Re-evaluating the 69/31 Hitter/Pitcher Split in Auction Leagues
Whether one agrees with the practice or not, people in most standard 5 x 5 auction leagues in which I've participated or known about, spend a good deal more on hitting than pitching. This is odd on its face as pitchers count for just as many standings points as hitters - after all it's 5 x 5, not 7 x 3 which would be more in line with the distribution of dollars.
In years past, I used to assume more money was spent on hitting because it was more reliable. Pitchers seemed to get hurt more easily, and their performance seemed to fluctuate more drastically, making them less sound investments. But is that still the case?
After a debate on the SXM show and Twitter, Fangraphs' Eno Sarris weighed in on the topic
, arguing while top-100 overall ADP pitchers finished in the top-100 at year end at a slightly higher rate than top-100 hitters last year, it's been shown that pitching categories are harder to project accurately than hitting ones. This makes sense when you consider how much wins fluctuate based on run support and saves depend on team circumstances. (Although saves mostly confuse the issue if we're comparing hitters to starting pitchers).
Peter Kreutzer also weighed in here
, arguing hitters are a better overall return on investment. In three AL and NL only leagues, he totaled up the money spent, and it turned out they returned 88 percent of it by season's end. Pitchers, on the other hand, returned only 32 percent.
This is complicated by two factors, however: (1) Pitchers' poor returns were in part due to more of them generating negative values due to two ratio categories (ERA and WHIP), while hitters have just one (AVG); and (2) the results were drawn from AL and NL only leagues, rather than mixed where at-bats are scarce and low-end pitchers often get destroyed. In fact, Kreutzer discovered top pitchers (those who went for $20 or more) actually outperformed $20 hitters, 89 to 83 percent.
This makes sense because top pitchers - even if they underperform - are likely to do so due to injuries (which don't generate negative values in ERA and WHIP), and even if they have healthy off years like Justin Verlander
, Stephen Strasburg
or Cole Hamels
, they're still not likely to give you negative value.
Kreutzer also makes the point that there's a lot more "free loot" (unbought players who come off waivers to contribute) when it comes to pitching than hitting. That's undoubtedly true, but likely far more so in AL and NL only leagues than in mixed ones where more players with regular at-bats are still available. Also, part of the reason the percentage of the free loot (25%) is so much higher in only leagues for pitchers than hitters (5%) is the 69/31 split in the first place. In other words, spend more on pitchers and less on hitters, and those percentages move toward each other somewhat - even in only leagues.
Even so, it's clear from Kreutzer's and Sarris' work there are good reasons for an uneven split, especially toward the middle and lower ends of the pitching spectrum. At the top end, especially in mixed leagues, however, their work indicates the 69/31 distribution is probably excessive.
To that end I ran an experiment - I downloaded the Steamer projections
and turned them into dollar values. (I used those over Jeff Erickson's because while Jeff's are probably better for drafting - and will get much better as more information is incorporated into them during spring training - they also include his subjective views about players. The Steamer ones, if I understand it correctly, are derived from an algorithm, and should be more generic - which is better for my purposes here.)
First, I used a standard 69/31 split for a 5 x 5, 15-team NFBC-style mixed league, and I got the following values for the top-25 players:
These values seem a bit strange to me, too, but keep in mind it's a 15-team mixed, not a standard 12-team only league, so that should bump up the top players significantly. And don't worry about Puig, Pujols, Hosmer and Rizzo being so high - that's just how this projection system values them (Again, Erickson's, i.e., RotoWire's, is probably better).
But take a look at the enormous difference when we use a 55/45 split for the top-10 pitchers. (I didn't deduct from the hitters because any loss in their value is negligible when we bump up only the top-10 pitchers). Kershaw goes from 16th overall to third, Yu Darvish
goes from 22nd to fifth and Cliff Lee
goes from 24th to sixth.
Again, it doesn't really matter which pitchers we're discussing - if you have Strasburg over Lee, that's fine. But it does show how much the pitchers' projected stats are worth in a 55/45 environment. If you're drafting in NFBC, maybe it makes sense not only to take Kershaw in the first round, but Darvish and maybe your No. 3 starter too.