Beyond our fantasy baseball content, be sure to check out our award-winning slate of Fantasy Baseball Tools as you prepare for your draft this season. From our free mock Draft Simulator – which allows you to mock draft against realistic opponents – to our Draft Assistant – which optimizes your picks with expert advice – we’ve got you covered this fantasy baseball draft season.

Men lie. Women lie. Numbers don’t.

We’ve heard it countless times, but the truth is, numbers do lie. Well, they can when they’re reframed to us to fit a narrative or an argument.

When you’re doing your preparation for the upcoming season, whether you listen to a podcast, read articles, or buy a draft guide, you’ll have several analysts giving you breakouts, busts, and sleepers across all platforms. To form their arguments on said players, they’ll bust out some stats in support of their argument.

Often, you’ll see small samples offered in support of their arguments. And if the experts want to justify the claim, they’ll offer reasons for why those numbers spiked or declined. In those cases, you’ll read or hear a lot of “X player put together a good second half after a bad first half” or something along those lines.

We turn to first- or second-half splits like they are the end-all, be-all, but in truth, they’re just an example of cherry-picking.

Are the fantasy analysts misleading you? Not at all. They’re using numbers to support their stance, which is what they are supposed to do. But as the consumer, you should look at the other end of the argument, too, to see if data exists to support a case against the player. 

We ran a series here at FantasyPros, where three analysts made the case for, against, and provided the overall draft approach for the top-100 players. It forced us to make those cases even if we personally didn’t believe them.

Here are some ways you should approach your own research into arguments about first- and second-half splits.

2020 Draft Kit: View printable cheat sheets, sleepers & mistakes to avoid

Full-season results are more predictive

If you want to point out what a player did in the first or second half, that’s fine, but it’s better to look at what they did throughout the season. Any player can put together a two-week fluke — in either direction — that can skew the numbers in a small sample.

If you’re looking at a month-by-month breakdown, take it further and look at it week-by-week, then day-by-day. Try to find the why, not just the what. Was it pure regression — positive or negative — that caused the change in results? Was it a change at the plate or on the mound that a beat writer reported? Looking at the totality of the season can give you a better gauge on each player than a shorted sample.

Extend it out to include multiple seasons

Do you know what will paint an even better picture than a full-season sample? Try an even larger sample. The rule of thumb is to use three years as the sample size if applicable (majors or minors) since you’ll see how a player performs year over year and eliminate outliers as much as possible.

Often, a player can put together a good or bad season and dramatically change how they’re drafted in the next season. Again, if there is a noticeable change, find out the reason why. If a player had a huge year but also a .400 BABIP, you’re only going to leave disappointed if you invest in them to repeat their success. 

Getting caught up in recency bias

Recency bias is one hell of a drug, and it’s something that we all fall victim to in fantasy baseball. It’s especially true with first- and second-half splits. 

For example, look at home run derby participants. Every single year, there is a discussion over whether or not a player will be affected by participating in the contest. The analyst will look at past years and how players hit fewer home runs after the All-Star break, but they’ll ignore that a player got chosen to participate in the contest because of the early-season success that they had where they, more often than not, overachieved with their power. Elevated results due to a small sample cause players to be selected.

The same can be said about second-half performances as opposed to first-half performances and how we view the players after them. Look at 2019, for instance. Both Yu Darvish and Jack Flaherty put together pretty poor first-half stats per FanGraphs, but they ended up with phenomenal second halves to end the season:

Darvish
First half: 5.01 ERA, 1.34 WHIP, 26.5 K%,11.7 BB%
Second half: 2.76 ERA, 0.81 WHIP, 37.8 K%, 2.2 BB%

Flaherty
First half: 4.64 ERA, 1.23 WHIP, 26.4 K%, 7.9 BB%
Second half: 0.91 ERA, 0.71 WHIP, 33.9 K%, 6.3 BB%

Both were phenomenal, and managers who held on to or invested in either were pleased that they didn’t cut bait on them with the results they provided down the stretch.

If you looked at their ADP heading into 2020, both pitchers were being taken early, with Flaherty going as high as the late-second round. It begs the question, though, as to why the second half for both pitchers is considered to be the true outcome and the first half isn’t?

In Flaherty’s case, especially, with his microscopic ERA and WHIP, there’s no conceivable way that his second-half numbers are repeatable. The fellas at PitcherList have talked in-depth about him, and they took the time to do the leg work to explain the why for Flaherty’s second-half performance.

But you have to wonder, if Flaherty and Darvish’s seasons were reversed and their success came in the front half of the year instead of the second half, would the outlook be as sunny for either player? The answer is no, and it’s because of recency bias. 

Practice fast mock drafts with our fantasy baseball software

Subscribe: Apple Podcasts | Spotify | Google Podcasts | Stitcher | SoundCloud | iHeartRadio

Michael Waterloo is a featured writer at FantasyPros. For more from Michael, check out his archive and follow him @MichaelWaterloo.