The Z Files: Winning Tendencies, Part Six

So far in the Winning Tendencies series, the focus has been on the draft and its repercussions. It's time to shift to end-of-season rosters. The 570 rosters from the final transaction period have been added to the database, providing a treasure trove of analysis.

Let's start with the basics. Keeping in mind a handful of season-long contributors were released the last few weeks for strategic purposes, an average of 15.5 players remained on a roster all season. Here's the breakdown based on league finish, showing the total number of players kept per team and then breaking the draft into thirds:

Not only did contending teams draft a better foundation as shown previously, they retained fewer of their drafted players. These results don't account for the number of individual moves, reflecting how hard each team worked in-season, but it does suggest the top teams were less married to their players, willing to cut bait sooner or adroitly releasing them at the end of the season, especially those with injury concerns. For example, Ketel Marte was only on 24 final rosters since he was out of action for most of the final two weeks.

For those curious, the most players replaced was 24, while one team finished with their original 30. The bottom two teams in each league dropped fewer players than everyone else, which is not surprising based on previous data showing they added the fewest stats to their squad during the season. Maybe a high-stakes entry fee isn't the ultimate motivator to keep active all season.

When players were picked up is important to the big picture, but that information isn't available, or at least it's not easily tracked for 570 rosters. That said, some player data can be presented, to exemplify the type of acquisitions that contributed most to competitive teams.

Here's the list of free agent acquisition picked up by at least 25 teams throughout the course of the 2019 campaign. They're ranked by how high their aggregate teams finished in the overall. Also included is their aggregate league finish.

By means of perspective, using this ranking system, Gerrit Cole finished on top with an aggregate team rank of 169.2 in the overall, with Ketel Marte leading the league rank with 4.2.

Adding more context, here's a rough breakdown of the type of players acquired. This is a loose designation, but it helps elucidate who was targeted.

Since the theme of this research is the tendencies of winning teams, here's a list of the top-100 players ranked by aggregate overall rank, picked up by at least 20 teams last year.

While there's a good deal of overlap, this table shows a shift in designation as the percentage of each classification is displayed.

Drawing conclusions on the starting pitcher data isn't viable since they were churned throughout the season. However, it's interesting to see contending teams acquired rookies and veterans nearly equally while overall, rookies were favored. This reinforces the notion it isn't necessary to take chances to win. Sure, top teams picked up prospects, but not as frequently as the rest of the league. Further, the better clubs fortified closers at almost double the rate of the rest. This corroborates data discussed previously.

Next is a table of 114 players released by at least 10 teams who were acquired by someone else later in the year.

Let's break them into the same classifications as above while also including how their drafted and final teams fared in the overall.

The differences in aggregate team finish aren't huge, so take this analysis with the appropriate grain of salt. Veterans were dropped by teams finishing a little lower than those acquiring them, as were rookies. Perhaps more patience was warranted, but again, the delta is small. Steals were acquired by slightly more competitive teams, making intuitive sense as previous data illustrated contending teams acquire pilfers in-season. The only designation drafted by better teams to be picked up by lesser squads is catcher, with the takeaway being better teams are more diligent cycling backstops until they find a reasonable pair.

Let's finish this installment by looking at painful drops. Using earnings isn't the best approach, since players called up in-season are treated as if they played all year and thus are penalized, but this section is more for entertainment purposes and drudging up bad memories (sorry).

Here are the players released and picked up at least 10 times each, earning at least $5 using conventional valuation methods.

There's narratives with each, but my favorite is Lucas Giolito. He was drafted by 14 teams with only three of those clubs keeping him, thanks to his 5.32 ERA and 1.50 WHIP after five starts. There were a lot of lucky squads that picked up some portion of his next 153 innings, during which he sported a 3.12 ERA and 1.01 WHIP with 198 whiffs and a dozen wins. At the time, dropping Giolito was defensible as he was coming off a 2018 season with the worst ERA among qualified starters. However, given recent major improvement in data analysis, mainly Statcast, a look under the hood would have revealed Giolito using a different pitch mix with promising underlying metrics (increased velocity and spin rate). It's easy to say now, but a little patience would have been huge. The lesson is to take advantage of the next-level analysis now available before giving up on a player.

Lance Lynn is another cringe-worthy drop. After five starts, he had posted a 6.51 ERA and 1.55 WHIP. Lynn's peripherals were a little better, with just eight walks and three homers allowed in those 27.2 frames, though 24 strikeouts were a little low. Lynn then registered a 3.24 ERA and 1.17 WHIP with 222 punchouts in his next 180.2 frames. The 14 impatient managers are likely still kicking themselves.

Here are a few honorable mentions.

Mike Soroka, along with a pair of receivers, Mitch Garver and Christian Vazquez, stand out. They weren't dropped by many, but those who did regretted it.

Thus concludes the research portion of the Winning Tendencies series. Next time, a review of the most pertinent findings will be discussed, with an eye towards the observations that transcend the NFBC Main Event. Sure, the data was derived from the 570 teams in last season's Main Event, but there are multiple takeaways germane to private leagues.