THE HOT CORNER: Let's call it conventional wisdumb

A favorite saying of baseball men and broadcasters is "good teams win one-run games." Katsuya Nomura trotted this one out last week in his latest pot shot at the Tohoku Rakuten Golden Eagles.

The Eagles, whom Nomura managed to a second-place finish last year, lost each of their first three games this season under new skipper Marty Brown by a single run.

Nomura, who now holds a spurious position as their honorary manager, was asked by a Yukan Fuji reporter if his old club might be better than its poor won-lost record because the losses came in close games.

Nomura, who has always preached the use of data to learn about the game, responded with the logic-defying piece of conventional baseball wisdom.

"When a good team wins, the games are close," he said. "When they lose, it's in one-sided games. For bad teams it's the opposite."

Huh?

If one extended that line of thinking, one might expect the best teams would wind up allowing as many runs as they score.

Two weeks ago, Nomura said the Eagles were lucky to do as well as they did under him in 2009. That seems an accurate assessment considering they finished second despite scoring 598 runs and allowing 609.

No matter what sport you are in, being outscored by your opponents is not a sign of quality. A team that allows 18 percent more runs than it scores, as the Eagles did, can be expected to win 49 percent of its games. The Eagles won 54 percent.

Nomura's 2009 team won 77 games and had a 22-17 record in one-run games. Does this mean they were lucky or good?

One could study this issue by looking at the best teams to see how well they do with different margins of victory. If one does this, one quickly realizes the truth is the exact opposite of Nomura's assertion.

Over the past four seasons, six teams have had winning percentages in excess of .590 (the Lions, Fighters, Dragons and Tigers in 2006, and the Giants in '08 and '09). At the opposite end of the spectrum were six clubs with winning percentages lower than .410 (the Eagles in 2006, the Buffaloes in '06 and '09, and the BayStars in '06, '08 and '09).

The six best teams went 135-98 in one-run games for a winning percentage of .579, while their weak sisters went 99-130 (.432) in similarly close games.

Nomura's 2009 team did well in one-run games last season but went 14-20 (.412 winning percentage) in blowouts (games decided by six or more runs). If losing one-sided games is a mark of quality, as Nomura hinted, then his Eagles were first rate last year despite having little in common with clubs that actually win lots of games.

The six good teams in the study had a .674 winning percentage in blowouts, while the six weaklings were .322.

Of course, Nomura might turn around and argue that having an extremely good record is no real indicator of quality. Take that illogic a step further and one might go as far as to say Nomura's 2006 Eagles, (.356 winning percentage) were in fact a good team just because they went 18-18 in one-run games.

Rather than being the best indicator of a team's quality, records in one-run games are the worst. That is where the records of good teams and bad ones vary the least.

The reason mediocre clubs often have good records in one-run games is obvious. The closer a game is, the greater the chance that a tiny amount of luck will turn it one way or another. These are the games in which a team is most likely to outplay the other but still end up losing.

That's why you'll never hear a manager who loses by 10 runs say: "We outplayed them for most of the game and didn't deserve to lose."

So why the fascination with close games?

It's probably because they allow for so much more drama. It's better entertainment to talk about thrilling wins and how clubs fight back from adversity.

Taking a five-run, first-inning lead and mercilessly shutting the other club down is more of a sign of quality but not much fun to talk about.

On the other hand, if one is in the business of shooting one's mouth off, sometimes a catchy line is better than a good argument.