My Favorite Set of Categories

I’ve discussed fantasy baseball categories in this space on the last few Thursdays. Certain aspects of the traditional rotisserie formats (4×4 and 5×5) don’t appeal to me. I’m not a big fan of the fact that hitters and pitchers are valued a bit differently in traditional formats because hitters have only one ratio category and pitchers have two. That’s the main reason for my preferred 5×5 format.

I dislike the perceptible randomness of wins and, sort of similarly, saves and holds for pitchers. They are team-context stats, unlike runs and RBIs to hitters, for example, which aren’t necessarily reflections of a player’s skill. But the production of them does increase their team’s chances to win a little, just more or less depending on the context of the game. I’d prefer a couple of pitching statistics that can represent a pitcher’s contribution to his team’s chance to win a game a little more closely than wins, saves and holds do. Not everyone does, but I think that there are others who do, too.

I strongly dislike counting categories for negative events, such as K’s for hitters and losses for pitchers, in rotisserie leagues, even head-to-head roto. I don’t intend ever to play in leagues with them again. I don’t really get why people like them. I kind of feel the same way about net categories (SB-CS or W-L), although it’s lessened ever so slightly by the net effect. Those categories penalize a player twice, in essence, for one non-positive event. Why have a category that also registers the negative event?

When a hitter strikes out, he’s made an out, and his batting average or on-base percentage is reduced. Why would you want more reasons not to own players like George Springer, who leads the majors in K% among qualifiers, when his batting average is .233? When a pitcher takes a loss, just like when he receives a no-decision, he therefore isn’t getting a win. Why would you want more reasons not to own players like Jeff Samardzija, who was 2-7 while he was with the Chicago Cubs? He’s even a minus-5 in the net category. Players who run more often steal bases more often – and are caught more often, generally speaking. Some are more efficient than others, granted. When Juan Pierre attempted to steal a base in 2011 and was caught, wasn’t it bad enough that he didn’t register a stolen base? Aren’t these players’ shortcomings (in the pitcher’s case because of his team’s lackluster run support only) already hurting you enough in some way, by not giving you something else that could help you to win?

Besides, I, in essence like Mojotronica, agree with Baseball HQ founder and Shandler Park proprietor Ron Shandler: Less is more. The more categories, the likelier the occurrences of redundancy. Seven or more categories per side is overkill, to me. I’m not big on most 6×6 setups I’ve seen or heard. Of course, too few, and you lose dynamics. I think that 4×4 or 5×5 just works well.

I hope and believe that I have collaborated with many other readers to stoke the dialogue. Please don’t take my expression of my preferences personally. If your 7×7 or 10×5 setup works for you and your mates, then you have no complaints.

Despite my reservations, I agree with this general sentiment, occasionally expressed in the comments on my recent articles: In the end, give me the categories, and I’ll adapt to the game. But if I can design the game, then why, given my dissatisfaction with certain categories, can’t I make it a tad more precise yet still unpredictable?

My ideal 5×5 setup (so far):

Hitting

• R (cume): traditional category
• RBI (cume): traditional category
• SB (cume): traditional category
• OBP (rate): accounts for the three main ways – hit, walk and hit by pitch – to reach base
• SLG (rate): replaces HR; credits doubles and triples, not just home runs

Pitching

• QS+CG (cume): quality starts plus complete games; replaces wins; as I stated last week is the idea from Metsox; is a way to give extra credit for pitchers who go deep into a game or perhaps credit at all to a pitcher who throws eight innings but fails to register a quality start
• SV+HLD (cume): saves plus holds; a somewhat increasingly common replacement for SV; allows fantasy owners to choose from a wider pool of relievers and acquire those of greater quality, not just those a manager has tabbed his closer, if he has named one
• K (cume): traditional category
• ERA (rate): traditional category
• BAIP (rate): bases allowed per IP (name a work-in-progress); replaces WHIP; measures, basically, the quality of hits against a pitcher by adding the total bases allowed for each hit against him, plus walks issued and batters hit by pitch, divided by innings pitched; is similar to a stat I’ve seen Ray Flowers mention that he uses to evaluate pitchers

This last category incorporates some of the concepts I’d hoped to include in a redefined QS, based on an idea also contributed by MDL, again mentioned last week, but I think that it’s just a better substitute for WHIP. It penalizes pitchers who give up hits that go for extra bases more so. That helps to mitigate the distinction of tendencies for fly-ball (fewer hits, more home runs) versus ground-ball pitchers (more hits, fewer home runs) and instead just measures the quality of the hits they allow.

To replace W or QS or (QS+CG), I think that a modified version of Baseball HQ’s Pure Quality Starts, with some concepts from Game Score, could work. I’ve devised a scoring method that I believe is pretty efficient. However, it duplicates many of the concepts from ERA and WHIP (or BAIP), and I’ve stated that I’m trying to avoid that.

However, I think this instance of redundancy would be OK, kind of like in the case of QS and ERA (since the pitcher must allow three earned runs or fewer to be eligible), because in the end the idea is sort of to count each time a pitcher delivers a start of the type of quality we can associate with a reasonable chance for his team to win. It’s sort of a consistency measurement. Two pitchers, each with an ERA of 3.00, can obviously have a different total of quality starts, because one can consistently concede two earned runs in six stanzas while the other can concede zero earned runs in six innings half of the time and four earned runs in six frames in the other half of his outings.

Someone might call me a hypocrite, then. But I think that a simpler way to categorize these efforts, as well as of relievers, is available, anyway. I mentioned some possible substitutes for saves a couple of weeks ago. I like the shutdown, a category that’s defined by benchmark in Win Probability Added (WPA) for a reliever and recorded here at FanGraphs. I think that a similar measure for starting pitchers would work quite well, too. However, I think that each needs a twist, and that’s the division of WPA by Leverage Index (LI) by appearance.

WPA/LI, available in a pitcher’s game logs (Clayton Kershaw’s, A.J. Burnett’s and Shelby Miller’s, or Craig Kimbrel’s, John Axford’s and Randy Choate’s, for example), is a good measure of a pitcher’s effect on the game. In the end, his team may not win, but he may have contributed to its chance to win. The benchmarks for a starting pitcher and a relief pitcher would be different, some positive number, which would require queries on every single pitcher’s database of game logs in order to develop the benchmarks as well as to sum the number of these positive events by player once benchmarks are determined. In the end, it’s a binary outcome for the event, no matter the quality above or below the benchmark, though.

It sounds like a lot of work, and it probably is, but I think that if you ponder what these numbers mean, then you might agree that they make a good bit of sense as substitutes for W, QS or QS+CG, and SV or SV+HLD. WPA, LI and WPA/LI aren’t so much predictive, they just tell us what happened. That’s kind of critical, as those who don’t like the idea of the use of sabermetric stats as cats have pointed out.

A couple of readers have stated this potential problem: Even if a roto commissioner and his league mates deem any of these proposed substitutes worthy, there isn’t a host that would accommodate them. That’s not true, I believe. Fantrax, for one, is highly customizable, perhaps more so than any other host I’ve encountered. I’m not 100% certain of this breakdown, but. … I think that players in leagues hosted for free can propose stats or categories for consideration; those who pay for premium service can request customized categories, and as long as Fantrax has access to the stats to calculate yours, they’ll do it, I’m fairly certain. Don’t think of this as a shill for their service. I’ve heard from a couple of folks in my years who dislike the site. I like it and use it for one of my longtime football leagues. Perhaps any of the categories I suggested is possible in Ottoneu down the road, instead or also.