We last ran our Cal football season predictions following spring practice [programming note: our Pac-12-wide predictions will go up next week—if you haven’t yet participated, get your votes in by tomorrow!]. Back then, it seemed like the Bears’ biggest questions may be settled by the end of fall camp: some semblance of a functional passing game may emerge and we would build some depth into one of the best Cal defenses many of us have ever seen. Since then, Brandon McIlwain left the team to focus on baseball, Devon Modster’s eligibility became doubtful, and defensive tackles Aaron Maldonado and Siulagispai Fuimaono have gone missing (has any checked to see if they’re lost in Dwinelle?). So we have some lingering questions heading into the season opener. We’re still expecting a decent season—again, it helps to return nearly everyone (and add some new faces) to a stellar defense—but forecasting the future is still fuzzy. Can we pull some upsets and get 9 wins? Will the Cheez-It offense return and drag us down to the precipice of bowl eligibility? Your guess is as good as mine. And thanks to all of you who participated, we had about 200 guesses as to how the 2019 season will unfold.
The Road to 7.21 Wins
In the table below we have the Bears’ average chance of winning each game according to the predictions that you all submitted. For most games we don’t see much change in either direction compared to our spring predictions, but the standard deviations have grown in nearly every game. If you were too dazzled by Leland’s graphs in last week’s Pac-12 power rankings to remember his definition of a standard deviation, allow me to refresh your memory: a standard deviation captures how much variation is in the data. A larger standard deviation indicates we have more variation (and more uncertainty) in our results, while a smaller one suggest more agreement. Notably, the standard deviation grew in 11 of our 12 games (Oregon State was the lone exception), and the SD for the win total grew nearly 30% from 1.10 in our spring predictions to 1.39. Interestingly, our win total has barely changed (up to 7.21 from 7.20 in the spring). Our average predictions have not changed, but the range of possible outcomes has grown notably, for better or for worse. To get a better sense of how the season may unfold, let’s go through these results game by game.
|Opponent||Avg. Win Chance||Standard Deviation|
|UC Davis||94.2% (-2.2)||9.1|
|at UW||39.2% (+1.6)||18.0|
|North Texas||85.4% (-2.2)||12.4|
|at Ole Miss||59.5% (+1.9)||16.7|
|at Oregon||33.3% (-0.3)||16.0|
|at Utah||38.0% (-4.8)||16.9|
|at LSJU||56.6% (+4.1)||23.6|
|at UCLA||55.6% (+2.5)||18.3|
|Total||7.21 wins (+0.01)||1.39|
We start out with the likeliest win of the year, a 94% chance of victory over the Aggies. Then we have our first coin flip of the year, a 39% chance of beating UW in consecutive years for the first time since 2006. Next is our second-likeliest win of the season, an 85.4% chance of beating North Texas. You all are quite optimistic about our chances of beating a team that won 9 games last season and returns nearly all its skill players from an offense that put up 35 points and over 450 yards per game. Finally, we close out the OOC schedule with the Ole Miss rematch: a game in which we favor the Bears, but not enough to move this out of coin-flip territory. Add up all those OOC games and we expect to get about 2.4 wins. Include the intriguingly early UW game and we expect to have have 2.8 wins in the first 4 games. Rounding up, that’s 3–1. I wouldn’t be disappointed with a 3–1 start.
Herm Edwards’ random number generator of a team visits to close out September and we give the Bears a 63.5% chance—our fourth-likeliest win of the year. The fact that a 7-win ASU team is our fourth-likeliest win speaks to the difficulty of this year’s schedule. Speaking of difficulty, the trip to Eugene eight days later earns our lowest win likelihood of the year. Perhaps I watched too many dropped passes, low snaps, and odd bouts of inaccuracy from Justin Herbert last year, but I think that game is much more winnable than the trip to Utah.
After a bye week, we host the Pac-12’s worst defense as Oregon State visits Strawberry Canyon. As it has been for the past five years, this is our likeliest conference win of the year. Then we have a trip to visit the reigning Pac-12 South champions Utah, where we have a 38.0% chance of winning. The last couple trips to Salt Lake City have been eye-gougingly bad (Goff’s five turnovers in 2015 and in 2012 we saw a blowout loss culminating in Keenan Allen’s season-ending injury on a meaningless, garbage-time onside kick). Hopefully that’s another inglorious tradition Wilcox can end.
At this point in the season we ought to be 5–3 with four remaining coin-flips in the schedule. We favor the Bears in every one of the November games, but we would be quite fortunate to win all four. When it all ends, we should finish the season with 7.21 wins. This is remarkably close to our three most recent projected win totals: 7.20 in our predictions earlier this year, 7.20 in our preseason predictions in 2018, and 7.23 wins in our spring 2018 predictions. Will we be satisfied with another 7-win season? If it includes losses to all three California schools and another 4–5 conference record, probably not. If it’s a 5–4 conference record with a Big Game win, another win over USC, and an upset over Oregon, that might keep the fanbase complacent enough to wait another year for a real offense to emerge. Whatever happens, we’ll probably endure an agonizing loss or two (remember those excruciating back-to-back losses to Arizona and UCLA last year?) as well as an exhilarating win or two (I know you remember those nailbiting wins over Washington and USC).
We’ve spent a lot of time looking at text and a table. Let’s shake things up and look at a chart. Below I’ve plotted all the predictions for each game (except UC Davis because that one is uninteresting and, more importantly, it breaks the y-axis of the other plots). This helps to show how our votes were spread out over the range of win probabilities for each game.
Before we move on, I’d like to personally thank each and every one of you who contributed to that little blip at 100% for the Big Game.
Simulating the 2019 Season
As you may or may not have noticed, football games produce binary outcomes. Either we win or we lose. As a result, all this 60% chance of winning and 30% chance of winning discussion is a bit divorced from reality. So I devised a way to turn our messy predictions into meaningful outcomes: simulation.
The simulation process is simple: I take a prediction from the UC Davis game at random and use that to determine whether we win or lose. For example, if the prediction is 90% then I have an 90% chance of drawing a win and a 10% chance of drawing a loss. After drawing that outcome, I move to the Washington game, repeat the process, and track our record after that game. After that, I move on to North Texas. The process continues until I get to the season finale against UCLA. After that, I save the total number of wins over the course of the season and then start the process all over again. After 1,000,000,000 simulated seasons I think we have enough data to make some meaningful conclusions. So what do the results tell us? We’ll win 7 games. Good thing we went through this simulation exercise, right?
Actually, the simulations provide some nuance into some of the possible outcomes for the 2019 season. Obviously, we can’t win 7.21 games. So how likely are we to win 7? What about 8? What about the case where all those coin flips land in our favor and we win 10 or 11 games? The table below shows us how likely we are to finish with anywhere from 0 to 12 wins.
According to the simulations, we have over an 85% chance of earning bowl eligibility. We’re likelier to win 9+ games than we are to stay home in December. And we have a 40% chance of exceeding last year’s win total. Of course, that all depends on those coin flips.
In addition to looking at the final outcomes, we can use the simulated seasons to see what our record might be over the course of the season. In the chart below I show all possible win outcomes after each game. For example, we have a 94% chance of being 1–0 after Davis and a 6% chance of being 0–1. After Washington, we have a 37% chance of being 2–0, a 59% chance of being 1–1, and a 4% chance of utter disaster with an 0–2 record. Over the course of the season likelier outcomes have darker blue cells (but as we play more games, those probabilities get dispersed over more possible outcomes).
Again, the simulations help provide some nuance to the results to show us how we end up with 7 (or 8 or 6 or 12!) wins.
As always, we have recognize three categories of predictions: the most optimistic, the most pessimistic, and those closest to the group average.
First, we have those for whom the sun never sets.
|1. David Shaw slaps babies||12.00|
|1. OCBear 1983||12.00|
|4. Oski Disciple||11.32|
|6. Class of 87||10.05|
|8. the ghost of joe roth||9.70|
|9. Old Grumpy Bear||9.68|
These may seem overly optimistic, but we do favor the Bears in 9 games this season. So these are not so far-fetched. Shout out to our ATQ friend KuhlDuck for being so optimistic about his ATQ South brethren this year.
Next we highlight the lowest predictions.
|4. No account||3.91|
These are grim, but 3 or 4 wins are possible if every single coin flip goes awry. That would be an agonizing season.
The Voice of Reason
Finally, we have those whose predictions were closest to the community average.
|3. Bearly Legal||4.76|
|9. GoldenBear 77||5.23|
In a little more than 48 hours all this speculation will be a distant memory as we will have real, live football to watch. Thanks for participating! Enjoy the season!
How would you feel about a 7-win regular season?
This poll is closed
CONTRACT EXTENSIONS FOR EVERYONE!