Reading Spring Training Numbers for the Yankees Isn't About Batting Average
You open Statcast, you filter for Cactus League or Grapefruit League games, and you stare at a table of numbers that look completely normal and completely useless at the same time. This happens to everyone who tries this. The key is knowing which columns to ignore and which ones to actually trust. The framework I use is straightforward. I pull data from three sources: the main Statcast page filtered to spring games only, Brooks Baseball for pitch-level spin and movement details, and the team's own media guide for any context about rest days or injury listings. I cross-reference the two public sources against each other because they don't always agree on pitch counts or game dates, especially for those exhibition games that start at odd hours due to TV scheduling. What matters most for the Yankees right now is the pitching depth pool. The roster is set, but the bullpen battle isn't. Guys like Clarke Schmidt, Luis Gil, and any late-cut relievers will define whether this staff actually functions in April. The metrics I track for them are far more specific than you might expect.
I look at strikeout-to-walk ratio within context, meaning I filter for appearances where the runner is on second or third with fewer than two outs. That is where spring games actually reveal something. Velocity maintenance across multiple batters in a single appearance is another signal. If a reliever throws 96 mph to the first hitter, 94 to the second, and 91 to the third, that is a red flag regardless of how many strikeouts he accumulated. On the hitting side, I am not interested in batting average. I track contact quality through Statcast's barrel percentage and hard-hit rate, specifically broken down by pitch type. If a guy is lighting up fastballs but swinging at sliders in the dirt, the spring numbers will not show that distinction unless you dig into the pitch-by-pitch data. That is where the actual work happens.
Common Pitfalls That Waste Afternoon
The biggest mistake people make is treating spring stats as predictive. They are not. A .400 batting average in March tells you nothing about May. What it tells you is that the guy is making contact against non-top-shelf pitching, which is valuable information about his mechanical approach but not about his regular season production. Another trap is ignoring the quality of competition. Spring opponents vary wildly. A pitcher who dominates Non-Roster Invitees and minor league guys is not necessarily ready for a full season. I always check who he is facing. If his opponent is a AAA called-up reliever or a fringe MLB player, that stat line needs a heavy discount applied. I also learned the hard way that spring games have irregular scheduling. The Yankees sometimes play back-to-back exhibitions against the same opponent on consecutive days, and the rotation order gets shuffled for pitchers. This means a player's stat line might be incomplete if you only pull data once. I schedule two data pulls per day during camp—one before the afternoon games and one after the evening slate—to catch everything.
Get the Full Details
Specific Edge Case I Deal With Regularly
There is a recurring problem with Statcast filtering where certain spring games are miscategorized or missing from the default view. Specifically, exhibition games that are considered "extended spring training" or scrimmages sometimes do not appear in the standard dashboard. I found this when cross-referencing my data against the official Yankees spring training schedule and noticing gaps where games clearly happened but no pitch-level data existed. The workaround is manual. I go to MLB.com's game log for each spring opponent, pull the date and opponent, and search for those specific game IDs within Statcast's advanced search. It adds maybe twenty minutes to the data collection process, but it prevents you from missing entire games where your target players appeared. Without that step, your exit velocity averages and swing data are systematically incomplete, and the conclusions you draw from them are wrong.
What Actually Predicts Regular Season Performance
Exit velocity trends are the closest thing to a reliable predictor. If a Yankee is consistently producing 95+ mph exit velocity on ground balls and fly balls alike during spring, that is a legitimate signal. The swing path is correct, and the power is there. Conversely, if a guy is posting high batting averages on weakly hit balls that fall in between defenders, that is regression waiting to happen. Strikeout and walk rates on a per-pitch basis are the second most useful metric. These are more stable than batting average because they reflect the underlying approach. A guy who walks at a 12% rate and strikes out at 18% in spring is showing the same discipline profile he will likely carry into April, even if his batting average fluctuates wildly. For the Yankees specifically, the catching situation deserves attention. Austin Wells and Jose Trevino both have spring stat lines that can be misleading if read at face value. Catcher framing numbers do not appear in standard box scores, so the offensive production behind the plate is all you have to go on. I track their pitch framing tendencies by watching film clips when available rather than relying on whatever proxy metrics exist, because the publicly available framing data is neither complete nor accurate for spring games.
Limitations You Need to Accept
This entire exercise has a fundamental flaw that no amount of data cleaning solves. Spring training games are not representative of regular season conditions. Pitchers are working at 85 to 90 percent intensity. Hitters are taking early pitches off if they have already accomplished their goal for the day. The competitive pressure is minimal, and the managerial strategy is experimental rather than optimized. The sample size is another issue. A full spring training roster might accumulate two or three hundred total at-bats across all exhibition games. That is enough to spot trends, but it is not enough to confirm them. Any individual player's spring stat line should be treated as a hypothesis, not a conclusion. For anything beyond broad trend identification, I recommend supplementing spring data with minor league performance history. The Yankees system produces prospects who have large sample sizes in AAA and AA, and those numbers are far more predictive than anything that happens during a Cactus League exhibition. Use spring training to understand mechanical adjustments and approach changes, not to project regular season outcomes.
The data exists. The tools to access it are free and publicly available. The difficulty is in knowing what to do with it, and the even greater difficulty is resisting the urge to pretend that March numbers mean anything in October.
