Are The Chicago Bears Going To Suffer From Regression In 2026?

Hello, Chicago Bears fans! There is a certain argument that purports to be rooted in one of the fundamental laws of statistics, namely regression, that casts doom and gloom upon the horizon of the Chicago Bears, forecasting a return to mediocrity in the 2026 season.  Once such example, and the social media feeds on the NFL and the Bears are quite full of them, is seen here from Carson Caldwell on X:

Before I address this argument in the context of the Chicago Bears, let us talk about what regression is, as opposed to this distorted version employed by Mr. Caldwell. I think part of the problem comes from the colloquial definition of regression, as Merriam-Webster defines it: a return to an earlier, worse, or less advanced state.  We are ingrained from common usage of regression and “to regress “to think of it as to, essentially, “get worse” or go backward.  If we were to talk about the Chicago Bears’ offense “regressing,” we would naturally take that to mean that the Bears’ offense had taken a step back from its current level of success and achievement.  Such use of the word “regress” is not wrong – it is an accurate invocation of the definition and usage of the term in common parlance. 

But it is not what we mean when we talk about “regression” as a statistical phenomenon with regards to sports performance and statistics. So, what is regression in that sense?  The common example we use in statistics classes (FYI, I have a Master’s in Statistics and have taught statistics for over 20 years) is height, as Sir Francis Galton observed it as a phenomenon in the 19^th century while studying population statistics.  The average height for men in the United States is about 5 feet 9 inches, and for women, approximately 5 feet 4 inches.   Let’s assume, though, an NFL linebacker gets married to a WNBA basketball player.  Our NFL linebacker is 6 feet 4 inches, and our WNBA player is 5 feet 11 inches.  Both represent extreme “outliers” in terms of height – about 2.5 standard deviations above the mean for both, or approximately the 99.4th percentile (they are both taller than all but 0.6% of the population of humans).  What are they outlying from?  The mean – the average we have already described.  So, let’s say our married couple have a boy.  What height should we expect the boy to reach in adulthood? 

Many of you may expect two outliers who have a child are likely to produce an even greater outlier – a 6’11” future NBA center, perhaps. Or perhaps you have reasoned that the boy will be equivalent to the height of their parents (say, 6 feet 5 inches).  Well – your intuition that the expected height of the boy is likely to be determined by the height of his parents is correct – we know that from genetics.  But in fact, due to regression to the mean, which is 5 feet 9 inches for men, we would expect the boy to likely be approximately 6 feet 1.5 inches tall – shorter than both his mother and father in terms of their unusualness – the boy would only be approximately 1.25 standard deviations above the mean, or in the 88.44th percentile in height. Still an unusual height – but not as unusual as his parents. 

Regression is fundamentally an observation that unusual things are unusual for a reason.  And if the unusual becomes commonplace, then it is no longer unusual.  In height, a man or woman who is unusually tall who have children would be expected to have children less unusual than themselves – the progeny would ‘regress’ to the mean of the human population in height.  But there is another side to this story.  Let’s say our man was, in fact, 5 feet 2 inches, and our woman was 4 feet 7 inches – unusually short.  Our male is approximately 2.5 standard deviations below the mean height for men (a bottom 1% of the male population in height), and our female is actually 3.3 standard deviations below the mean height for women. Regression to the mean here works in the positive direction.  Their children would be expected to be taller than their parents.  In fact, we would expect the male child of this couple to be approximately 5 feet 3.5 inches tall, or 1.87 standard deviations below the mean (about the 3rd percentile – only 97% of the male population is taller). Positive regression relative to a “low” outlier is just as valid a form of “regression” in the statistical sense as is negative regression relative to “high” outliers.  It is a testament to the power of the center – a return to the normal after something unusual happens. 

And it is positive regression that is most often left out of these pessimistic takes on the Chicago Bears for the 2026 season.  It is, of course, easy to look at the ways that the Bears were unusually successful in the 2025 season and to confidently predict that the Bears are unlikely to reach those heights again.  The two most commonly cited examples are the fact that the Bears’ defense led the league in turnover margin in 2025 (+22) and that the Bears won many games that were close and late (they were, in fact, 8-5 in games decided by one score).   We see one of these explicitly in the Caldwell tweet, and the reference to point differential is an implicit way of referring to the Bears’ success in close games.  And as such – the point is fair enough.  We should expect the Bears to regress in turnover margin.  We would expect the Bears turnover margin in 2026 to be drawn back towards the average (which is zero) and for the win rate in close games to be pulled back towards the mean (which is 50%).  And this is where our interlocuters (many of which just so happen to be Green Bay Packers fans) end their inquisition into the Bears’ prospects and pronounce the Bears doomed to a “regression” – in win total and perhaps even a losing season this year. 

But there is a fundamental flaw in this analysis, and I bet you can now, having a full understanding of what regression is in the statistical sense, deduce it.  That’s right – they are ignoring that the 2025 Chicago Bears were negative statistical outliers in several important categories related to winning games in the NFL.  Let me count the ways.  The Chicago Bears were 27th in the NFL in rushing yards allowed per game, giving up 134.5 yards per contest.  The Bears were 22nd in the NFL in total sack production with just 35 total sacks, tied with the Cincinnati Bengals.  The Bears were 21st in opposing passer rating, they allowed 32 passing touchdowns, and in total defense rank the Bears were 29th.  Huh, well, maybe the Bears were just bad on defense.  And sure – they were.  They were unusually bad on defense.  You cannot rely on negative regression to the mean when talking about the areas where they were unusually good without also accounting for positive regression in the areas where they were unusually bad.  The Chicago Bears doomers only look at one half of the equation. One-half of the story of regression to the mean. They assume the Bears will be frozen in amber on the things they were unusually bad at, while experience statistical regression on the things they were unusually good at. As the meme goes: that’s not how this works. That’s not how any of this works.

And it doesn’t stop with the defense.  The Bears were only 21^st in fourth-down conversion percentage – they converted only 51.7% of their attempts.  Failure to convert those fourth-downs are effectively turnovers.  So fully accounting for regression means tallying the positive and negative regression.  And it is not only in statistical performance that the Bears can hope for positive regression. The Chicago Bears were 1^st in NFL games missed due to injury during the 2025 regular season, accumulating over 240 missed games prior to December.  Key absences were, you guessed it, on defense and in the secondary.  Kyler Gordon and Jaylon Johnson missed most of the season, and were shells of their former selves when they did return.  The Bears lost their starting defensive end (Dayo), a rotational defender (Shemar Turner) for the season, and Booker for eight games to start the season. They lost Tremaine Edmunds, who was having a great season, for four games, and when he returned, he didn’t play nearly as well.  Positive regression would mean we should expect the defense to be healthier in 2026.  And that’s putting aside all of the new starters on both the offense and defense sides of the ball and how they may affect performance.

So, is it fair to note that the Bears are likely to “regress” in turnover margin or their win rate in close and late games?  Sure.  But that is only half of the story – and jumping from that to predicting the Bears will have a worse record in 2026 is the kind of misuse of statistics that has led to such sayings as “there are lies, damned lies, and statistics.”  The Bears are likely to have a great deal of positive regression in areas where they were unusually bad or weak, which will counterbalance the negative regression in those unusually good statistical areas.  And that’s entirely aside from the fact that the 2026 Bears are not the 2025 Bears – new players, new development and growth, and new schemes and strategies will be deployed in 2026. 

As Ben Johnson is fond of saying, the 2026 Chicago Bears are “back to ground zero” as they head into the season.  From a regression perspective, that means the Bears will not get to count their unusual successes, but also not suffer from their unusual failures that boosted them on the one hand and which they suffered from on the other.  So the next time someone starts talking about “regression” and how it is simply a statistical fact, the Bears will “regress” in the win column in the 2026 season, you give them a good talking to on the proper use of statistics.  And you might also mention that the Green Bay Packers were second in the NFL in third-down conversion percentage in the 2025 season, and you can ask what that foretells about the Packers’ offense in 2026. 

What do you think about the discussion of statistical regression and the 2026 Chicago Bears?  Tell us in the comments below!