We plumbed the heart of controversy in examining the MVP award last week, and this week we'll pick through the season's numbers to look for the most numerically deserving pitchers for the leagues' Cy Young awards.

A review of past Cy Young winners on Baseball-Reference.com indicates guidelines for snaring one.

Starters have a strong advantage, getting 44 of the last 50 awards (one in each league for the last 25 seasons). With 21st-century tactics, a fully used starter will produce about three or four times as many innings as a fully used reliever. Innings are opportunities to create excellence, so getting one-third as much work makes it hard to elbow past very good starters.

Voters also favor starters who play on good teams. Over the last seven years, 13 starters have won. Nine came from teams that went to playoffs, and eight from teams that were so good, their win-loss mark was a winning record even deducting the record of their Cy Young winner. Four of those five exceptions were already-famous pitchers who had won before: Randy Johnson and Roger Clemens.

And voters give extra pieces of flair to starters with gaudy win-loss records. Starters better have at least 20 wins. If they don't have at least 20, they have to have won twice as many games as they lost — there has been only one exception among the last 17 winners (Randy Johnson, 1999, 17-9).

The pitchers who do well in the sabermetric stats like the ones on The Times' Tuesday table usually fare well in the actual voting, so we'll use Component earned-run average as the foundation for comparing pitcher quality.

I'll note in the table what other flair each pitcher has: likely to win 20 (w20), a chance to have at least twice as many wins as losses (2x), in contention for the playoffs (P), and was the team good enough to have a winning record even subtracting the candidate's record (G).

Two of these candidates are toast. Roy Halladay hasn't pitched since July and Kevin Millwood, even though he's the current leader in the retro stat ERA, loses a lot of flair for having a losing record (8-11). The pure sabermetric case has to give the award to Johan Santana. Compared to Colon, the Twin's CERA is a significant 7/10ths of a run better, he has pitched more innings, and the home park in which he starts half his games is more hitter-friendly than Bartolo Colon's. Still, I think the voters will reward Colon with the hardware for his wonderful season.

For the National League award, the focus is on three candidates: Roger Clemens from the Astros, the Cardinals' Chris Carpenter and the Marlins' Dontrelle Willis.

The voters love Clemens (seven Cy Youngs) and his CERA is almost inconceivably low. But his win-loss record is a slender 12-8, and to them that looks like a coffin full of cockroaches to a "Fear Factor" contestant.

Dontrelle Willis has the 21-9 record, the game-winning hits and baserunning to complement his pitching, and one of the most radiant personalities in pro sports.

Pedro Martinez is the stat monster who's getting the shaft because voters like pitchers on winning teams. His record is 15-7 for a lukewarm Mets team that blew four games they were leading when he left, costing him extra wins.

I think Carpenter's combination of personal achievement and his team's overwhelming regular-season superiority will put him at the top of the list.

Jeff Angus writes on the new baseball statistics, describing what they reveal about the game and how that affects the Mariners. He is the author of "Management by Baseball" and of the Web log at http://cmdr-scott.blogspot.com. He's a member of the Society for American Baseball Research, a worldwide group that fosters the study of baseball past and present. Reach him at jangus@seattletimes.com.