(Note: all statistics are as of August 21st)
Earlier this month, Jayson Stark of espn.com wrote a column advocating that Aroldis Chapman, despite being a reliever, deserves the National League Cy Young award. In principle, I am opposed to the idea of relievers getting the nod for the Cy Young and my reasoning is simple; overall contribution.
Once upon a time I wrote, when comparing two candidates for pitching’s highest honor, would you rather have 250 innings from a pitcher with an ERA of 2.50 or 200 innings? If you had to give one of them an award, which would you give it too? All else being equal, the decision is easy, you give it to the guy who threw more innings. He matched the ERA of the other pitcher across a larger sample size and in my book that always equals a better performance.
However, in the case of Chapman, all else is not equal. His numbers this year have been quite otherworldly:
1.33 ERA
0.72 WHIP
4.3 Hits allowed per 9 Innings
16.5 Strikeouts per 9 Innings
Should he be considered? Absolutely. Should he win the award? That’s what I need to find out.
A while ago I developed a metric to attempt to quantify the best pitching performance in a season. It was based upon three components; Bill James’ Game Score, ERA+, and innings pitched. The first problem I had to overcome was the Game Score; it’s designed for starting pitchers and doesn’t work for relievers. My first iteration of my new and improved metric just plugged in relievers numbers as if they were starters. The results, predictably, were not good.
So I started fiddling (very technical term) with the Game Score metric and came to a method that I like quite a bit. Essentially, I removed all of the rewards for innings pitched and I took out the constant of 50 points that a pitcher starts with. Basically, all that was left was this:
Hit = -2 points
Unearned Run = -2 points
Earned Run = -4 points
Walk = -1 point
Strikeout = 1 point
At the point the next problem is that if we look at all of these statistics on a gross basis or on a per game basis, starters will fare far better than relievers. I solved this problem by calculating how many of each of those stats that a pitcher accumulates per 9 innings pitched and then I did the above calculations with those numbers.
If you crunch a few of those numbers, you’ll realize that for the vast majority of pitchers, that resulting number will come out to be negative. I then took this resulting number and compared it to the best and worst figures league-wide; this gives me a percentile rank of each pitcher. At this point, I calculated my metric by multiplying the percentile rank, the ERA+ value (divided by 100), and innings pitched. The results, to say the very least, were fascinating.
If the Cy Young voting went according to this metric, a reliever would win the National League Cy Young award… and the American League Cy Young award. These are the top five scores from the NL:
195.20 – Aroldis Chapman
183.88 – Johnny Cueto
158.51 – Clayton Kershaw
158.07 – R.A. Dickey
148.32 – Cole Hamels
And in the American League:
236.36 – Fernando Rodney
204.48 – Justin Verlander
193.82 – David Price
187.98 – Felix Hernandez
150.60 – Chris Sale
At this point, I feel the need to widen my gaze by quite a bit. Does this new method unfairly favor relievers over starters? The only way to look at that is to look at many, many more seasons and see how some of the best seasons of all time stack up against each other using this new method.
THE CY YOUNG ERA
The first Cy Young Award was handed out to Don Newcombe of the Brooklyn Dodgers and since then, 101 awards have been handed out for pitching excellence and thus far, only 9 of them have gone to relievers. To see how well my formula does over the course of Cy Young voting, I applied the above method to all seasons where a pitcher threw 50 or more innings from 1956 to 2011 and while some results were surprising, others were exactly what I expected them to be.
To give you an idea of what kind of numbers this formula produces, the overall average for all of the 15,125 pitchers in the Cy Young era was 81.24. The following are where a pitcher would have to perform in order to make it into various percentiles.
Top half (50th percentile) – 81.24
Top quarter (75th percentile) – 107.57
Top 10% (90th percentile) – 159.50
Top 5% (95th percentile) – 194.41
Top 1% (99th percentile) – 280.46
Below is a graphic representation of where this formula (cleverly named the “Murphy Score”) calculates all pitchers to be. My reason for this is simple; when I say that a reliever posted a score of 250.0, quickly referring to this chart will show you that that season was in about the top 500 of the Cy Young era. While that doesn’t sound great, remember that this sample has 15,125 data points which means a score of 250.0 would be better than 98.2% of pitchers in the past 56 years.
You will also notice while perusing this chart that there are a few data points on the extreme high end. Only 23 pitchers had a season that produced a Murphy Score of over 400 and of those, only 3 broke 500. The overwhelmingly vast majority of pitchers fell into the lower end of the scale while a few had truly transcendent performances and frankly, that’s about what you should expect.
But the question that I want to examine is how do relievers stack up against starters? To separate the relievers from the starters, I simply looked at the percentage of a pitcher’s appearances that were starts. If that was over 50%, I considered him a starting pitcher and if it was under 50%, he was a reliever. It’s a bit on the simplistic side but I had to draw the line somewhere.
As you can imagine by simply looking at my formula, the numbers will be heavily skewed towards starters. Of the 684 pitchers who produced a Murphy Score in excess of 200, 93.1% of them were starting pitchers. Say whatever you want, I believe that equal performance over more innings should be considered a better performance. Therefore, if two pitchers rank in the 80th percentile on the rate stats and both have an ERA+ of 150, the one that throws 200 innings should have a better Murphy Score (240.0) than the one that throws 50 innings (60.0).
BEST SEASONS EVER
Now that I’m finished on my high horse, let’s take a look at the best seasons in the Cy Young era. This table shows the top 20 seasons according to Murphy Score (MS) in the past 56 seasons with some of the more traditional stats that people are accustomed to.
As you can see, there is only one season from a reliever in the top 20 and it’s not even the one you might expect. Dennis Eckersley won the AL Cy Young award in 1992 but he led the league in Murphy Score two years earlier with one of the most dominant pitching performances of all-time. In fact, in 1990 Eckersley was one of just four relievers to ever lead the league in Murphy Score but none of those four won the Cy Young. However, of the nine relievers to win the Cy Young, only three finished in the top ten in Murphy Score and Steve Bedrosian in 1987 finished 96th that year.
One can easily poke fun at my methods and say that they are flawed but many analysts have looked at the Cy Young voters and found flawed (or outright missing) logic, at least as far as statistics are concerned. Recently there has been a bit of a shift in thinking amongst the voters and this analysis agrees; in the past three years, the Cy Young winners have finished first and second in Murphy Score.
THE VERDICT
Do relievers deserve to be considered for the Cy Young award? Yes, I believe they do. However, I personally will always give a nod to a starter for the sheer sample size. As far as I’m concerned, a reliever has to be great (or beyond great) in a season where no starting pitchers are dominating. The problem for both Fernando Rodney and Aroldis Chapman is the fact that there are starting pitchers (Justin Verlander, Felix Hernandez, David Price, and Johnny Cueto) who are pitching extremely well.
When handicapping the Cy Young races down the stretch, you need to also remember that the margin of error for a starter is much larger. If Rodney goes out and gives up three runs in an inning (hardly a stretch for a closer), his ERA would jump from 0.78 to 1.23, an increase of 57%. If his ERA+ were to go down 150 points (about where Aroldis Chapman is right now), his Murphy Score would go from 236.36 to 157.89. Justin Verlander would have to have a bad month for his rate to go down 79 points.
Relievers deserve to be considered but that doesn’t mean that they should be winning every other year. I just can’t support a guy who throws less than one out of every 18 innings his team plays.
