The Z Files: Points League Pointers

As a fantasy baseball analyst, dipping my toes in all formats is necessary. However, with the National Fantasy Baseball Championship and Fantrax offering points-based contests, it's time to jump into the deep end.

The key to points scoring is understanding that ranking by raw points is a huge mistake. The proper method incorporates replacement level. Fantasy football gamers may recognize this as value-based drafting. The irony is, it's originator, Joe Bryant, is said to have brought the notion to fantasy football after using it in his fantasy baseball league.

In short, ignoring transactions for now, everyone is going to have the points produced by the lowest scoring player at each position on their total. In a league with no moves, if the worst pitcher scores 300 points, his owner gains nothing from him; every other pitcher earned more than 300 points. What matters is everything above 300. This adjustment is made for all the positions, using their respective replacement-level player. The result is a ranking based on useful points, not raw points.

Just like in rotisserie rankings, it's necessary to determine how to parse the player. This is the notion commonly referred to as scarcity. On paper, each position requires its own replacement level. However, it gets fuzzy when middle infield, corner infield, utility and multi-eligibility players are added to the mix. As I've been espousing for over 10 years (others are just catching on now), it's only necessary to account for catchers. That is, in almost all formats, you're fine breaking the hitting pool into catcher and non-catcher, yielding a three-portion pool when pitchers are included.

One of the issues with writing about points leagues is while the calculation of useful points is a common thread, the makeup of the resulting rankings differs from league to league as there's no single, standard scoring system. The aforementioned game sites all use different scoring. Even within the same system, the adjusted patterns differ based on the size of the league and constitution of the active roster. This will no doubt frustrate those of you in points leagues, since I'll soon reveal a trick, but it's only relevant for the format in the ensuing model. The combinations are infinite. Hopefully I can explain the process well enough for you to exercise it on your own league.

In almost all instances, the raw rankings and adjusted three-pool rankings are markedly different. Doing this adjustment is the easy part. The hard part is taking advantage of those not following suit. Well, it used to be, until I unearthed the trend I teased earlier.

To demonstrate adjustment process, I'm going to use my projections for the 2017 season as though I were playing in a standard ESPN points league. This is a 10-team format with 13 hitters (one catcher) and nine pitchers. As such, the 10th highest catcher, 120th non-catcher and 90th pitcher are all adjusted to zero useful points. The points associated with each of those players are subtracted from everyone else at the respective positions.

These tables demonstrate how vastly different the rankings are before and after the adjustment at different player pool penetration levels:

Before

After

Raw points favor hitters but post adjustment, pitchers are brought into each group of 50 until the pool is properly aligned at 220 players: 22 players multiplied by 10 teams, with each position pool having exactly enough.

Those following a raw points list will likely end up with mostly hitters, until they realize they better draft some arms. Those doing the adjustment no doubt find their cheat sheet with an abundance of hurlers at the top. If they keep taking a pitcher, the hitting pool thins very fast since everyone else is using raw ranks.

Admittedly, this was frustrating. I know doing the adjustment is the correct approach. However, because most leagues permit weekly, if not daily roster transactions, there needs to be a few spots dedicated to playing the matchups. It's bugged me for years; what's the proper balance between trusting the ranks and skipping down to a hitter?

Then it hit me. What if the adjusted rankings are wrong? I don't mean theoretically or mathematically, but rather in practical teams. What's the composition of the season-ending player pool within each segment, and how does that compare to what's projected?

Here's the table for final numbers for the 2017 season, using adjusted ranks:

Sure enough, the distributions are different. There are more hitters than projected, even after the replacement-level adjustment. For those wondering, I looked at data from 2015 and 2016 and the results are consistent. Projections routinely rank pitching a little high after adjustment. That said, the adjusted top-end pitching always outpaced the adjusted top-end hitting, so I don't want to give up that advantage. I want to continue to grab arms early, but I want to know when to shift to sticks.

Again, things differ by league size and scoring system, so the following adjustment is only apropos to standard ESPN leagues. For those skilled in Excel, the process is straightforward, from making the useful points adjustment to this.

The pattern observed for all seasons investigated is around 150 players deep, and the projected and final positional pools are in sync, or at least close enough not to worry about. Once you're that far into the draft, differences in player evacuation mask ranking inefficiencies.

This is strictly empirical, but adding 20-25 points to the pitching replacement level for the top 100 players, then dropping that to 15 points for the rest lowers the ranks of ample pitchers to even out the projected and final adjusted pools. The best part is the top pitchers remain ahead of the top hitters, even with the lower adjusted total.

I now have a cheat sheet meshing projections and likely outcomes. My team will begin with the elite arms, but once that tier is exhausted, the hitters and pitchers are evenly distributed so the roster isn't overloaded with hurlers or devoid of bats.

For those not wanting to go through this process, just about every points system I've encountered follows this trend. That is, after adjustment, pitchers float to the top of the list. It may seem counter-intuitive to most advice you read or hear, but focus on top pitching for your first couple of picks, move to hitting, popping in pitchers you favor as the rounds progress. You'll organically come to a point hitters and pitchers are clustered together to fill out your roster.

For what it's worth, while the numbers presented were based on my projections, I reran them with two other sources and the conclusions are the same.

While crunching the numbers, I observed some cool, albeit intuitive trends. Shoot, what's a few more tables between friends.

It's no secret offense has picked up the past couple of campaigns. It's reflected in projections, but not rotisserie values since hitters are compared to hitters, pitchers to pitchers with each allotted a budget amount that hasn't changed. In other words, hitting pricing isn't higher now than in 2015, while pitching isn't priced more cheaply. As an aside, the early NFBC satellite auctions are all running at a 63:37 split, but that's a subject for a different day.

Since hitting numbers have improved while pitching has declined, the raw points for batters should reflect that, and they do. Correspondingly, raw projected pitching points have decreased. However, it's adjusted points we care about. Here's the data from 2015:

Before

After

The distributions are almost identical. Just a couple of years ago, it was all about player evaluation and proper ranking, not so much about draft flow and deciding between a hitter or pitcher. The raw points of the hitters still bested those of the pitchers, so those using unadjusted cheat sheets still focused too much on hitting, but the point at which the distribution was equal occurred sooner, so those making the adjustment weren't staring at a roster with three pitchers and a cheat sheet with eight more at the top of the list.

There's still a lingering factor to consider. How does replacement level change with roster transactions? On paper, it seems like activating pitchers with favorable matchups should raise replacement, since we're really putting value on a role not a specific individual, and multiple pitchers will move through the the same roster spot.

Unfortunately, I don't know the answer. My sense is since not all activations turn out to be favorable, the adjustment wouldn't be that much, maybe equal in points to a win, since it's the win likely contributing the points that increase replacement level. I plan on tracking the weekly results in a couple of points leagues to see if this can be quantified.

Again, I apologize there isn't a one-size-fits-all strategy for points leagues. Proper ranking transcends all formats. After that, it's looking at how 2017 played out, but that assumes you have your projections/rankings from last season. That said, just comparing a raw ranking list (the points league equivalent to ADP) to a useful points list should give you an idea of what many others will be doing, helping you to formulate a plan. In the likelihood your adjusted sheet is populated with pitching at the top, draft a couple, then skip down, as previously discussed.

Next week: A look at early auction trends.