Every probable starter on BullpenEdge has a projection page with a short paragraph of numbers. This guide explains what each one means, how it is built, and how much weight to give it. For the definitions of traditional stats such as ERA and WHIP, see how to read pitcher stats.
The three rates: K%, walks plus HBP, and home runs
A projection starts with what happens per batter faced, because that is steadier than what happens per inning. Each page lists three rates:
- **Projected strikeout rate (K%):** the share of batters the pitcher is expected to strike out.
- **Projected walk plus hit-by-pitch rate:** the share of batters who reach on a walk or a hit batter, the two outcomes a pitcher fully controls without a ball in play.
- **Projected home-run rate:** the share of batters who hit a home run against him, which is the least stable of the three.
As an example, at the time of writing the page for Nick Pivetta (San Diego) shows a projected strikeout rate of 28.0%, a walk-plus-HBP rate of 7.8% and a home-run rate of 3.40%. Strikeout and walk rates settle faster than home-run rates, which is why the model leans on them most. The reasons are covered in what is FIP, xFIP and xERA.
Why the projection is not just this season's stats
A pitcher's season line is a small sample, and small samples are noisy. The projection blends several seasons, weighting recent ones more, and then pulls the result toward the league average by an amount that depends on how much information there is. The page reports this as the number of batters faced in the weighted sample: 1,783 for the Pivetta example. A pitcher with more batters faced moves less from the league average in a short stretch, and one with few batters faced is pulled harder toward it. The method is described in projecting pitchers with the Marcel method.
Expected innings per start
This is how deep into a game the pitcher is expected to work. It decides how many batters the starter faces and how many the bullpen faces, and bullpens are usually worse than starters, so a short-outing starter shifts value to the relievers. The Pivetta page shows about 4.6 innings per start, derived from his recent outings. See starters vs bullpens for how innings are shared in today's game.
The season line
Each page also lists the current season's starts, innings, strikeouts, walks plus HBP and home runs. It is context, not the projection itself. A pitcher with 10 starts has thrown too few batters for the raw rates to be trusted, which is why the page leans on the weighted history instead. See early-season small samples for why.
Statcast context
The page adds a few Statcast measures from the previous full season:
- **xwOBA allowed:** the expected weighted on-base average from quality of contact and strikeouts and walks, independent of defense and luck. Lower is better.
- **CSW%:** called strikes plus whiffs as a share of all pitches. Higher means more pitches that win the count.
- **Whiff rate:** swings and misses as a share of swings.
- **Barrels allowed and hard-hit rate:** how often batters hit the ball in the best (barrel) or hardest (95 mph or higher) way. Lower is better.
These do not change a pitcher's projection by themselves. They describe the quality of the pitches behind the results, which helps you judge whether a good or bad run of results is likely to continue.
Recent starts
The page lists the last few starts with innings and pitch count. This is where you see workload: a pitcher throwing 66 to 87 pitches per start is being used differently from one reaching 100. Recent pitch counts also tell you how likely a starter is to go deep in the next game, which feeds the expected-innings number.
How it feeds a game
The three rates and the innings estimate feed the expected runs for each team in a game. From there the projection model turns expected runs into win probabilities and totals, and we grade the results on the track record page. Browse every probable starter on the pitcher projections page.
What to remember
- Rates per batter are steadier than per-inning stats.
- Strikeouts and walks are more reliable than home runs.
- The projection is regressed toward the league average; a small sample moves it less.
- Projections are estimates and can be wrong; they are not betting advice.