2010 NFL Playoff Predictions: Should We Really Choose Six New Teams?
I’ve read a lot of articles lately, such as this one, about the inability for half of the playoff teams from one year to make it back the next and the chances of teams with poor records to “rebound” the following season.
While it is a virtual certainty that some (and often times, many) different teams will make the playoffs in a given season, I disagree with the notion that it is rational to displace a talented team with a mediocre one in one’s playoff predictions simply to accommodate the “six new teams will make the playoffs” trend.
The reasoning is simple. For the sake of argument, let’s suppose that all of the playoff teams from last season really do have a 50 percent chance of making the playoffs again this year, while the non-playoff teams have a 30 percent chance of making it (which would be the case if 12 teams are each a coin flip). We’ll label last year’s playoff teams as P1, P2. . .P12, and last year’s non-playoff clubs as N1, N2. . .N20.
Which group of 12 teams is the most likely to make the playoffs in 2010? It is actually the same group as last season (in this hypothetical example). Of course, the chance of that exact group of 12 making it again is incredibly small. . .but that isn’t a reason to not predict it will happen.
The reason is that, while we can be fairly certain the group of playoff teams will contain some newcomers, we don’t know which newcomers it will be, and we don’t know which teams they’ll replace. To predict that a team with a 30 percent chance of making the playoffs will do so at the expense of a team with a 50 percent chance is simply bad math.
The logic is the same when you are filling out your NCAA brackets. Assuming the teams with the higher seeds are better (which is generally the case), the most statistically plausible bracket is one that picks every favorite to win. What are the chances this comes to fruition? Probably a billion-to-one. What are the chances that multiple underdogs win? A virtual certainty. So why pick all favorites? Because we don’t know which underdogs will win and, statistically speaking, choosing all favorites to win is the most likely scenario (as compared to all other individual game combinations).
Think about it. If we choose 62 favorites and one underdog (there are 63 games ), the chances of winning the bracket are just slightly less than choosing all 63 favorites. As we choose more underdogs, our chances diminish.
I really want to drive this point home, so here is another analogy. Let’s say you have a lopsided die that is just slightly weighted toward the number ‘six.’ Numbers one through five have a 16 percent chance of showing up on any given roll, while the number six has a 20 percent chance of coming up.
Now, let’s say you must predict the outcome of 100 future rolls of the die. What would you choose? Nearly everybody I know would select some “random” combination of numbers, perhaps with slightly more of the number six to account for the slight increase in likelihood of it coming up.
That prediction would be wrong, however. The most likely scenario (as compared to all specific alternatives) is that the number six comes up all 100 times. Even though you could roll the die your entire life and never see the number six come up even 25 straight times, much less 100, that is the statistically correct prediction.
People have a tough time believing this because the chances of six not coming up, at least once, are a lock. Why bet against certainty? The answer is because we aren’t betting that six won’t come up 100 straight times, but rather we are asked to predict a very specific combination of numbers. If we were asked to predict whether the die will show either six or ‘one through five,’ we’d obviously choose the latter, but we aren’t making vague forecasts.
The same is true in predicting playoff football teams. While teams obviously change from year-to-year and randomness has its affect on seasonal outcomes, the fact that the most likely group of 12 playoff teams is the 12 best teams (or the best in each division and the next two best in each conference, to be exact) is statistically irrefutable.
So will I pick the same 12 teams from last year in my 2010 predictions? Nope, because one season isn’t nearly a large enough sample size to determine a team’s level of talent. In fact, in any given season, the chances of the 12 best teams actually making the playoffs are almost zero.
Look at the AFC North last year, for example. Three teams (the Bengals, Ravens, and Steelers) all finished within one game of each other. There were probably thousands of seemingly random phenomena throughout the course of the season that could have altered the order of those standings. Maybe a crucial fumble took a weird bounce and ended up in one team’s hands over another. Maybe one team’s headsets lost power for a minute during a vital drive. Maybe Joe Flacco ate a bad meal the night before a game and consequently didn’t play well.
Whatever the reasons, the group of 12 playoff teams from 2009 is almost certainly not the best 12 teams from the season. In the die example, we know the probability of each number showing up. In football, we don’t know the probabilities. We must infer them from the previous season, personnel changes, and so on. If teams played 1,000 seasons, then I might have a different opinion, but we have just one to examine.
So when making your 2010 NFL predictions this season, don’t worry about the “50 percent rule” or the “NFC South loser always wins the division” rule. Instead, use a combination of statistics and your gut to select the six best teams from each conference, disregarding who “should” miss the playoffs this year because they made it last year.
Remember, the previous outcomes of random events have no bearing on future ones. Making choices using that methodology is no better than neglecting to choose the number six because it came up on the last roll of the die.