Triple Option

After listening to hours of college football preview podcasts (I tagged all my “sources” whom I listened to frequently to consider different opinions), I have a couple of confident winners for season win totals, but I also have taken a different approach to the win total market than what is traditionally applied. The traditional approaches include going through the schedule game by game and projecting wins and losses, or to look at returning and lost production. These methods are highly erroneous, and I wanted something more based in finding value with teams that can drastically outperform their number, hence the graph and the “differential method.”

37% of teams finish within 1 win above or below their projected output, 65% finished within 2 wins of their projected output, 88% finished within 3 wins above or below their projected output.

The Differential Method

For lack of a more original title, I will just borrow this phrase, because it is exactly what it sounds like. The graph above features the calculation: 2021 actual win total- 2021 preseason win total. In other words, the change in bookmakers expectations. The above graph shows that the majority of team finish +/- 2 wins of their projection. Taking the bell curve shift, I expect this data set to behave like a standard distribution.

Practically, every team has a 4 game range they can be expected to finish in. The 2021 data has this percentage around two-thirds. I believe this number will be higher because the 2020 COVID season skewed all the projections last year, so it was harder to forecast teams in the preseason. My hypothesis is that for 2022, 75% of teams will finish within that +/- 2 game range.

So why does this matter in a practical setting? Well, if two-thirds of teams are going to be right around their win total, there is not much value in picking a side in the preseason. Take the most middle of the road team, maybe an Oregon State 6-6. From what I have just stated, they can be expected to win four games and lose four games, but the other four could go either way. Even if I have a strong lean towards an over, they can get unlucky. This is why the schedule method is flawed.

Considering the Oregon State case, using FPI win expectancies for probability, I will arbitrarily set cut-offs at less than 35% as a loss and more than 65% as a win. This gives them wins against Montana State, Washington State, Colorado, and California. Losses are Utah, Washington, Arizona State, and Oregon. So they have four toss up games: Boise State, Fresno State, USC, Stanford. I have no way of projecting turnover timeliness, coaching mistakes, weather implications, injuries, or any of the luck factors. If best case, they get lucky and go 3-1, they still only go over the total by one.

Therefore, my method is an attempt to identify “outlier” teams that can finish in the minority third of teams who significantly over-perform and under-perform. 

Liquidity Considerations 

I have a limited amount of money to “invest” into college football season win total futures.  I have stated my theory that the long term profit opportunity comes with identifying teams that finish significantly above or below their projected output. From a practical standpoint, I want to avoid situations where I have to sweat out a result until the last week of the season, purely because that money is tied up. 

If I identify a team with a high push potential but have a lean one way or another, the smart path would be to wait and see if the team performs relative to expectations. Case in point: Oregon State. I am better off just playing my lean in one of those toss ups rather than open myself up to the risk anything goes wrong in the schedule. 

On Cover 3’s ACC Atlantic Preview Danny Kanell applied this method to Louisville.  He believes Louisville will be good and go over the total, but his advised play is just to take Louisville -3.5 against Syracuse in Week 1, because if they win that the schedule opens up for a good likelihood that they go over. So instead of waiting eleven more games, take the money now. As much as I want to, I do not have to take every preseason over. In some cases, that money is better invested week to week. Compare to day trading volatile stocks versus buying and holding methods. Both work sometimes, and there are advantages and times to do either. But one ties up money for too long, while the other makes everything sporadic. 

So with that in mind, I want to outline my win total projections as part of an investment plan to monitor (as I will with tracking all my picks as part of the season long thought experiment) by identifying a few key factors and examples of teams that are poised to go significantly over or under their projection.  

System Opportunities

A system when it comes to betting is a universal approach to every scenario that applies to a set of events. This could be something as simple as “play every favorite on the money line” (or vice versa), or “play every game to go over for the week.” Because the sports world operates with a high degree of randomness, usually the simple ones do not work. For instance with the over system, if over the course of an x amount of season, oddsmakers determine that 55% of games will go over, they will either raise game totals overall to skew that number back to 50%, so make adjust the payouts so that overs pay out less in the aggregate than unders do, so the money split comes out at 50-50. 

Any good handicapper or aspiring analyst like myself is essentially looking for systems that return a positive profit over the long-run. Let us then apply the system logic to the season win totals. Last year, 62 teams went over their projections, 62 teams went under their projections, and 6 teams pushed. At least for a one year sample size, there is no system opportunity in playing every over or under. I only have tracked data for 2021 to this point, but I would expect if I evaluate over a five year sample, the splits would be around 50/50. Logically, this makes sense too because of the interaction in every team’s schedules that factor in making the numbers. Simply put, if one team goes over, they have won a game they were not projected to, so another team is on the other side of that who loses the game. Take one instance from 2021: True college football fans will remember the Friday night matchup in late October between Tulsa and Navy, in which Navy upset Tulsa infamously without throwing a single pass. Navy’s preseason win total was 3.5, and they ended up 4-8, hitting the over. Tulsa on the other hand was projected for 6.5 wins, and finished 6-6. In the alternate universe where everything remains constant except for Tulsa winning this football game, Tulsa goes over and Navy goes under, and there is no net change with the total amount of overs and unders. Think of every game as having a balancing act like a seesaw. Even in another example if Army were to take down Navy and go over their 8 projection instead of pushing, the splits are 63-62-5, not enough to significantly conclude that taking everyone’s over is profitable. 

Therefore, something more detailed is necessary, so I will propose certain systems that revolve around outlier factors for season totals. Some of which I will use to bet, some of them are merely to monitor and build a data collection. For reference, the 2021 outliers (+/- 3 wins from projection) were: For overachievers: Utah State, Michigan State, UTEP, San Diego State, Northern Illinois, Baylor, Wake Forest, Michigan, Oklahoma State, NC State, Pitt, Purdue, Houston, Fresno State, UTSA. For underachievers: Indiana, Washington, USC, North Carolina, FIU, Ohio, Florida, Northwestern, Nebraska, Tulane, Buffalo.

There certainly is not a greatest common denominator, but there are a few commonalities. I have five proposed systems: Championship contenders, overreaction, low numbers, new coaches, bad luck last season.

System #1: Play Alabama, Ohio State, and Georgia to go Over 11.5 at plus money

You make this bet because this is an instance where you want to tie your money up for a good number. Neither of these three will be an underdog this season, and the and game will be a huge number which I want nothing to do with. All three of them? Probably not, but that’s why you take this number now and hedge it later.

Unless Michigan State impresses me, the only time I see Ohio State losing is maybe to Michigan, so in my mind this is an Ohio State money line play on The Big Game that is better value than I will get that week. And if I wager Michigan ML that week as well, I guarantee a profit. Georgia I do not see a loss so I will just wait and see if Florida or one of Mississippi State/Kentucky can catch them in a tough spot in the back to back road games in November. Same deal with Alabama when they visit LSU and Ole Miss in consecutive weeks. As long as none of them go worse than 11-1, there is almost no way I can mess this one up. 

System #2: Overreaction; just play the opposite of whatever those outlier teams did last year

Simple enough. The teams that had the biggest change in wins last year have the biggest change in expected wins, so why not take advantage of a market overreaction to a fluke year? I used this for the North Carolina teams. There is a lot of hype around NC State, and I think they’ll be good, but finishing 9-3 in consecutive seasons? I’m skeptical. North Carolina was the hyped team last year, and I think there’s a good chance these two’s 2021 seasons switch for 2022. Best worst case? In their rivalry week matchup, NC State is 8-3 and UNC is 7-4, and I need a Tar Heel win for both to hit. In that case I just hedge NC State. 

System #3: Overs on teams with really low numbers

Most of the teams with totals from 1-3.5 had really bad seasons a year ago, and for the ones that did not make a coaching change, they are probably headed in the right direction. Utah State and Northern Illinois were two of the worst teams in the COVID year: Both hit their over by mid October, and ended up winning their conferences. Especially in the group of 5, where there is plenty of parity, not many teams will stay terrible forever. 

I am using this logic to play Vanderbilt, Akron, and Charlotte on the over, with Kansas, Louisiana-Monroe, and Arizona worthy of consideration as well. In my mind, last season was so bad that it almost has to bee better. And if 3-9 cashes an over? I will take my chances. Most teams play one FCS opponent, which should be one win. Clark Lea wants Vanderbilt to be “the best program in the country.” For now, a one game improvement works for me. Take care of Hawaii, Elon, and NIU and this cashes by the end of September. Akron hired an established offensive coach, and in one of the worst FBS divisions from a year ago, they have four winnable games on the schedule. Charlotte has been trending the wrong way, but they lose two of their tougher opponents in Marshall and Old Dominion in the Sun Belt, and avoid UTSA. They can find 5 wins sprinkled in. They quit in their final two games last season, which hopefully is a wake up call for the guys who stuck around. 

System #4: Hope for Chaos in the Coaching Changes

This was such an unprecedented carousel that I am split on it being a good one to find profit in or a bad one. From an odds-making perspective, more continuity in a coaching staff would seem to bring more consistency and less season to season variance. For this reason, it would be smart to stay away from teams with longer tenured head coaches (arbitrarily say, more than 5 years) because those teams are bound to get lucky or unlucky, and any preseason inklings are probably useless. This gives more of a stayaway strategy than anything. So for this reason, I would rule out of any projections: Clemson, Wake Forest, Pitt, NC State, Ohio State, Michigan, Wisconsin, Iowa, Penn State, Northwestern, Indiana, Oklahoma State, Iowa State, Utah, Stanford, Alabama, Georgia, Cincinnati, Navy, Army, Air Force, Wyoming, Tulsa, Tulane, Middle Tennessee, North Texas, Miami OH, Ball State, Eastern Michigan, Western Michigan, and Toledo. One of those teams will probably have their best year in a decade. A handful will probably make coaching changes. To be within 2 games above or below their projections is not worth sweating out a lose total or hoping for a dumpster fire.

I caution against assuming too much with newer hires. We can easily apply the principle to Utah State as they hired an established head coach in Blake Anderson, but so did Auburn, who went the wrong way under Brian Harsin. The benchmark I would use here is to bet on established head coaches at their previous school. That means over on: LSU, Florida, Miami, USC, Washington, TCU, Colorado State, Akron, Georgia Southern I lean under on coaches with coordinator experience, as there is more of a learning curve: Oregon, Oklahoma, Texas Tech, Virginia Tech, Virginia, Duke, Nevada, Hawaii, Temple, Troy.

System #5: Support teams who were unlucky or underperformed relative to their tradition

Michigan State and Baylor have low key been top ten programs in the past decade, with a few down seasons (and ugly scandals) obscuring the fact that they both have a lot of very good seasons. 

The five “close-loss, bad turnover luck” this category applies to are Nebraska (obviously), Iowa State, Mississippi State, Florida State, and South Alabama. Nebraska is a huge risk, but those other four were in the -2 deviation where none of the swing games go their way. The good programs who had down years: Clemson, Penn State, UCF, Boise State. I lean over on all those teams (only played Miss State).

So who is bound to come back to earth? One possibility is to go under on the Group of 5 champions because of the greater level of parity. So Cincinnati, Utah State, Louisiana, NIU, and UTSA. Otherwise, Minnesota finished 8-4 despite losing to Bowling Green and not passing the ball. I’ve been on the Coach Fleck bandwagon for a while, and this team is just too hyped up for my liking. The Big Ten West is a huge question mark this year, and Purdue could also be prone to a drop off, while Nebraska and Illinois might steal some wins. UCLA I believe benefitted from a really down year in the Pac 12, which I believe skewed the number a bit as I expect Cal, Stanford, Arizona, and Washington to all be better. 8 is too high of a number for a team not names Utah, USC, or Oregon.

Conclusion

I played 13 preseason win totals (about 10% of FBS teams), which I tweeted as official. Obviously I want to go above 50%, and at the end of the season I will track that with the various system techniques, as well as the net unit increase and then the overall differentials, with on average how may games over/under the selected team went compared to the average. 

Until then, expect weekly picks and write ups as I preview as fit. My on the record/off the book predictions are as follows. Playoff: Ohio State, Alabama, Utah, Georgia. Champion: Ohio State. Heisman: CJ Stroud, Will Anderson and Bryce Young and Duece Vaughn as finalist. Group of 5 champions: UCF, Boise State, Coastal Carolina, Miami OH, Florida Atlantic.