Insight Into Ranking Discrepancies

We’re getting ready to post our March fantasy rankings. I believe they’ll roll out starting next week. RotoGraphs isn’t the only place to publish my rankings though. Over at RotoBaller, we’re on our fourth round of updates, having started in December.

To accompany the February round, we ran a series of rankings disputes. Players with wide discrepancies like Albert Pujols – 60 picks between the highest and lowest (me) rankers – were debated. The point was to show how different player evaluation approaches can produce different projections and rankings. Instead, I discovered that I usually agreed with the other guy. Our different rankings had nothing to do with the expected stats. The issue was our managerial preferences.

Sticking with the Pujols example, we both agreed about his expected production. Healthy Pujols posts well above average home runs and RBI with slightly below average numbers in the other three categories. Relative to other first basemen, he’s a solid core producer. With the steady upside comes chronic foot issues, an aging body, and a luck neutral low BABIP. He’s been doing the fly ball thing for awhile.

As I explained at RotoBaller, there are scenarios when I’d take Pujols 60 picks earlier than my #158 ranking. They involve a dire need of home runs and RBI at first base. In a typical snake draft, I’ll generally feel very comfortable with those categories by the time Pujols shows up on the queue. There are also late-round alternatives who I prefer in a relative sense. Tommy Joseph, Eric Thames, and Greg Bird come to mind. My rankings reflect when I would start to feel compelled to take Pujols.

To me, there are a few lessons here. When an expert releases his rankings, it’s difficult to infer exactly how he or she produced the list. For example, Jeff Zimmerman built custom projections and ran them through a system. That’s why you see some stark discrepancies with the RotoGraphs consensus in his rankings. As far as I know, he gave no consideration to ADP. It’s good to have that list and an ADP aware list.

Jeff’s relatively easy to interpret. How about a trickier example. Paul and I are both “down” on Rougned Odor. Speaking only for myself, I actually like Odor. A lot. However, I see considerable risk that he earns a bottom of the lineup role due to his free swinging and low OBP. I also worry he’ll be given a red light on the bases. I’d love to have Odor on almost any roster not featuring Robinson Cano or Trea Turner or Daniel Murphy. But I won’t ever reach soon enough to make it happen. Looking at the pool, I view second base as a good place to gain relative advantage in runs, AVG, and steals. Odor isn’t that player.

Above, I’ve implied that Pujols has to be a value for me to draft him in most settings. Seemingly, I’m basically unwilling to roster a fair value Pujols. This is a weird discovery – I’ll happily reach for some players while only considering others if they’re left on the board too long. And this isn’t a conscious strategy either.

Over the years, I’ve developed a rule of thumb: as league owner quality increases, the optimal risk profile decreases. In plain english, I’m more likely to draft a steady guy like Pujols in an expert setting like LABR than in a casual setting like my college league. Lower owner quality means it’s easier to recover from mistaken reaches.

The draft pool flattens around where you’d target Pujols. This is when managerial preferences come into play. After picking up the foundation of my power and run production, I usually find myself filling unresolved skill positions and pitcher slots around picks 100 through 160. Then I pivot back to finding my last few lead-footed power sources later in the draft. My rankings are obviously going to reflect my typical draft.

In summary, I learned four lessons from investigating ranking discrepancies at RotoBaller.

• Managerial preferences inform these discrepancies at least as much as statistical/scouting analysis.
• Since experts have biases, it’s important to use consensus rankings. Multiple consensuses if possible. And it doesn’t hurt to also have a list that completely ignores ADP.
• Some players might not be worth rostering at fair value. The rightness of this statement may be related to the quality of your league.
• Your typical draft cadence can and should inform how you value players. However, you also have to be prepared to alter your values if the early rounds don’t go as expected.