Should we still care about projections?
Somewhere in Philly, there is an Eagles fan laughing their ass off at the pre-season projections. Why shouldn’t they be? Football Power Index rankings before the 2017-18 season had the Eagles at a .500 win percentage with just a 19.3% chance of winning their division and a measly 3.7% chance of earning the spot at No. 1 seed. If the odds weren’t stacked against them enough, the Eagles had a 100:1 chance to hoist the Lombardi Trophy in February.
The Philadelphia Eagles were a perennial underdog. Their biggest critics were not shy to inform you at any given opportunity that they haven’t won anything since they invented face masks. With one of the more reactionary sports cities, it was a constant struggle of never winning paired with the obvious poking and prodding of smug sportscasters and pundits alike. That all changed the first Sunday of February 2018.
Not only were the Eagles one of the bigger underdogs in not only their division but also in the larger National Football Conference. In addition, the team they snatched their first Super Bowl victory from was sitting on the opposite side of the odds. The Patriots had an overwhelming likelihood of winning their division. FPI granted New England a 96.2% chance of making the playoffs, a 92.3% chance of winning their division along with a 51.1% chance of being the No. 1 seed. If that wasn’t heavy favoriting enough as it was; the Patriots had a 34.7% chance of winning the Super Bowl!
So, as ‘Football Feens’, how are we supposed to feel towards projections? We have two different sides playing themselves out right in front of us. On one hand, we were shown that projections likely do not matter in the case of the Eagles. On the other hand, in the case of the Patriots, we have a team, who more or less, proved how correct they can actually be. I imagine when weighing both sides equally, we can confidently conclude that projections are not the end-all, be-all.
Most of our readers at this point are probably thinking something along the lines of “Yeah, no shit.” (And you would be right). The answer is quite wishy-washy for the average sports feen. To those that hate projections, I’ll refer to this chart provided by Fangraphs.
We can clearly see there is a positive relationship between projected wins versus actual wins. In the case of the MLB. The degree to which is about a 0.36 R correlation. It remains unclear as to how significant that really is. For those that didn’t pay attention in statistics class- a perfect positive correlation is 1.00, with anything on the positive side indicating a correlation. Why does this matter if it is as imperfect as we all suspected it to be? It shows that there is some usefulness in using projections as a guide.
The median error for the MLB over the course of 162 games is about 6 wins/losses. The reason why I decided to pair the MLB’s projections which are relatively linear as opposed to the NFL’s whose were comparatively lackluster is simply due to sample size. It is much more difficult to predict the outcomes of 16 games and all the minute swings therein compared to a league with literally over ten times as many games. There is certainly a lot more to consider when weighing the factors pulling in each direction for the outcomes of football games.
So what is the takeaway from all this? In summation, the projections for the NFL have the biggest opportunities to be correct and incorrect on a large scale. So we should not be surprised when a team like the Eagles slips through the cracks. We can see that the distribution for wins and losses in the MLB is not nearly as flat as it is for the NFL. The Eagles, who were once apart of the meatiest section of the bell curve, then found themselves on the outskirts. Which is something the MLB doesn’t typically see as often. So before taking projections as law or as trash, keep in mind which sport you are referring them to.
So rejoice in beating the odds Philadelphia! But never forget that if the Phillies are ever projected to go .500, their likelihood to still win the World Series is much, much lower.