Monday, August 02, 2004

The Steelers and the "bounce-back" in '04

This post is a follow-up to my post last Friday about whether the Steelers are actually rebuilding or if their 2003 6-10 record was just a fluke.

After writing what in retrospect turned out to be a rather long and convuluted post about the what the Steelers can expect in 2004 (as well as the rest of the AFC North for that matter), I revisited the numbers and noticed another disturbing trend (at least if you're a Pittsburgh fan).

I looked at the records of the four teams that make up the AFC North for the last 25 years (obviously the Ravens have only been around since 1996 and the Browns were an expansion team in 1999 -- there was no team in Cleveland from 1996 to 1998. In these two cases, only the years the teams were in existence are considered) and here's what I found:

Of the teams winning 3 or fewer games from one season to the next, their third season they average only 6.7 wins (for example, if in season 1 the Steelers win 11 games and in season 2 the Steelers win only 8 games, then on average they would only be expected to win 7 games their third season).

Of the four teams in the AFC North, the Ravens have the best "bounce back" record (let's call "bounce back" a measure of how well a team does after having a bad season. And a bad season is defined as losing 3 or more games one season when compared to the previous season). Specifically the Ravens, in their seven-year existence, have had only one season where they won three fewer games than they did the season before (in 2001 they won 10 games and in 2002 they won only 7 games). However, they were able to "bounce back" to 10 wins in 2003.

The results aren't so rosy for either the Steelers or the Browns (of course, no one cares about the Browns) -- both teams won three fewer games in 2003 than they won in 2002. The Ravens and the Bengals actually improved substantially in 2003 when compared to their 2002 records.

Over 25 years, the Steelers bounce back numbers look grim (remember, bounce back just means how many games they won after winning three fewer games last season when compared to the season before). Specifically, they only win (on average) 7.3 games the following year. The Browns win an even more dismal 4.6 games (and for completeness, the Bengals win 7.5 games).

What's interesting (or maybe disturbing) is that of the three different approaches I've looked at in terms of how the Steelers will fare in 2004, I've gotten similar results (8.5 games won using the time-series model; 8 games won using the binomial distribution; 7.3 games won using the bounce back statistic). What you also have to keep in mind is that "guessing" eight wins is about as safe a prediction as you can make. It's seldom that a team wins as many as 13 or as few as 3 games and usually the majority of teams end up somewhere in the 9 to 7 win range.

Nonetheless, the three methods I've examined have the Steelers winning between 7-9 games, and in all likelihood they'll win 8 games (with the caveat that I'm not considering really important stuff like player personnel moves, injuries and performance. I know, I know, these are important, but gathering the data aren't conducive to quick, witty posts -- Ok, at least give me quick), but as I've said several times before, I'm banking on a 16-win season (don't forget, the chances of the Steelers winning 13 games in 2001 was roughly 5% -- so anything's possible).