The CGB community is riding an impressive streak of accurately predicting the number of wins the Bears will achieve each season. In 2014 we predicted 5 wins. Cal finished 5-7. In 2015 we predicted 7 wins. Cal won 7 in the regular season and capped the year with a bowl win. This past year we predicted another 7 wins for the Bears.
7 wins was a rather bold prediction for a team that was losing Jared Goff and his top 6 receivers. However, 7 was not inconceivable given the team’s recent upward trajectory. If we had known the Bears would beat Utah, Texas, Oregon, and UCLA, a 7-win season would seem downright modest. On the other hand, if we knew we’d lose to Oregon State, San Diego State, and a woeful ASU team, 7 wins would sound quite improbable. Alas, our misguided predictions proved to be too optimistic, as the Bears stumbled to 5 wins and set in motion the events that brought us to the new Justin Wilcox era. While this season may have disappointed, this momentary setback appears to be better for the program’s long-term health. Before we start dreaming of Rose Bowls and National Championships under Wilcox (was I just saying something about misguided optimism?), let’s look back at how this past season fared against our expectations.
Below I have listed our predictions for each game along with the outcome.
|Hawaii (7-7)||87.84%||W (51-31)|
|at San Diego State (11-3)||69.31%||L (40-45)|
|Texas (5-7)||59.87%||W (50-43)|
|at Arizona State (5-7)||57.90%||L (41-51)|
|Utah (9-4)||54.53%||W (28-23)|
|at Oregon State (4-8)||80.36%||L (44-47)|
|Oregon (4-8)||40.92%||W (52-49)|
|at USC (10-3)||30.24%||L (24-45)|
|Washington (12-2)||49.91%||L (27-66)|
|at Washington State (8-5)||54.48%||L (21-56)|
|Leland Stanford Junior University (10-3)||49.36%||L (31-45)|
|UCLA (4-8)||50.94%||W (36-10)|
|Total||6.86 wins||5 wins|
6.86 projected wins and 5 total wins—it’s pretty clear Cal fell short of expectations in 2016. In our pre-season predictions we favored the Bears in 8 games; we won 4 of those. We shouldn’t necessarily expect the Bears to win every game in which they’re favored, especially if we only give them a 50% or 60% chance to win. Still, losing half the games in which we favored the Bears was a disappointment. On the other hand, Cal did win one game as a pre-season underdog. However, we shouldn’t expect Cal to lose every game as an underdog. In fact, based on the predictions above we were projected to win 1.70 games as underdogs. Winning only one means we even managed to underachieve at overachieving!
Comparing our predictions to binary win-loss predictions is a bit simplistic. If we give the Bears a 99.999% chance of beating some cupcake and we win 3-2, it will be a much less satisfying win than if we give Cal a 50.00000001% chance of winning and we win 105-0. How the final score compares to our prediction offers a bit more insight into how well our predictions match reality.
Predictions vs. Reality
In order to compare our predictions to reality, we first have to define reality. So, what is reality? Unless your name is Donald Trump, you recognize that reality has depth and nuance, and it may be interpreted differently from different perspectives. As such, measuring the Bears’ chances of winning based on their performance on the field is not a problem that is easily solved.
One approach to deriving the Bears’ chances of winning each game is to use the final score to derive some measure of how likely the Bears are to win if the game were played again. If Cal beats the Lobsterbacks 99-0, we can expect that they’d be much more likely to win again than if they had won 35-34. Using this strategy, I define reality by using the Pythagorean Projection. For each game, the probability of winning is calculated as follows.
In the following chart I have applied the Pythagorean Projection to each of our games in the 2016 season. I’ve plotted reality in gray and our preseason predictions in blue. When reality falls below our predictions, the Bears have underachieved. Likewise, when gray is higher than the blue line, the Bears exceeded our expectations.
Cal underachieved in the first two games of the season in a we-shouldn’t-have-given-up-31-to-them win over Hawaii and another disappointing loss in San Diego. We met expectations in the win against Texas and then faceplanted against ASU. The win over Utah slightly exceeded expectations and then we suffered the biggest disappointment of the year in the loss against Oregon State. Despite giving the Bears an 80% chance of winning we went out and delivered the most embarrassing loss of the Sonny Dykes era. Fortunately we bounced back with an expectations-exceeding win over CGB North. Then things went pear-shaped.
After beating Oregon, the Bears underachieved in four consecutive games to squash our bowl hopes and likely seal the end of the Sonny Dykes era. Not even our first win over a California school since 2012 could save Sonny, although it was an unexpected and pleasant surprise. Of course, over the course of the season we may have re-calibrated our expectations for the team and we may have updated our expectations for our opponents. By November we would have been much less optimistic about our chances against Washington while we should have been more optimistic about our chances against a hapless Bruins squad. When we compare our pre-game predictions to reality, does the season still seem like a disappointment? To answer that, I’ve added our pre-game predictions (taken from our weekly report card series) to the chart in dark blue. Once again, if that dark blue line is above the gray line then the Bears have underachieved. If the gray line is above the dark blue line, the Bears have overachieved.
For many years we’ve had a habit of exhibiting volatile reactions to wins or losses. Wins get those sunshine pumps going at full speed while losses portend a smothering, endless doom. There’s something quite amusing about watching our predictions wildly swing back and forth over the first half of the season. After the Hawaii win we became stuck in an oddball pattern where we lost each time we favored the Bears and won each time we picked the Bears to be underdogs. After six weeks of wackiness peaking with an overtime win over Oregon, reality finally set in and we never picked the Bears to win again. We had loved the Bears and they broke our hearts too many times. Let that be a lesson, Cal fans: never love anything.
When looking at our wacky week-to-week predictions, the Bears fared slightly better than our preseason predictions. They underachieved 7 times (Hawaii, SDSU, ASU, OSU, USC, UW, and WSU) and overachieved 5 times (Texas, Utah, Oregon, LSJU, and UCLA). Underachievement in wins is obnoxious but acceptable; however, we only won one of our underachieving games. That’s thoroughly unpleasant. At least our overachieving wins were quite memorable. Not that it was enough to save Sonny Dykes’ job...
In our final chart I’ve added Las Vegas’ predictions, derived from the money line posted before each game.
Other than the Texas and UCLA games, Vegas’ predictions were remarkably accurate. Vegas closely matched our accuracy over the second-half slump and avoided the volatility we succumbed to over the first half of the season.
If, for reasons I will never understand, you prefer tables to charts, please enjoy the data from the above plot in the table below. We’ll wait here patiently while you peruse the results.
|Reality||CGB Preseason Predictions||CGB Pregame Predictions||Vegas Predictions|
|Total||5.40 wins||6.86 wins||5.45 wins||4.76 wins|
All done? Great! Now let’s move on to the best part of the post (again, unless you really love that table up there)...
This was a fairly unpredictable season: we had several uncharacteristically satisfying wins and some eye-gougingly bad losses. So which of you were able to predict all this madness? To find out, I took each of your predictions from our final round of preseason predictions and compared your game-by-game predictions with our measure of reality. After some fancy magic involving subtraction, exponentiation, and summation, I rated you all from most accurate to least accurate. Behold, the most accurate among you:
First we have the Ursadamus award for the ten predictions closest to reality.
|Name||Total Deviation from Reality|
|1. Antiguo Azul||0.456|
|3. webb's hog||0.562|
|9. Nik Jam||0.667|
Congratulations to Antiguo Azul! chruppel and webb’s hog round out the podium. Proving that we CGB writers aren’t completely removed from reality, Nik Jam finished with the 9th-most accurate predictions! Thanks Nik!
Next we have those whose predictions were furthest from reality
Look at that, a four-way tie!
|Name||Total Deviation from Reality|
|1. Old Bear 71||4.392|
|6. Oski Disciple||2.885|
|7. Ben Lynch Ruled!||2.270|
|9. TKE Prytanis 79||1.711|
C98, FeedtheBeast, Old Bear 71, and DavidShaw’sPussyStinks all gave the Bears a 100% chance of winning every single game, so they share the dubious honor of having the least accurate predictions. Don’t lose hope, though. I’m sure one day Cal will win every single game by shutting out its opponents and you all will prove to have excellent foresight. It may not happen today. It may not happen tomorrow. But it’s definitely going to happen on September 2nd in Chapel Hill. GO BEARS.