Method · cross_sport_dominance@v1
Cross-sport dominance
How far a player sits from the middle of his own sport's distribution, so that distance can be compared with a player from another sport. Two numbers, because they answer slightly different questions: a z-score is how many standard deviations above his peer group's average he finished, and a percentile is the share of that peer group he beat.
The filter, exactly as it runs
season_type = 'REG'
Parameters
nfl_qb_dropbacks | 250 |
nfl_rb_rushes | 120 |
nfl_wr_targets | 60 |
mlb_plate_appearances | 400 |
mlb_batters_faced | 400 |
Where this is arguable
- Peer groups are different sizes. Roughly 34 quarterbacks qualify in a season against roughly 205 hitters, and a small group has thinner tails, so the same z-score is harder to reach in baseball than the raw number suggests. This is the single biggest reason not to read a z-score gap as a verdict.
- Baseball plays 162 games and football plays 17. A full MLB season is close to a settled estimate of a hitter's true level, while an NFL season still carries a lot of luck, so more of a football z-score is noise. Read an NFL figure as the shakier of any cross-sport pair.
- The long season cuts the other way for teams. A baseball club has time to recover from a bad month and is rarely eliminated by one game, while a single NFL result can end a season. Individual dominance therefore buys a football player more leverage over his team's outcome than the same dominance buys a baseball player.
- A z-score assumes a roughly normal distribution. Baseball's wOBA is close to normal; per-play EPA in football is not, because a handful of plays carry enormous value. Where the z-score and the percentile disagree, the distribution is skewed and the percentile is the safer number.
- Qualification thresholds decide who is in the distribution, and therefore what average means. Raising the bar for quarterbacks removes the replacement-level starters and makes every good quarterback look less exceptional.
- Partial seasons scale the threshold down proportionally, so an early-season leader is being compared against a smaller, noisier field than a full season would give.
- This measures dominance over contemporaries, not greatness. A player in a weak era scores higher than the same player in a strong one, and this cannot tell the difference.
- It says nothing about positional value. A quarterback at the 95th percentile is worth far more to a team than a running back at the 95th percentile, and this deliberately does not adjust for that.
These are the objections I already know about. If you have a better one, or you think a different threshold tells a truer story, say so and I will re-run it with your numbers and publish what comes back.
Version 1. Older versions are never deleted, so a piece that citescross_sport_dominance@v1 stays reproducible after this definition is revised. Source: Data: nflverse (nflfastR), CC-BY 4.0