Fantasy Index

header banner img
Win here.


Roster Management Thoughts

Todd Zola reviews some of his approaches to in-season roster management.

Use a sketchy two start pitcher, or play it safe with a one start guy with a favorable matchup?

Is the newly promoted prospect worth my top waiver priority, or a hefty FAAB bid?

Do I play a hot hitter over a better guy that’s in a slump?

Decisions, decisions.

In-season management is a grind. Most of the decisions are common sense, but the others can be the difference between winning a championship and there’s always next year.

Here are some thoughts to help navigate the six-month trek.

The amount of time one spends researching players and moves is different. It depends on other priorities and the type of league. Some are more serious, while others are casual. The level of effort one expends is a personal decision. Just keep in mind that as hard as you think you’re working, someone is more diligent.

Pitching analysis is contextual. Points leagues are much easier to address. Just come up with a reasonable expectation based on matchup(s) and convert it to points. Rotisseries leagues are far more complicated. Of course, some of the decisions are based on team needs, but it’s still early, so let’s assume the goal is to be competitive across the board.

Relying on steaming pitchers is a viable, if not necessary ploy. The extent is league dependent. In 10-team Mixed Leagues, up to half of starts can be of the streaming variety. In 12-team Mixed Leagues, there are usually some spot starters available for pickup. There may be a couple each week in a 15-team Mixed League, but most of the decisions are between pitchers on your active and reserve roster. In AL and NL only formats, the choice is usually between a starter with a sketchy matchup and a solid reliever.

A weekly-league strategy relying on streaming two-start arms is flawed. In today’s landscape, there is an insufficient number of viable candidates. Last season, there were more starts on five days of rest compared to four. Not to mention, the MLB schedule was expanded to include three more off days, which eliminated three chances for a team to have a double dipper.

The focus should instead be on one-start hurlers with the favorable matchups. The problem is defining favorable. There is data detailing the strength of opposition, but it comes with many caveats. Splits facing left-handed starters are full of noise, even further into the season. Players are obviously inconsistent, but teams also incur slumps. For example, through April, the Atlanta Braves boated the second best wOBA in MLB. So far in May, they sport the seventh lowest. Are you comfortable streaming your starting pitcher against the Braves lineup?

Pitching decisions should incorporate the season’s worth of data, seasoned with the current lineup. If a key player or two is missing from a strong offense, it’s safer to deploy a starter against them. Similarly, a perceived weak offense could be more productive if players are returning from injury, or perhaps acquired in a trade.

Something to consider is the time of the season. Historically, scoring is down in April and May, then it picks up in the warmer months before falling again in September. This season, the average runs per game in May is lower than it was in March and April. This speaks towards being more aggressive with pitching decisions and trusting a starter more than a reliever. Build up wins and strikeouts without sacrificing ratios now. If you need to chase wins and strikeouts later, you’ll be doing it when ERA and WHIP are theoretically higher.

Don’t overreact to small samples is cliché. The conundrum is there is a plethora of next level data. In most cases, it remains descriptive. Sure, we know the exit velocity and launch angle of every batted ball. We know the mph, spin, and movement of every pitch. This allows us to reverse engineer every event, but it doesn’t always generate a definite outlook. The data is accessible, and it’s so cool. However, it’s often a mistake to apply it in a predictive manner.

An exception is using batted ball metrics to discern if a perceived hot or cold streak is real. This doesn’t change the fact that a streak can flip at any time, but knowing if a batter has been snake bit or lucky aids decisions. Hot streaks supported by a high HardHit% and high average exit velocity at least reflect the batter is displaying elevated quality of contact. It doesn’t reveal if he can keep it up, but trusting a player hitting the ball hard is safer than relying on a so-called hot player without the backing of solid batter ball metrics.

On the flip side, as of this writing, Matt Olson’s line is .209/.319/.367. His slugging is 22 points lower than last season’s on base mark. YIKES! That said, Olson is hitting the ball harder than he did last year; he just isn’t getting the same results. Part is he overperformed his skills last season, whereas he’s underperforming them now. Benching Olsen until he gets it going is a big mistake.

Personally, the most difficult aspect of roster management is the balance between wanting a reason to pick up a newly available versus doing so because if I don’t, someone else will. This decision is a lot easier if you have a couple of roster spots dedicated to churning. Even so, I’m wired to need a tangible cause for making a move, but in most cases, there simply isn’t enough data.

This is where the newfangled metrics come into play. Some are misapplying the information as rationale for making a move. They think they have a tangible reason, but it’s via flawed analysis.

I understand that waiting for proof is a fool’s errand. Forget that the sample may never be ample this season, someone else is riding the hype and making the move. The explanation is essentially fear of missing out. However, statistically, prospects (and other emerging players) end up failing more than succeeding. Sure, when you land on a gem, you greatly improve your chances of winning, but what about most instances where the player struggled? Many play in multiple leagues, so they just care about the positive results. Not everyone has that luxury, time or bankroll. Drafting Wyatt Langford or Jackson Holliday in one of 10 leagues is no big deal. For those playing in just one or two leagues, it’s a different story.

Apologies, but I don’t have the answer. It’s as much of a personal decision as driven by the numbers.

In general, my preference is proper application of available data. Play the probabilities, and if there is a greater than 50% chance the move plays off, go for it.


Fantasy Index