Bug: implied_total calculation reversed for home teams in ep_preprocess.R

[cschupbach]

Issue

In R/ep_preprocess.R[42-50], the calculation of implied_total in .preprocess_common_fields() reverses the logic for home and away teams. The result assigns the away implied total to both teams when the home team is favored.

Current output

Example using 2025_07_TB_DET:

game_id	posteam	posteam_type	spread_line	total_line	implied_total
2025_07_TB_DET	DET	home	6.0	54.5	24.25
2025_07_TB_DET	TB	away	6.0	54.5	24.25

• In nflverse play-by-play data, spread_line is from the home team’s perspective (+6 = home favored).

Expected behavior

• Home (DET): implied_total = (total_line + spread_line) / 2 = (54.5 + 6.0) / 2 = 30.25
• Away (TB): implied_total = (total_line - spread_line) / 2 = (54.5 - 6.0) / 2 = 24.25

Proposed fix

Replace the existing case_when() logic with:

implied_total = dplyr::if_else(
  .data$posteam_type == "home",
  (.data$total_line + .data$spread_line) / 2,
  (.data$total_line - .data$spread_line) / 2
)

[tanho63]

Hmm, that’s a great catch! Probably would need to retrain and backfill values, not sure when we’d be able to take that on