Utes to win and cover.

No. 9 Clemson opens the season with a potentially sneaky road trip to face Duke (+11.5). The FPI is confident in a Tiger blowout, while my computer suggests that the Blue Devils will cover the spread at home.

I also have my eye on a quartet of games involving Pac-12 schools facing potentially tricky Group of Five Opponents. UCLA, Washington, Oregon State, and Washington State are all roughly 15-point favorites against Coastal Carolina, Boise State, San Jose State, and Colorado State, respectively, but something tells one at least one of those teams could find themselves in trouble.  The FPI data suggests the Washington should be careful with Boise State, while my computer sees Colorado State pushing Washington State, especially in Colorado Springs.

Finally, there are two games involving Group of Five teams that could ultimately impact the Group of Five participant in the New Year's Six. Houston has moved to the Big 12, but the Cougars are only a 2.5-point favorite over Texas San Antonio, now of the AAC. My computer likes UTSA in the upset.

Meanwhile Texas San Antonio's AAC rivals, Tulane, has a tricky opener against Sun Belt foe South Alabama (+7). There is a very real chance that the Green Wave could go undefeated in the regular season, but they will need to pass this early season test in order to get there.

Picks of the Week

Every week in this space, I plan to share the most current calculations from my computer in regards to the my projected margins of victory in every game involving two FBS teams. Over the years, I have developed a method that allow me to use the result of my calculations to make educated recommendations for upset picks, picks against the spread, and more recently, point total (over/under) picks.

Figure 1 below gives an overview summary of my computers projections for the week in comparison to the opening Vegas spread.

Figure 1: Comparison of the projected final margins of victory based on my algorithm to the opening Vegas lines for Week One.

This figure contains a lot of information. Each data point represents a single game where the Vegas spread is on the x-coordinate and the margin of victory predicted by my algorithm in on the y-coordinate. At first glance it is striking how much agreement there is in the two data sets, as the points generally cluster around the solid diagonal, which represents the situation where my algorithm perfectly agrees with the opening point spread.

Any data point that lies above the solid diagonal line represents a game were my algorithm suggests that the favored team will cover, and vice versa. There is also a pair of dotted diagonal lines both above and below the solid line. 

I have tracked the performance of my algorithm over a 14 year span. In that time, I determined that when the prediction of my algorithm differs from the point spread by more the 12 points, my algorithm can correct predict the team which will cover the spread a significant percentage of the time. 

The dotted lines represent this threshold. When a data point falls outside of these dotted line, my algorithm triggers a "recommended bet" for that game. Over the past six years, this method has correctly picked the winner against the spread (ATS) 53.8% of the time (276-237). 

Figure 1 also features a vertical red line on the left side of the figure corresponding to a spread of zero. Any data point that falls to the left of that line indicates a game where my algorithm predicts an upset relative to the opening spread. Historically, my algorithm correctly predicts an update 42.4% of the time, which is notably better than a random number generator experiment that I ran several years ago (36%).

With this background in mind there is only one data point that lies outside of the dotted lines (Iowa over Utah State) and one data point that lies to the left of the red vertical line (Texas San Antonio over Houston). Therefore, these are the only two recommendations from my algorithm this week.

As a comparison, I conduct the same analysis using data from ESPN's FPI. Figure 2 below is the result of this analysis for Week One.

Figure 2: Comparison of the projected final margins of victory based on ESPN's FPI to the opening Vegas lines for Week One.

My analysis of FPI data over a similar time period led to the conclusion that a deviation of only six points between the FPI and the Vegas spread is enough to trigger a recommend bet. So Figure 1 and Figure 2 can be interpreted the same way, but here the dotted lines are close to the central, solid and dotted line.

While Figures 1 and 2 give a nice visual summary of the full week, it is difficult to spot all of the recommended bets in both figure. So, I will summarize those in the table below, starting with the upset picks In Table 3.

Table 3: Summary of upset picks for Week One.

As stated above, there are only two upset picks total this week: Texas San Antonio to beat Houston and Northwestern to beat Rutgers. I will also note that each week I perform a simple simulation on the total number of upsets that are likely to occur. The results for this week suggest that we will see a total of X plus-or-minus Y upsets.

Table 4 below summarizes the recommended bets ATS for Week One, based on Figures 1 and 2.

Table 4: Picks against the opening spread for Week One. The picks are listed in order of confidence score.

My algorithm only suggests a bet on Iowa, but the FPI suggests an additional eight picks against the spread including bets on UMASS, Virginia, East Carolina, Clemson, and South Florida all to cover.

Finally, last year I started also looking at the predictions made using Bill Connelly's SP+ system. In this case, however, I am focused only on point total bets (over/under). I once again developed a system to make a series of recommended bet and in this case there are two tier. Each week I will generate basic recommendations and a smaller group of picks (called "locks") at a higher confidence level.

I only have one year of data, but last year the recommended picks were 105-84 (56 percent) for the recommended picks and 58-33 (64 percent) for the locks. Table 5 below gives my point total picks for Week 1.

Table 5: Recommended total points bets (over/under) for Week One. Note that the picks labeled as "locks" are at a higher level of confidence.

My computer may be going slightly overboard with picks this week, as over half of the games involving two FBS teams are on the this. The new running clock rule may be impacting the Vegas lines, which is why there over so many "overs" on the list. For this week, it might be safest to stick with the five "locks" (two of which are on "unders") or to just sit this one out entirely.

That is all the advice that I have for you this week. Thank you for attending my Ted (Lasso) talk. Until next time, enjoy the games.