A bit of an advance apology; you're going to have to read on a bit to get to the actual rule. This is a blog, not a text book, so you don't get to read the title, skim the subheadings, and call it a day. Also, I'll get back to societal problems and the way we treat people with special needs in a few days. For now I want to post some topics that I've had in the works for a while* but haven't gotten to yet.
Super Bowls 51 and 52 were both great games, for totally different reasons.** For those who don't remember (or don't follow football***) during Super Bowl 51 the Patriots were down by 28 to 3 with only eight and a half minutes left in the third quarter. They came back to tie the game, then win in overtime.
That made me wonder: for any given deficit in a football game, how much time is needed for a team to have a reasonable shot at a comeback victory? To the Business Intelligence cave!****
First stop: ArmchairAnalysis.com. I obtained an awesome data set, play-by-play data for every NFL game (including playoff games) from the 2000 season through Super Bowl LI. And I do mean play-by-play. What happened in the play on both offense and defense, plus environmental data, and more.
Using Access and Excel, I added some measures and slicers to the main data, then set out to plot the relationship between deficits, comeback wins, and time. The primary question: given a deficit of X score, how much time is needed for the team that's behind to have at least a 50% chance of winning?
I pictured a result showing something like, "A team behind by 3 only needs 5 minutes in the game, a team behind by 10 generally needs 12 minutes, etc." However, the results quickly exemplified one of the first rules of business intelligence: your first business question is probably not the right business question.
Why the focus change? According to the pivot, 65% of teams that scored first went on to win the game. And even more significantly, if that first score is a touchdown, the win percentage jumped to a whopping 70%.
That's seriously heavy information. Of course, the first thing I did was re-check the data, make sure I hadn't made any mistakes with calculated fields. Nope. Sure looks like a team that starts with the lead has a serious advantage.
That fact changes the priority of my business question. I'm no longer concerned with the time needed to overcome a particular deficit -- I'm more concerned about the importance of the first score, and of making that score a touchdown.
The unexpected information also raises a new business question: if scoring first is so significant, why does the team who wins the coin toss almost always elect to kick first? Is it because the entire NFL grossly misunderstands their own statistics? (Don't be so quick to scoff that possibility -- go read Moneyball.)
Or is it because the situation is too complex to be explained by a single statistic? That's more likely the answer, and the subject of my next BI-related blog.
* Sure, you might consider a year a bit more than "a while," but...
** Despite the Minnesota Vikings not being in Super bowl 52, as they should have been. Sorry, Tim VandeSteeg, it was a good try!
*** Yes, I said, "football." You know, a game characterized by an oblong goal, very large men, and not being rugby. The round-ball game played with no hands is called "soccer," Europe.
**** Okay, BI isn't really that exciting. We work hard, but we rarely literally spring into action, and we don't have theme music.