In the 20-plus years of the Madden football game franchise, infamy has surrounded those selected to grace the annual football simulator’s cover. Since the San Francisco 49er’s Garrison Hearst broke his ankle after being named Madden ‘99’s cover athlete, rumors have circulated of a “curse.”
The “curse” grew in infamy as an ever-increasing numbers of Madden cover athletes suffered high-profile injuries: after being named to the Madden cover, Michael Vick broke his fibula during the 2003 preseason, while Vince Young, Troy Polamalu, and Shaun Alexander all lost time due to injury. Other cover athletes simply had terrible seasons compared to their normal successes: Drew Brees had a down year after his selection, and Marshall Faulk, usually so reliable, failed to pass 1,000 yards rushing. Now, fans are so superstitious about the curse that they actively campaign for their favorite player to not get selected for the cover.
Of course, other publications and experts have weighed in and dismissed the Madden “curse” as nothing more than superstition. The drop off of players featured on the cover, so the argument goes, is due to two factors: 1) regression fallacy-- basically, people assume that players have regressed because of outside forces (the curse) instead of taking into consideration natural fluctuations in football. If a player does well one year, people expect they’ll play that well next year. 2) the spurious correlation between a player’s appearance and their statistical dropoff. Essentially, people attribute the cover selection and its curse as leading to a player having a bad season the following year.
So who is right? How do you even prove a curse exists?
We initially tried to consult occult experts to determine why there would be evil surrounding the cover of a video game series, but, unlike the billy goat’s curse that haunted the Cubs, we could find no compelling backstory to explain what diabolical forces Madden might have tapped into. So we turned to math.
While we can’t prove the existence of a curse, we hoped that a thorough look at the numbers would prove once and for all whether players who were selected for the cover were statistically more likely to have bad seasons the year after their selection. If cover selection = bad following year, then the numbers should show it. So we asked Nick Hatley, a research analyst out of Washington, DC, to help us build a mathematical model.
Using ESPN’s fantasy numbers, we gave each offensive player a numerical value per season-- incorporating things like touchdowns, yards, yards thrown, etc. For example, Tennessee Titans running back Eddie George accrued 1,509 yards rushing and 14 touchdowns in 2000, which translates to 342.2 points in our system. In the data below, using NFL records and ESPN’s numbers, we tracked each Madden cover athletes’ output over their careers.
A quick note: Our fantasy-focused model has some limitations. We only included offensive athletes in our analysis because comparing offensive stats with defensive stats proved too clumsy. With only a handful of defensive players featured on the Madden cover, creating a combination model proved too unwieldy. Apologies to Richard Sherman.
First, we looked at the type of player who is selected for the Madden cover. These are, with few exceptions, some of the top players in the league. Drew Brees, Marshall Faulk, and now Tom Brady have all graced the cover. To illuminate whether or not players’ decline is simple regression of top talent, we compared the regression of Madden cover athletes with the regression of MVPs.
But in our research, players who received the MVP the year prior regressed more than Madden cover athletes. In other words, you were more likely to have a comparatively bad year after being an MVP than being a Madden athlete. Of course, “We don’t talk about the MVP curse,” Hatley explains. Why not? Players are selected for the cover when they’re likely at their peak, as the thinking goes, so they’re more likely to regress. MVPs and Madden cover athletes follow the same pattern, but Madden athletes didn’t even regress as much: Madden athletes declined by an average of 71.4 points the year after their cover year compared with an average decline of 91.8 points for MVPs.
But in our research, players who received the MVP the prior year regressed more than Madden cover athletes.
So it looks like yes, the Madden curse can somewhat be attributed, in aggregate, to larger player regression trends. So the curse is disproved, right? Not so fast.
Madden cover athletes are not just the top players in the league, but are often selected by fans for a variety of reasons. These players sometimes aren’t in the MVP conversation at all and may still be early in their careers. Peyton Hillis is a great example-- a running back whose career was on the upswing but by no means a dominant force in the league. The Madden cover looked like the start of a long successful career. Surely, players like Hillis won’t regress as much as players at their peak, right? Our model needs to take that into account.
So comparing Madden athletes to MVPs is too narrow. MVPs rarely repeat in the league, and are doomed to regress after their peak year. So what about comparing Madden athletes to a fantasy top ten performers at each position (QB, RB, WR)? This allows us to compare Madden players to a wider range of players that had a successful prior season.
Surprisingly, in our model, Madden athletes experience a 24.5 point decrease more than a top ten player. To put it another way: compared with their previous season, a Madden athlete is more likely to have a worse year the next year being on the cover than a regular player coming off a top ten fantasy point season. To go back to our Hillis example, Hillis never achieved the heights of his pre-Madden cover season. Instead of being the mark of an up-and-coming player, he peaked and regressed.
Hillis’ case turned out to not be just an anecdote. The difference in a Madden player’s decline and a Top 10 position player's is statistically significant, Hatley explains, meaning it’s further away from random variance. So we can’t dismiss the curse outright as random chance. Madden athletes have worse dropoffs than the average top ten player. That’s strange.
Perhaps the most high-profile aspect of the Madden curse is the perception that after getting awarded the cover, players are afflicted with bizarre and enduring injuries that costs them playing time. We would be remiss if we didn’t take that into account with our calculations.
So we compared how much time a player who had just achieved a top ten fantasy season averaged in missed games the next season compared to the missed games of a Madden athlete. But again, the curse didn’t hold up in our numbers: Madden cover athletes missed FEWER games on average the year after their selection than a top ten athlete.
But perhaps we’re again injecting bias into our model by comparing the high-profile injuries of Madden athletes to fellow athletes playing at a high level. So we decided to compare Madden cover athletes to the average player.
Surprisingly, compared with an average NFL player, a Madden cover player misses a half a game more on average. That’s it. In our research, Madden cover players missed 1.2 games on average after being on the cover, and top ten players from the year before missed 1.57 games in their next years. For average players, they missed .079 games in the following year. It’s not looking good for the curse.
The Madden Curse has captivated our interest because of several high profile incidents. Hillis never played another great season. Gronk got hurt. Hearst’s ankle didn’t hold up. But in aggregate, these strange seasons are given context. Football is a highly variable and injury prone sport. Our research doesn’t show a series of strange data points that can only be explained by the supernatural. Instead, it shows largely reversion to the mean.
It’s not looking good for the curse.
Players who are not injured for long periods of time (and therefore can accrue a large amount of stats) are likely to regress back to the norm. But that won’t stop people from attributing these regressions to other things. Indeed, the sport is rife with superstition that tries to explain these spurious correlations: wearing a particular color increases a chance of victory, just as not following the unwritten rules of a sport will likely lead to a “curse.”
We set out to prove or disprove evidence of the curse, and our verdict is mixed. While Madden cover athletes don’t regress more than MVPs, our research predicts that they’ll have, on average, worse seasons compared with the year before than other top tier players-- even if they don’t miss as much playing time. If that’s enough for you to believe in a curse, maybe don’t draft Tom Brady in your fantasy league.
Christian Holt is IGN's Custom Content Editor. You can follow him @bicyclesquid
Aucun commentaire:
Enregistrer un commentaire