I’ve been a fan of Syracuse basketball for a long time. While other teams might take a 25-point lead and never relinquish it, any double-digit lead is massive for the Orange. So whenever I watch one of their games, I can get incredibly anxious, as ‘Cuse just never quite manages to pull away from their opponent.
Northwestern football has a similar effect on me. They even have a well-established moniker that well predates my fandom — the Cardiac ‘Cats. As I watched the game against Purdue a couple of weeks ago, I noticed the same statistic flash across the screen a couple times. The commentators kept telling us that Northwestern is tied with Navy for the most close wins in FBS play since 2006.
While I understand that 2006 is the year Pat Fitzgerald took over coaching the ‘Cats, why is 2006 the cutoff? The original Cardiac ‘Cats team was the 1996 team that played six close games and won five of them, one of which was against top-10 Michigan. But is playing six close games in a season a lot? Since 1996, have the ‘Cats actually lived up to their nickname? Do they actually play a lot of close games? Let’s run the numbers and find out.
Scrape! That! Data!
After some searching, I found this database of historic FBS/Division 1-A football scores from James Howell. The scores go through 2019, so it doesn’t include this year’s data. But that’s just fine for our purposes. I decided to include all current FBS teams and the Idaho Vandals, who dropped to FCS in 2017 but spent enough time at the FBS level for me to justify including them. As for the data, here’s what each entry contains: the team, season, conference and coach, the win-loss record and point differential, the number of close wins and losses, and most importantly, whether or not the mascot of each team is a cat. That’s a lot of information, so let’s break it down.
I’m going to assume you know what a win-loss record is. One thing to note about this is since sometimes a coach retires in the middle of the season, I decided to use whichever coach physically coached the most games. For instance, LSU fired head coach Les Miles after just four games in 2016, so LSU’s coach for that year is recorded as Ed Orgeron. Now to the stuff that isn’t obvious:
The point differential is the amount of points the team scored in a season minus the number of points they let up. A close game is any game that was decided by one possession, which is eight points or fewer (TD + two-point conversion). As for the mascot, there are 19 teams that are definitely cats. They include Wildcats, Bobcats, Panthers, Jaguars, Tigers, Nittany Lions (curiously, no regular lions) and Cougars. And there’s one timely team that is questionably a cat team — so let’s briefly discuss the Cincinnati Bearcats:
Also called a binturong, the bearcat is native to Southeast Asia. It’s neither bear nor cat (although it is very closely related to them), which is a pretty solid argument for why they shouldn’t be counted as cats. But how did this random South Asian pseudo-cat end up as the mascot of a university in Ohio, you might ask? According to the University of Cincinnati’s website, the nickname dates back to 1914. During a game versus the Kentucky Wildcats, a cheerleader made a reference to UC fullback Leonard Baehr by chanting “They may be Wildcats, but we have a Baehr-Cat on our side!” Cute.
Cincy won the game, and it took a while, but the nickname caught on, and the university eventually adopted the Bearcats moniker. Cincy has had a live bearcat that cheers the team on since 2008, but the true origin story means the name was originally a pun on the word wildcat. In fact, the original student newspaper cartoon that memorialized the game spelled it spelled Bear Cats, two words, and because that’s definitely not a reference to an actual bearcat, that imaginary animal is at least part cat. As a result of all this, I’m counting the Cincinnati Bearcats as a cat team, for a grand total of 20 teams with cat mascots.
So let’s run the numbers now. I did my data analysis in R, and I linked both the data and my code for data analytics nerds out there to check my work!
The best individual teams
I want to start by looking at close games. Is six a lot? I feel like it is, but let’s check.
Well, the graph is kinda helpful, but we’re not quite there. These distributions look really normal, which is a statistics term that means they’re really really predictable, and as a result, we can do fancy statistics on them.
Let’s use the standard deviation to figure out just how unusual the ‘96 Wildcats were. The standard deviation tells you, on average, how far spread apart the values of a dataset are, and when combined with a normal distribution, we get a really neat trick: the 68-95-99.7 rule. On any normal distribution, like the ones you see above, about 68% of all values will fall within 1 standard deviation of the mean.
So let’s do it. The standard deviation calculation is a little complicated to explain, so I’ll just tell you that the computer spit out a standard deviation of about 2.13 games. If we add to and subtract one standard deviation from our mean of 4.25, we get 6.38 and 2.12, respectively. That means that any season with between 2.12 and 6.38 close wins (so between 3 and 6 speaking practically) is not that out of the ordinary. That 1996 Northwestern team was at the higher range of normally anxious but unfortunately wasn’t really that special.
Other Northwestern teams, however, are more special. Since 1996, eight NU teams have played at least six close games, and three, 2004, 2009 and 2010, played eight close games. But how does that stack up against the rest of the cat teams? And what about the rest of FBS?
Pretty well, as it turns out. There have been 15 cat teams that played eight close games in a single season, and four teams — ‘03 Cincy, ‘11 Kansas State, ‘16 Ohio and ‘19 Pitt — with nine close games played. LSU and Pitt are the only schools that challenge Northwestern by having multiple entries on the list. In all of FBS, Northwestern is up there as well. Other competitors include Big Ten foes Wisconsin and Purdue, the aforementioned Pitt, Georgia, Wake Forest, and Miami (FL). But the winner, by far and away, is North Carolina, which has played three seasons of nine one-possession contests. The Tar Heels are the only team to play nine close games in multiple years.
But playing close games isn’t everything—you have to win them. The Cardiac ‘Cats didn’t get their nickname just by losing heartbreakers, after all. Only 33 times has a team played at least five close games and won them all. The luckiest of them won just seven, which happened three times. Two of those teams won the national championship. The luckiest of this list must be 2009 Wyoming, which played six close games and won all of them, while winning only seven total games. The unluckiest teams are just as interesting — only three teams have ever played six close games in a season and lost all of them.
The best overall teams
Let’s turn to the overall teams now. It’s one thing to be anxiety-inducing one year. It’s another thing to make a habit of it. So who has the most close games since the ‘Cats gained the nickname 24 years ago? Counting the games up seems like a decent idea until you remember there are teams like the South Alabama Jaguars, who only joined the FBS in 2012. To even it out for these newer teams, let’s look at the proportion close games they played. I’ve put the cat team data on the left and the data for all FBS teams on the right.
Proportion of Close Games Played
|South Alabama||40.40%||3||Notre Dame||43.14%|
|Florida International||36.60%||8||Eastern Michigan||40.64%|
|Kansas State||33.99%||15||Air Force||38.93%|
By this metric, NU is indeed cardiac, playing 122 close games, or 41% of its contests over the last 24 years. But it isn’t at the top of the list — it’s five close games behind Pittsburgh, or about one and a half percentage points. Both NU and Pitt rank very high in all of FBS too, with Pitt coming in fourth and Northwestern sixth. Anxiety champion Coastal Carolina has played just three seasons in FBS, and small sample size could be a problem. We can look behind the Chanticleers to find UNC, which has played 129 total close games, good for 43% of its games since 1996.
Here’s that average close games per season graph again, now that we’ve seen the results on a long-term scale. I’ve added in some vertical lines this time. The light blue line represents UNC’s average, which I’m using in place of Coastal Carolina because of its small sample size; the purple line is Northwestern; and the dark blue line is Boise State, which has the lowest proportion of close games in FBS, with only 26% of its games (3.4 each year) being close.
So Northwestern has a cardiac competitor in Pittsburgh. If the Panthers are the most Cardiac ‘Cats, they’ll need a different alliterative nickname. I like the Pulsating Panthers™️. I asked my friend at Pitt, who suggested the Panic Attack Panthers, so pick and choose, Pitt. Just be sure to credit me when you’re done. Since this debate hasn’t been officially settled yet, I propose two ideas: using past head-to-head results to decide who’s actually more anxious, or simply starting a regular rivalry game against Pitt so we can settle it once and for all. And since Northwestern beat Pitt (in a close game, of course) back in the 2016 Pinstripe Bowl, I have no choice but to award NU temporary possession of the Cardiac Cat trophy until the rivalry comes to fruition.
So we’ve figured out who plays the most close games. But again, I must remind you that the Cardiac ‘Cats have to be good at winning games too. In other words, what proportion of close games does a team win? Here’s that table:
Proportion of Close Games Won
|Georgia State||42.31%||20||Brigham Young||56.52%|
Northwestern isn’t present in the top 10 of FBS schools by this metric, but no fear, they’re number 11! First-ranked Liberty has an insanely small sample size of just two seasons, so I’d be okay with saying NU is number 10 for now. What’s even more shocking is that NU is up there in the cat data with powerhouses like LSU and Auburn.
If you’re curious just how big of a difference it is, LSU and Auburn have combined for three national titles since ‘96. LSU’s win-loss percentage since then is 74.6% with Auburn’s at 65.7%. Northwestern’s, for comparison, is 51.5%. This suggests that LSU and Auburn underperform in close games, since their close games proportion is lower than their win-loss percentage, while Northwestern overperforms — the ‘Cats win more close games than their overall win-loss percentage would predict.
Since I know you’re wondering, the team that overperforms the most is Old Dominion, with an overall win-loss percentage of 45.9% that jumps to 63% in close games. The underperforming leader, by a wide margin, is Alabama. Since 1996, the Crimson Tide have won just under 75% of its games but wins just 48.4% of close games. In fact, the worst teams on this graph are all some of the best teams in college football, presumably because they blow everyone else out and then lose only in close games.
Backing up the anxiety
So far, I’ve shown you that there are quite a few teams that could deserve the cardiac cats moniker. Cincy, Kansas State, Ohio and Pitt have played the most close games in a season. Pitt plays the highest proportion of close games overall. LSU and Auburn perform incredibly well in tight games. Northwestern is fighting with all these teams for the top spot in every one of these categories but isn’t the top dog in any of them.
But when you aggregate these stats, NU’s Wildcats prove to be the truest of cardiac cats. I’m going to do that in the simplest way I can think to do so: proportion of close wins to total games played. Of all the games Northwestern has played since 1996, just under 25% of them ended in a close win, the highest proportion for a cat team and second highest in FBS. If you watch a Northwestern football game, there’s a one-in-four chance your heart will be racing until the very end as the ‘Cats narrowly defeat their opponent.
Proportion of Close Wins in All Games Played
|Florida International||15.98%||15||Wake Forest||20.96%|
This is irrelevant to the larger analysis, but of all the schools to be above NU, of course it’s that darn school from South Bend.
The Northwestern Wildcats toy with your heart like a tabby playing with a ball of yarn or a tiger stalking its prey. They don’t kill you quickly. They are patient. They will watch you agonize over every little mistake before finally doing you in. Except they don’t even do you in! More often than not, they just let you go! And then you come back the next week to do it all over again! Voluntarily! I guess what I’m trying to say is, we’re all masochists here.
In my research for this story, I couldn’t find any evidence that a sports fan has ever actually had a heart attack and died as a result of watching sports. There were plenty of studies about how it can lead to increased risk for high-risk people, but no reported deaths. I guess it’s good the Cardiac Cats are only a nickname.
Housekeeping, Acknowledgements and Data
Since I know someone is going to mention this: Yes. I know you’re not supposed to use standard deviation on a set of categorical variables. However, because the number of close games played is in fact numeric, we can still use things like percentiles to calculate exactly how strange a season is. And while it’s a far from perfect method to use the standard deviation to do that, I think the standard deviation gives a decent enough estimate to tell us the ‘96 season is kinda strange but not weird weird.
Second, this data is definitely not perfect. I think the initial assumption is that every close game is double counted, but that’s not true, because the data is only for when the teams actually played in the FBS level. So while the 2019 Northwestern-Purdue game will count as a close W for Purdue and a close L for Northwestern, the 2007 Appalachian State-Michigan upset will only count as a close L for Michigan since App State wasn’t playing in FBS at that point in time.
Third, I’m aware the definition of a close game is a little shaky. There are games that will be within eight points up until the very end, when a pick-six or lateral fumble returned for a touchdown on a potential game-winning drive pushes it out of the data set or something along those lines, like NU-Michigan State this past week. Sorry to bring it up. As a result, I admit I’m actually slightly undercounting the amount of close games. It’s unfortunate, but I wouldn’t even know where to start if I wanted to apply nuance to how each individual game played out. Plus, some garbage time scores probably even things out.
Fourth, when I began working on this, I thought I was going to have to manually fill in every single year of every single team, a process which my back of the napkin calculations told me it would take 30 hours to do. That sounded like a lot, so I found a better option. I’m indebted to my friend Nick Elkin, who spent a Thursday night on a discord call with me writing a python script to scrape the scores into something I could use. Without him, this analysis wouldn’t have happened. Thanks, Nick!