Previous discussion in this space outlined the vast advantages of tying wins and losses to a matrix of what constitutes quality starts. Doing so demonstrates how Jason Hammel of the Orioles, who sports a 7-4 record, in fact has provided his team with only five good chances to win in 14 tries. Conversely, The White Sox' Chris Sale, whose 5-5 record suggests middling performance, should be 10-2 based on the quality of his starts.
Here are some other insights that Quality Wins provides on starting pitcher records:
The Phillies' Cole Hamels is 2-10. He should be 10-5. His won-loss record not only tells us nothing about his performance, it completely misleads us.
The Cardinals' Lance Lynn is 9-1. He should be 9-5. He's avoided four losses because his teammates have bailed him out.
Here are some other Quality Win revelations with the pitcher's name, his Quality Win record in bold and his actual W-L tally in parentheses.
Bud Norris 10-3 (5-6)
Patrick Corbin 10-3 (9-0)
The Astros' righty and D'backs lefty have performed similarly, but W-L aficionados would perceive them quite differently.
Joe Saunders, 8-6 (5-6)
Yovani Gallardo 6-9 (6-6)
An above average Mariner has a losing record while a bad Brewer appears to be middling.
Stephen Strasburg, 8-5 (3-6)
Gio Gonzalez 9-5 (3-3)
Evidently the Nats' problem is more offensive than defensive.
Clayton Kershaw, 12-3 (5-4)
Jon Lester, 6-9 (6-4)
The
Dodgers' ace and the Red Sox' ace have similar records, but only
because one has been victimized and the other propped up, by his mates.
Brandon Morrow 3-7 (2-3)
Phil Hughes 7-6 (3-5)
Again, similar records; divergent paths.
Mat Latos 9-5 (6-0)
Shaun Marcum 3-7 (0-8)
The undefeated Red should have lost five times. The winless Met has pitched well enough for victory a few times.
Lucas Harrell 10-5 (5-7)
Ricky Nolasco, 8-6 (3-7)
A.J. Burnett 9-5 (4-6)
Marco Estrada 4-8 (4-4)
Estrada has the best record of the group despite the worst results.
This is based on a 2000-2013 standard, which includes a goodly chunk of the longball era. You might be thinking that had we limited the benchmark to 2011-2013, more of these wins would be losses, or put another way, a starter would have to pitch more innings per run allowed in order to notch the W.
Currently, this is the matrix:
Go at least five innings. Give up two or fewer runs.
Pitch 6.3 innings and give up three runs or fewer.
Pitch eight innings and give up four runs or fewer.
Pitch 9.7 innings and give up five runs or fewer.
Subtract three percent from the scoring and the matrix becomes this:
Go five innings. Give up one run or none.
Go at least five-and-two-thirds innings and give up two or fewer runs.
Pitch seven innings and give up three runs or fewer.
Pitch eight-and-two-thirds innings and give up four runs or fewer.
As it turns out, it doesn't make much difference. After recalculating the Quality Win records of eight guys listed above, a grand total of one win transformed into a loss.
The bottom line is, if you're addicted to pitching wins, here's a formula that hews closer to the quality of the start. And as these examples attest, it can make a big difference in how you perceive a pitcher.
Showing posts with label Quality Wins. Show all posts
Showing posts with label Quality Wins. Show all posts
17 June 2013
15 June 2013
A New Way to Credit Pitcher Wns
Pitching wins are a team event credited to an individual who may have actually inhibited a team win. They too weakly correlate to good pitching to serve as a proxy when pitching acumen is being analyzed.
Efforts to find a single statistic that encapsulates good pitching have proved elusive and have been completely disengaged from the ultimate goal of a starting pitcher -- winning the game. They've also become a tangle of higher mathematics incomprehensible to the average fan.
Suppose we could invent one simple statistic that credited pitchers when they pitched sufficiently long and successfully to give their team a better-than-even chance of winning? That would trump mere pitching wins and be easy to use. You can see the whole discussion here in Matt Hunter's piece in Beyond the Box Score. We'll call his stat Quality Wins.
We know that since Year 2000, teams whose starting pitcher goes at least six-and-a-third innings and allows three runs or fewer win their games more than half the time. It's the reason that six-and-three is considered a quality start.
We can expand that matrix, starting with at least five innings of work, to create a continuum of Quality Win performances:
Go at least five innings. Give up two or fewer runs.
Pitch 6.3 innings and give up three runs or fewer.
Pitch eight innings and give up four runs or fewer.
Pitch 9.7 innings and give up five runs or fewer.
Keep in mind that this is runs, not earned runs. Defense is an issue on every ball hit, not just the errors.
Let's examine how that might have worked in the games of of June 14:
In the Brewers-Reds tilt, decided in the 10th inning, neither starter was credited with a win, though Cincinnati starter Bronson Arroyo earned one. He got through two outs in the eighth inning and surrendered three runs. His opponent, Kyle Lohse, pitched six full frames but allowed three runs to cross the plate, saddling Milwaukee with a better-than-even chance of losing (which they ultimately did.) He would be charged with a loss.
Chris Sale threw an eight-inning complete game for the White Sox in their 2-1 loss to Houston, allowing five hits and one walk while fanning 14. Under the Quality Win system, he gets a win for his effort. His counterpart, Eric Bedard, also earns a Quality Win for lasting six frames of one-run ball.
No Quality Win for either starter in the Phillies' come-from-behind 8-7 win over the Rockies. Kyle Kendrick and Juan Nicasio escaped without losses despite four-inning, seven-run and six-inning, five run debacles. Under the new system, each is charged with an L.
In four other games the losing pitcher would get credited with a Quality Win and in the Indians-Nationals contest, where both starters were gone when the outcome was settled, each would be credited with a Quality Win.
In all, on this low-scoring Friday night:
Five starters charged with losses would instead get credit for pitching well and giving their team a good chance of winning.
Three hurlers who no-decisioned would instead get wins.
Three others with no decision would instead take responsibility for a loss.
The seamheads would justifiably complain about various shortcomings of this statistic.
All true. This is not a leap towards the Holy Grail, just a refinement that gives us won-loss records that relate to actual performance.
Since no pitching stat is foolproof, and since this is replacing pitching wins, which is nearly useless, the standard is not perfection but mere relevance and ease of use. Moreover, statheads can refine it to account for many of those issues. For example, a pitcher could be credited with only the percentage of a win he provides in every game he starts. And the stat could be adjusted for ballpark, defense, etc., just as others are. These enhancements would strengthen the "win" statistic but render it difficult to calculate. The result would be a fielding-independent stat represented not as ERA, as is currently the case, but as a won-loss record.
It's a small step, but it would expose the starters with great records despite lousy performances (I'm looking at you, Jason Hammel of Baltimore, 7-4, 5.24 with a 1.5 WHIP; would be 5-9 in Quality Wins) and exonerate high achievers with bad teammates (Chris Sale, 5-5, 2.43 and a 0.9 WHIP; would be 10-2 in Quality Wins).
Efforts to find a single statistic that encapsulates good pitching have proved elusive and have been completely disengaged from the ultimate goal of a starting pitcher -- winning the game. They've also become a tangle of higher mathematics incomprehensible to the average fan.
Suppose we could invent one simple statistic that credited pitchers when they pitched sufficiently long and successfully to give their team a better-than-even chance of winning? That would trump mere pitching wins and be easy to use. You can see the whole discussion here in Matt Hunter's piece in Beyond the Box Score. We'll call his stat Quality Wins.
We know that since Year 2000, teams whose starting pitcher goes at least six-and-a-third innings and allows three runs or fewer win their games more than half the time. It's the reason that six-and-three is considered a quality start.
We can expand that matrix, starting with at least five innings of work, to create a continuum of Quality Win performances:
Go at least five innings. Give up two or fewer runs.
Pitch 6.3 innings and give up three runs or fewer.
Pitch eight innings and give up four runs or fewer.
Pitch 9.7 innings and give up five runs or fewer.
Keep in mind that this is runs, not earned runs. Defense is an issue on every ball hit, not just the errors.
Let's examine how that might have worked in the games of of June 14:
In the Brewers-Reds tilt, decided in the 10th inning, neither starter was credited with a win, though Cincinnati starter Bronson Arroyo earned one. He got through two outs in the eighth inning and surrendered three runs. His opponent, Kyle Lohse, pitched six full frames but allowed three runs to cross the plate, saddling Milwaukee with a better-than-even chance of losing (which they ultimately did.) He would be charged with a loss.
Chris Sale threw an eight-inning complete game for the White Sox in their 2-1 loss to Houston, allowing five hits and one walk while fanning 14. Under the Quality Win system, he gets a win for his effort. His counterpart, Eric Bedard, also earns a Quality Win for lasting six frames of one-run ball.
No Quality Win for either starter in the Phillies' come-from-behind 8-7 win over the Rockies. Kyle Kendrick and Juan Nicasio escaped without losses despite four-inning, seven-run and six-inning, five run debacles. Under the new system, each is charged with an L.
In four other games the losing pitcher would get credited with a Quality Win and in the Indians-Nationals contest, where both starters were gone when the outcome was settled, each would be credited with a Quality Win.
In all, on this low-scoring Friday night:
Five starters charged with losses would instead get credit for pitching well and giving their team a good chance of winning.
Three hurlers who no-decisioned would instead get wins.
Three others with no decision would instead take responsibility for a loss.
The seamheads would justifiably complain about various shortcomings of this statistic.
- First, it fails to credit a pitcher any more for giving his team a 97.6% chance of winning (nine shutout innings) than for giving them a 50.5% chance (six and a third innings; three runs).
- Second, it is divorced from actual wins, so a pitcher could be credited with 20 "wins" even if his team ultimately lost every game. (And both starters could be credited with wins or charged with losses in the same game.)
- Third, it fails to take into account all the other elements that affect run scoring, like defense, ballpark, relief pitching, weather conditions and luck.
- Last, it has no predictive value because it doesn't examine how the pitcher performed, just what resulted.
All true. This is not a leap towards the Holy Grail, just a refinement that gives us won-loss records that relate to actual performance.
Since no pitching stat is foolproof, and since this is replacing pitching wins, which is nearly useless, the standard is not perfection but mere relevance and ease of use. Moreover, statheads can refine it to account for many of those issues. For example, a pitcher could be credited with only the percentage of a win he provides in every game he starts. And the stat could be adjusted for ballpark, defense, etc., just as others are. These enhancements would strengthen the "win" statistic but render it difficult to calculate. The result would be a fielding-independent stat represented not as ERA, as is currently the case, but as a won-loss record.
It's a small step, but it would expose the starters with great records despite lousy performances (I'm looking at you, Jason Hammel of Baltimore, 7-4, 5.24 with a 1.5 WHIP; would be 5-9 in Quality Wins) and exonerate high achievers with bad teammates (Chris Sale, 5-5, 2.43 and a 0.9 WHIP; would be 10-2 in Quality Wins).
Subscribe to:
Posts (Atom)