Starting Pitchers Matchup Tool

[tfox1]

When will the HQ Pitchers Matchup tool begin this year? We already have 2 games in the books. Also, when will the weekly and daily guide columns begin? Thanks!

[RAY@HQ]

We didn't activate this stuff for the two games this week. And as you saw, the SPs didn't exactly work deep into the games anyway.

The Matchup Scores should start populating this weekend, as the projected starters for Opening Day and beyond filter into the tool.

We'll have Daily Matchups columns starting on Opening Day morning. And the two-start SP column will debut next weekend (3/31).

Also one more treat in store that we'll unveil, probably next weekend.

[Hoff40ma7]

Ray - Just as a quick refresher. Currently, the Matchup Scores are based on 2018 stats, correct? Is there any consideration for Spring Training stats, expected player growth, team/environment changes, and/or changes to repertoire/pitch mix? (for the record, I'm not expecting the moon here, I'm just generally curious) At what point do 2019 actuals start getting baked in? And in your professional opinion, at what point do we really start truly "trusting" these scores for 2019. I put the air quotes around trusting because I'm still using them, aka Dylan Bundy @ Yankee Stadium would have a pretty bad score no matter what stats you used.

[RAY@HQ]

It's a great question, and since I'm writing the Opening Day Matchups column for tomorrow, I had to go back and re-read the original research.

Spring training stats do not factor in. The system starts using projected stats, and shifts weight toward in-season stats based on batters faced thresholds. The tiers are: (TBF30=batters faced in last 30 days)

Tier 0: TBF30 < 10
Tier 1: 10 ≤ TBF30 < 50
Tier 2: 50 ≤ TBF30 < 100
Tier 3: TBF30 ≥ 100

I didn't explicitly test this when I validated the tool last year, but to me the threshold of 50-100 batters faced is the key one. And that should arrive pretty quickly... basically that's two normal-length starts, or three shorter ones. And since this system is using projections, where the old (pre-2018) system just used prior-year stats for the early-season, even these first couple of weeks numbers are pretty good. If you just go look at the distribution of scores for this opening week, they look very reasonable.