Predicting the Second Half of a NFL Season Using the First Half
By Scott Kacsmar
Cold Hard Football Facts’ Comeback King (@CaptainComeback)
At the midway point of the NFL season, a lot of attention shifts towards the playoff picture and which teams are legitimate Super Bowl contenders.
It is already obvious which teams are the worst (here’s looking at you, Jacksonville and Kansas City). There are 17 teams with a record of .500 or better, and those are the ones that will most likely be competing to get into that 12-team tournament where anything can happen.
But can we use the first half of the season to predict which of those teams will likely make up the 12, or is there really something to the playoff teams generating steam late in the season?
Many things can skew the numbers with half a season to go. Unpredictable injuries are a big one, and look no further than the 2011 Chicago Bears who were 7-3 before Jay Cutler was lost for the season. Chicago finished 8-8 and missed the playoffs.
But the 2011 Bears are not your average case. Having to start Caleb Hanie sealed their fate to be a bad team at the end of the season.
What does the data say for the whole league as we look to predict the second half of the season?
Using Pro-Football-Reference, we collected data on teams during the 2002-2011 regular seasons for how they did in games 1-8 and 9-16. That covers every season with 32 teams in eight divisions, meaning a total of 320 teams (120 playoff teams).
The correlation coefficient for winning percentage in games 1-8 compared to games 9-16 is 0.44, which would suggest there is neither a strong or weak relationship between the two halves of a season.
First half to second half
Interesting to find that points scored (0.57) have a much stronger correlation, or carryover effect, than points allowed (0.32), which would add statistical support to the idea that offensive performance is easier to sustain than defense.
If you built a regression model using the win percentage and scoring differential from the first half of the season to predict second half of the season win percentage, then the r-squared would only be 0.25. That means the model’s no good as nearly 75 percent of the variation is unaccounted for.
Here is the breakdown of how the 320 teams started the season, and the last five columns show the average number of wins they had in the second half of the season, the decline in wins, and splits for how many of the teams had a winning, losing or .500 record in the second half of the season.
Teams with a winning record in the first half continue with a winning record in the second half 60.3 percent of the time (79 of 131). They only finish on a losing run 18.3 percent of the time (24 of 131).
Likewise, teams with a losing record in the first half continue with a losing record in the second half 55.8 percent of the time (72 of 129). They only finish on a winning run 24.8 percent of the time (32 of 129).
If you start 7-1 or better, then your chances of having a winning second half are great. The only team to finish .500 was the 2006 Indianapolis Colts, who went on to win the Super Bowl anyway.
The teams at 5-3 or 6-2 have less sustained success, but still continue winning more often than not (54.1 percent of the time).
Teams who are 4-4 repeat that performance 20 percent of the time, and are almost just as likely to have a losing second half as they are a winning one.
We have seen three 3-5 teams rebound to make the playoffs since 2002, such as last year’s Broncos, and you can see they have more success than the teams buried at 2-6 or worse. The Broncos went to Tim Tebow at quarterback, while the 2002 Jets turned to Chad Pennington to spark their run.
The only 0-8 team to finish on a winning note was the 2008 Bengals, who finished 4-3-1.
The Division Races
Can we look at the division standings through eight games to get a good idea of what the final standings will look like?
Well the answer is difficult, and would require a deeper look into divisions that have more separation versus divisions that have four teams separated by a game. Obviously when the race is so close, the results could be the exact opposite by season’s end.
Regardless, here was a painstakingly mad attempt at getting the division standings for the first eight games of a season, and compared to how the teams finished at the end of the regular season.
Final Division Standing
There are 80 of each, so this means teams in first place in their division through eight games have gone on to win the division 71.3 percent of the time.
The most common result for each position is to maintain that position at season’s end. It happens most frequently for first (71.3 percent) and fourth (68.8 percent), while just under half the time for second (46.3 percent) and third (45.0 percent), which makes sense.
Last season seven out of eight teams in first place through eight games won their division. Only San Diego (4-4) was beaten by last-place Denver (3-5), who was a game behind the rest of the AFC West.
The only two teams to go from first to last were the 2010 Titans and 2002 Bills.
Who made the big leap from worst to first? It was the 2011 Broncos, the 2008 Dolphins, the 2002 Jets, and the 2002 Raiders.
Ultimately, everyone’s goal is to make the playoffs, which obviously is a lot easier to do when you set yourself up with a good start to the season.
We have had 59 teams start 6-2 to make the playoffs, while 61 were 5-3 or worse, showing you do not have to be far ahead early to still make it. You can also see a 6-2 start is no guarantee of the postseason, and it only goes down from there.
When looking at the correlation coefficient for just these playoff teams, there are some different results when it comes to points. The points scored are down to 0.41, while points allowed are now only 0.14.
You would have expected more correlation between season halves for a playoff team, but that is not the case. Often we see teams peaking late or early, producing a split in stats.
Playoff Teams 2002-11
Those numbers look nearly identical, but that is because half of the teams regress in the second half, while the other half improve.
Exactly half of the teams (60) scored more points in the second half, while the other 60 either scored the same or fewer. Likewise, 55 teams allowed fewer points in the second half, while the other 65 allowed more/the same.
In the end, they all still meet in the postseason.
Applying it to 2012
What about our current league standings in the 2012 season? A few teams have played nine games already as you can see here.
If we expect at least two current division leaders to fall from that spot by season’s end, giving us a 75 percent retention rate, then smart money would have to be on Baltimore and Chicago being replaced by Pittsburgh and Green Bay.
The Ravens have looked spotty while Pittsburgh continues to improve, and while the Bears have that awesome defense, check the schedule. The NFC North has all teams over .500, and Chicago is playing Houston this week, at San Francisco the next, and another game with Green Bay, who has the chance to sweep them and perhaps win the division.
The AFC appears to break down rather easily with Houston, Denver, and New England as projected division winners. Baltimore and Pittsburgh will both make it, and that leaves the stunning Indianapolis Colts (6-3 now) in great position with a head-to-head win over Miami (4-4) and only the Chargers (4-4) as the other team at .500.
The NFC is harder to judge right now, but you can already book a spot for undefeated Atlanta in the playoffs. No 8-0 team has ever failed to get a first-round bye.
We talked about Chicago and Green Bay both making it. The Giants have a great lead in their division, even if they go on another nosedive. San Francisco is the class of the NFC West.
It comes down to that last spot, and right now Seattle (5-4) holds it, but must fight off Minnesota (5-4), Detroit (4-4), Tampa Bay (4-4), and if any 3-5 team is dangerous, it would be New Orleans.
We are technically past eight games with some of these teams, and picking Seattle for the playoffs a week ago before they beat Minnesota would have been less likely, but it looks like a strong pick at the start of Week 10.
Will all 12 teams hang onto the playoff berths they currently are in position for? We will see, but as always, the NFL makes you expect the unexpected. Would you really be surprised to see Drew Brees rally the Saints to a 9-7 playoff season?
If some of these teams fail to make it, much will be said for the losses they suffer late in the season and especially in Week 17.
But just remember those losses in September and October put them in that position in the first place, as those games are very important too when you only get 16 shots a season.
The only time a game is meaningless in the NFL is when it does not impact the playoff picture, change draft positioning, or is won by Matt Cassel.
Here’s to a great second half, which will only partially be a repeat of the first.
Scott Kacsmar is a football writer/researcher who has contributed large quantities of data to Pro-Football-Reference.com, including the only standardized database of fourth quarter comebacks and game-winning drives. He probably just put the jinx on Pittsburgh for Monday Night Football. You can visit his blog for a complete writing archive. Please send any questions or comments to Scott at email@example.com, or you can follow him on Twitter at @CaptainComeback.