Does the NFL Preseason Matter
Last weekend, a friend of mine mentioned that it's important for NFL teams to do well in preseason, since the results can affect a team's performance in the regular season. I assured him that was a load of baloney, but my buddy insisted he'd seen a study proving it. A lifetime of following the NFL left me certain that there was either a very weak correlation between preseason and regular season performance, or — more likely — no correlation at all. He sent me the link
he'd seen, which is from almost a decade ago and kept the sample sizes pretty small.
So I did my own research, looking at preseason results from every season since the league expanded to 32 teams (2002-10), a total of 588 preseason games. Some of what I found surprised me, though most of it confirmed my prior belief that preseason results are effectively meaningless. Consider the 2010 preseason:
* The Super Bowl champion Green Bay Packers went 2-2. So did the New England Patriots, who finished with the best record in the NFL, and the Atlanta Falcons, who had the best record in the NFC.
* The only team to go 4-0 in preseason was the hugely disappointing San Francisco 49ers, who finished 6-10 in the NFL's weakest division.
* Two teams went 0-4: the Chicago Bears and Indianapolis Colts, both of whom won their divisions.
I could go on. The Seattle Seahawks, who won the NFC West, went 1-3 in preseason. Everyone else in the division went 3-1 or better. And so on. It's seldom that extreme, but most years, there are good teams who do terribly in preseason, and bad teams who look like world-beaters when the score doesn't matter.
Over the past nine years, the team with the worst preseason record is the Indianapolis Colts, who have gone 11-27 (.289). The five teams with preseason records worse than .400, along with their regular season records during those years:
1. Indianapolis Colts, 11-27 (.289) preseason — 109-35 (.757)
2. Kansas City Chiefs, 12-25 (.324) preseason — 67-77 (.465)
3. Houston Texans, 13-24 (.351) preseason — 55-89 (.382)
4t. Buffalo Bills, 14-23 (.378) preseason — 59-85 (.410)
4t. Philadelphia Eagles, 14-23 (.378) preseason — 91-52-1 (.635)
That's two of the best teams in the NFL (Colts and Eagles), two of the worst (Bills and Texans), and one that's been up and down (Chiefs). Tough to draw any conclusions from that. Fewer teams have really excelled in preseason over the last nine years, but there are two who are well over .600:
1. New York Jets, 25-12 (.676) preseason — 72-72 (.500)
2. Dallas Cowboys, 23-13-1 (.635) preseason — 78-66 (.542)
Again, you wouldn't really want to draw any conclusions from that. The team with by far the best preseason record since expansion is exactly average in the regular season, and the team with the worst record is in reality one of the most consistently successful teams in the league, making the playoffs during every season included in my study.
Ultimately, though, I did find one correlation that surprised me. Teams that do poorly in the preseason are below average in the regular season. Not dramatically so, but the correlation is there. Interestingly, I did not find the opposite to be true: doing well in the preseason appears to be meaningless. Since the 2002 expansion, 16 teams have gone 4-0 in preseason. That includes successful teams like the '03 Patriots and '05 Broncos, but it also includes terrible teams like the '03 Cardinals and the winless 2008 Lions. Overall, those 16 teams went 130-125-1, almost exactly average. A great preseason doesn't seem to be a meaningful guide to a team's regular-season performance.
When a team really bombs the preseason, though, that can sometimes be an indication that the team is in for a disappointing year once the real games begin. Since '02, 18 teams have gone 0-4 or 0-5 in preseason. Again, there are some good teams: six of those 18 teams won 10 or more games, including the 2005 Colts, who started 14-0. But overall, the teams that went winless in preseason came up with a collective record of 134-154 (.465), which isn't devastating, but it's not good, 10 games below .500. Eighteen is a pretty small sample size, so you wouldn't want to draw too many conclusions from the fact that these teams went an average of 7.5-8.5. I mean, it's not like they all went 5-11 or something.
But there's more. I've always thought of records between .250 and .750 as pretty average. A good team having a bad year might finish 5-11, and a pretty average team can win 11 games with a little luck. But 4-12 or worse is an awfully poor season, and anything 12-4 or better generally marks a pretty elite team. Looking at the years I studied, 40 teams went 12-4 or better. Those teams finished a collective 86-77 in the preseason (.528), which is a little better than average, but not at a significant level. Literally every season, there is a great team who had a miserable preseason.
* In 2002, the Eagles went 1-3 in preseason before finishing 12-4 and making it to the NFC Championship Game.
* The 2003 Rams were 1-3 in August, but they went 12-4 and got a first-round bye in the playoffs.
* Both 2004 Super Bowl teams, the Patriots and Eagles, finished 1-3.
* The Colts finished 1-3 or worse every year from 2005-09, a total record of 4-18 (.182). Their regular-season record over those seasons was 65-15 (.813), every season 12-4 or better, including two years at 14-2, with two AFC titles and a win in Super Bowl XLI.
* In 2010, the division-winning Bears, Colts, and Chiefs finished a combined 1-11 in preseason.
Doing well in the regular season simply is not related to doing well in the preseason. The correlation is extremely small, less than half a game. But doing poorly in the preseason may be a slightly different story. Over the past nine seasons, 42 teams have finished 4-12 or below. Those 42 teams posted a combined preseason record of just 70-92 (.432). That makes almost 100 points of difference between the preseason records of the best teams (.528) and the worst (.432). Is this data statistically significant? No, not really. The difference is still pretty small — .432 is 7-9, a pretty average season — and the sample sizes with these teams aren't large enough to inspire confidence in the findings. Here are the big numbers: the chart below shows the overall results, sorted by preseason record, since the league went to 32 teams:
* Includes a 3-0-1 preseason
** Includes a 2-1-1 preseason
^ Includes a 0-5 preseason
Sorted another way, by winning record, losing record, or exactly even:
The difference is not between teams that do well in preseason and those that do poorly; it's between those who do poorly and those who are exactly average. Looked at this way, the difference is small, but it is real and statistically significant. Bombing in August may be a negative indication about a team's actual strength. The causal nature of this relationship is not obvious: do bad teams lose in preseason because they're bad, or does a poor preseason affect a team's psyche going forward?
When this becomes interesting is from 2004 on. In the 2003 preseason, and to a lesser extent '02 as well, there was a dramatic connection between preseason and regular season performance. In '03, both Super Bowl teams (Pats and Panthers) went 4-0, as did the 12-4 Titans. Even the Colts had a good preseason, going 3-1 before a 12-4 regular season. Two teams, Houston and Atlanta, followed winless preseasons (0-4) with 5-11 regular season records.
Looking just from 2004-10, the results show the opposite of my friend's assertion: there is an inverse relationship between preseason performance and regular-season success. Teams that went winless in preseason had a better regular-season record (109-115, .487) than those who finished preseason undefeated (71-89, .444), and Super Bowl teams had a losing preseason record (27-29). Here's the same table as above, looking just at the past seven seasons:
* Includes a 3-0-1 preseason
** Includes a 2-1-1 preseason
^ Includes a 0-5 preseason
I don't normally like to shrink sample size, but the data appears to show a trend starting in 2004, with teams that were very successful or very unsuccessful in preseason marginally worse than the rest of the league. I also think the records of teams that played five preseason games are potentially instructive here. The 12 teams that went 3-2 or better were subpar in the regular season (.443), while those that went 2-3 or worse were exceptionally successful (.635), with four of the six making the playoffs. This is probably a statistical aberration rather than meaningful data, but it's still interesting, I think.
Preseason records should be ignored when predicting a team's regular-season fortunes. There is a correlation between preseason mediocrity and regular-season success, but it is too small to be of any predictive use, more likely to mislead forecasters than to assist them. From a fan's perspective, preseason games are almost entirely meaningless.