How it works
01. It ignores the box score noise
If a quarterback throws for 400 yards, but half of them came from tipped hail marys, you wouldn't expect a repeat performance. The algorithm does the exact same math for baseball. It ignores surface-level luck and looks strictly at contact quality and plate discipline—the physics of the play—to predict what a player will do tomorrow, rather than what luck gave them yesterday.
02. The gap is your trade signal
When a player's underlying expected metrics disagree with their actual stats, that disagreement is an actionable trade signal. Buy the player whose expected stats are better than their actuals. Sell the player whose underlying numbers are worse. The table shows you the exact differentials (in green and red) so you can spot positive and negative regression candidates at a glance.
03. History anchors the score
While the algorithm doesn't care about yesterday's luck, it does care about track record. A two-year historical baseline anchors every player's score. A proven veteran going through a two-week slump will hold their value, while a waiver-wire hero with no track record has to prove their underlying contact quality is elite before the algorithm fully trusts them.
04. The math is fully transparent
Skeptical that an unknown rookie is ranked higher than a consensus superstar? Click the exp link next to any Trade Value score. A diagnostic modal will pop up showing you the exact metric-by-metric breakdown, the expected vs. actual differentials, and the plain-English reasoning proving exactly why the algorithm scored them that way.
Baseline assumptions & limitations
- Standard 5x5 Roto Baseline: The algorithm assumes standard 5x5 categories. Saves are weighted heavily for closers, while Holds are treated as a secondary high-leverage metric. Adjust accordingly for your specific league settings.
- Injuries: The tool does not know about IL stints, pitch counts, or workload limits. A high score for a player who just tore their UCL is a fact about their contact quality, not a bug in the code. Context is yours to layer in.
- Sample size matters: Before ~150 PA or 40 IP, scores lean more heavily on historical baselines. As the season builds, the current-season signal takes over and trade comparisons become increasingly reliable.
- Baseball knowledge required: This is a sophisticated assistant built for serious managers. A high Trade Value is a starting point for your decision, not the whole decision.