Stats, the Enemy of Common Sense

I am writing as I watch the Redskins - Seahawks game (the post got lost in the review process for a week), hoping that the Redskins hold onto their 3-0 first half lead until halftime. Then, with any luck, Seattle will come back and beat Washington, thus ending Joe Gibbs' streak of consecutive wins in playoff games his team leads at the half.

Earlier today, I was watching Duke struggle with Clemson. Here's an astounding stat for you, Clemson shot 6-21 from the FT line. Yes, 28.6% (I rounded the repeating decimal) on free throws! Even more appalling than the Tigers efforts from 15 feet was Len Elmore's index for rating college basketball players. Unfortunately, I didn't have the tape recorder going. I need TiVo! Fortunately, I remember the gist of the index. Elmore made a comment that scoring average is not the only meaningful statistic. Thank you Captain Obvious! Elmore ingeniously (sense the sarcasm?) combined scoring average, rebounds, assists, blocks, and top 25 wins (multiplied by 3) to determine the best college basketball players. Duke players Sheldon Williams and J.J. Redick top the list, not necessarily surprising since they each got 15 points for their 5 wins against top 25 teams.

While Redick and Williams are probably two of the top college basketball players, contriving some index and pulling it out in the middle of the Duke game to show that they are the top two seems just a little, well, contrived. Is a player on a lesser team with similar numbers to either Redick or Williams less deserving of accolades because the name "DUKE" or "NORTH CAROLINA" doesn't appear across his chest? If Duke faces Miami and Redick and Guillermo Diaz have similar numbers but Duke wins, should Redick rate higher on people's charts because he has McDonald's All-Americans helping him out?

The major problem with Elmore's index is that he neglects a lot of statistics (FG %, 3-PT %, FT %, TO, etc.) and did not correlate the ones he used to wins and losses (I do not have any proof of this, but I'm assuming the weightings of 1.0 on all the player stats are fairly good evidence). What is the actual value of each point scored with regard to team performance? Is each point the same value as each assist, as Elmore is suggesting (even though he may not realize he is doing that)? Steve Nash's NBA MVP last year makes you think assists are pretty valuable. I doubt it is the other way around, i.e. because we think assists are valuable, Steve Nash won the MVP. Anyone who watches the Suns play knows how important Nash is to that team, and his value is seen through assists.

Statistics are fun to analyze, but in this era of fantasy sports, they are often misused. Statistics must be put into context to be meaningful, and this framework is often missing when stats are delivered on game telecasts, sports news, or sports talk shows. Roger Clemens had a very good year in 2005. However, if you just look at his 13-8 record, it's a solid season, nothing that blows you away. However, record is a team stat, incorporating what you do on the mound as well as how the offense does against the opposing pitcher. Additionally, wins can be lost if relievers don't hold onto your lead. Clemens had an ERA of 1.87, a 1.01 WHIP, and a 0.198 BAA. In 2001, Clemens was 20-3 with the New York Yankees. Was his 2001 season better? Probably not, his ERA was 3.51 and his BAA was 0.246. I doubt that Yankees team got shut out numerous times throughout the season like this most recent Astros team. Is Muhsin Muhammed (Chicago Bears WR) a worse receiver this year than last? Possibly. However, is it not possible that having Kyle Orton at QB and a conservative offense were more important to his decline in production? Does anyone really think Adrian Peterson (University of Oklahoma RB) was worse this year than last? Personally, I'd chalk his lack of production up to a high ankle sprain, one-dimensional offense (at least early in the year) and injured and inexperienced offensive line. Check out the second half against Oklahoma State is you have any doubts about his abilities.

Rebounds, steals and blocks do not tell the whole story when it comes to how good a basketball player is on defense. Like many things in sports, and other fields, it is hard to quantify defensive prowess using statistics, at least the ones that are readily available. Even adding opponent's points per game, shooting percentages, assists, and turnovers will not be enough, although it would invariably help. Hopefully, in the coming months we (my brother and I) can help create frameworks for analyzing the increasing mass of statistics available. Until then, try to use some common sense and take things with a grain (or pound) of salt, and watch the games. Often, what you see with your own two eyes is far more valuable than anything you can find on the stat sheet.