How do you determine THOME's Projected Standings?
It's a five-step process.
- We start with last year's performance for each team. But not their won-loss record. And not the Runs Scored and Runs Allowed, but rather, their peripheral stats. We look at every teams' combined:
- Walks / Intentional Walks issued to batters
- Batters hit by pitches
- Home Runs by batters
- Total Bases by batters
- Stolen Bases and Caught Stealing attempts by baserunners
- Batters who grounded into double plays
- Total number of At Bats by the team's offense
- Hits allowed by pitchers
- Walks issued by pitchers
- Home Runs allowed by pitchers
- Total number of team Innings Pitched
- Having collected that data for every team, we calculate their BaseRuns, to attempt to decouple the team's true run scoring and run allowing talents from the cluster of lucky / "clutch" hits stringing together. If a team hits three doubles in an inning, they're sure to score at least two runs, and probably more. But hit those three doubles in three seperate innings and you may end up scoreless.
- We then look at all the players that have left the team in the offseason, and all the players that have newly arrived (to the best of our abilities. Team's plans with minor leagues sometimes throw a monkey wrench in here.) We then add batting runs and subtract fielding runs allowed for arriving batters, and the opposite for departing players. Same thing for pitchers, adding or subtracting their ability to avoid allowing runs. We base these additions and subtractions on statistical projections for the upcoming season.
- Having a new set of Expected Runs Scored and Expected Runs Allowed for each team for the upcoming season, we calculate an Expected Winning Percentage using the Pythagenpat Formula. That could be where we stop. But it isn't.
- Treating every game in the upcoming season as a roll of a weighted 100-sided die, we simulate the entire season 100,000 times. The outcome of each game is weighted based on each team's own Expected Winning Percentage. This gives us the Win Distributions you see on each team's page, and the average of the nine most common outcomes for each team is recorded as their Projected Record for the Standings page.
What are the "Vegas Number Confidences"?
Each team has a "Vegas Number," which is the number of wins that you can place a bet on a team either finishing with more wins than (the "Over") or with fewer wins than (the "Under").
Our Vegas Number Confidences represent the number of simulations of the upcoming season that resulted in a team ending up Over or Under their Vegas Number (controlling somewhat for outliers, we toss out any results more then two standard deviations off the mean).
So when we say "Oakland Athletics: 83.0 Wins (95.0% Over)", what that means is that in 95% of THOME's simulated seasons, the A's finished with more than 83.0 wins. This is a pretty high confidence. On the other hand, if we say "Philadelphia Phillies: 83.0 Wins (55% Under)" then in 55% of THOME's simulations, the Phillies ended up with fewer than 83 wins. But this isn't very confident. The Vegas Number is probably very close to where the Phillies will end up after the season ends.
The idea here is that higher confidences represent a more attractive wagering opportunity. They're the teams on which THOME and Vegas most strongly disagree.
What is IMPACT?
IMPACT stands for Individual Measurement of Performance As a Component of THOME. It's how we project the, well, impact that a player will have when leaving or joining a team.
How do we calculate it? Well, the details are what we call a "trade secret"... we won't go into specifics. But we start with WAR, and we normalize over a full season of performance. We do that for two reasons: 1, it lessens the volume-dependency of WAR as a statistic, and 2, IMPACT is intended to determine how a player impacts his team when he is in the lineup. So, when a player is not in the lineup, his IMPACT drops (or rises!) to 0.
IMPACT is runs-scale, so the number represents, essentially, how many runs over a full season that player will either help his team score or keep the other team from scoring. It's useful to compare pitchers to pichers, hitters to hitters, and fielders to fielders. It's not, however, all that meaningful to compare Mike Trout's Offensive IMPACT of 79.6 to Max Scherzer's Defensive IMPACT of 54.9.