Field Position and the Syracuse Win

Week 1 2026 — exhibits for publication

library(tidyverse)
library(arrow)
library(gt)
library(ggrepel)
library(scales)

# This file holds ONLY the three exhibits used in the written piece. The full
# analytical workbook lives in 2026_wk01_review.Rmd -- that is where the story
# gets found; this is where it gets published.
source("../R/04_storyline_engine.R")
source("../R/05_viz_style.R")

TEAM   <- "Syracuse"
SEASON <- params$season
WEEK   <- params$week

FIG_DIR <- sprintf("figures/%d_wk%02d_story", SEASON, WEEK)

hist <- read_parquet("~/cfb-data/processed/team_game_metrics.parquet")
cur  <- read_parquet("~/cfb-data/processed/team_game_metrics_2026.parquet")

fbs_teams <- unique(hist$team[hist$is_fbs_matchup])

# The reference class: FBS teams playing FCS opponents, all seasons on record.
ref <- hist |>
  filter(team %in% fbs_teams, !is_fbs_matchup, !low_volume,
         is.finite(off_avg_start_ytg))

# This week's peer group: the same situation, the same weekend.
peers <- cur |>
  filter(team %in% fbs_teams, !low_volume, !is_fbs_matchup, week == WEEK)

su <- peers |> filter(team == TEAM)

# Helper: yards-to-goal reads as field position for a general audience.
as_field_pos <- function(ytg) sprintf("own %.1f", 100 - ytg)

n_seasons <- dplyr::n_distinct(ref$season)
su_rank   <- rank(ref$off_avg_start_ytg)[0]  # placeholder, computed below
su_rank   <- rank(peers$off_avg_start_ytg)[peers$team == TEAM]

Exhibit 1 — The historical baseline

q <- quantile(ref$off_avg_start_ytg, c(0.10, 0.50), na.rm = TRUE)

# Computed, not hardcoded, so the projection can never drift out of sync with
# the table around it.
proj <- build_expectations(hist, TEAM, SEASON - 1, fbs_only = FALSE) |>
  filter(metric == "off_avg_start_ytg")
proj_expected <- mean(c(proj$proj_form, proj$proj_own), na.rm = TRUE)

tibble::tibble(
  Measure = c(
    sprintf("Typical FBS team vs an FCS opponent (median of %s games)",
            format(nrow(ref), big.mark = ",")),
    "\"Notable\" bar, set before kickoff (top 10%)",
    sprintf("%s — projected before the game", TEAM),
    sprintf("%s — actual", TEAM)
  ),
  `Yards to goal` = c(q[[2]], q[[1]], proj_expected, su$off_avg_start_ytg),
) |>
  mutate(`Field position` = as_field_pos(`Yards to goal`)) |>
  gt() |>
  tab_header(
    title = md("**Syracuse started 13 yards closer to the end zone than a typical FBS team does against an FCS opponent**"),
    subtitle = md(sprintf(
      "Average drive start | %d seasons of FBS-vs-FCS matchups (%d–%d), %s team-games",
      n_seasons, min(ref$season), max(ref$season),
      format(nrow(ref), big.mark = ",")))
  ) |>
  fmt_number(columns = `Yards to goal`, decimals = 1) |>
  cols_align("center", columns = -Measure) |>
  cols_align("left", columns = Measure) |>
  tab_style(
    style = list(cell_fill(color = "#FFF3E6"), cell_text(weight = "bold")),
    locations = cells_body(rows = grepl("actual", Measure))
  ) |>
  tab_source_note(md(paste0(
    "**", su_caption("2026 Week 1 | Syracuse vs New Hampshire"),
    "** — lower yards-to-goal means better field position."))) |>
  gt_theme_su()

Syracuse started 13 yards closer to the end zone than a typical FBS team does against an FCS opponent
Average drive start | 12 seasons of FBS-vs-FCS matchups (2014–2025), 1,332 team-games
Measure	Yards to goal	Field position
Typical FBS team vs an FCS opponent (median of 1,332 games)	66.6	own 33.4
"Notable" bar, set before kickoff (top 10%)	58.1	own 41.9
Syracuse — projected before the game	66.4	own 33.6
Syracuse — actual	53.5	own 46.5
Data: CollegeFootballData via cfbfastR | 2026 Week 1 | Syracuse vs New Hampshire — lower yards-to-goal means better field position.

Exhibit 2 — Third of 47 this week

fp <- peers |>
  mutate(is_su = team == TEAM) |>
  arrange(off_avg_start_ytg) |>
  mutate(team = factor(team, levels = team))

p_fieldpos <- ggplot(fp, aes(x = off_avg_start_ytg, y = team, color = is_su)) +
  geom_vline(xintercept = median(fp$off_avg_start_ytg), linetype = "dashed",
             color = SU$muted, linewidth = 0.4) +
  geom_point(aes(size = is_su)) +
  ggrepel::geom_text_repel(
    data = filter(fp, is_su),
    aes(label = sprintf("%s — %.1f yards to goal\n(own %.1f)",
                        team, off_avg_start_ytg, 100 - off_avg_start_ytg)),
    nudge_x = 2, nudge_y=2, size = 3.5, fontface = "bold", segment.color = NA,
    lineheight = 0.95
  ) +
  scale_color_manual(values = c(`FALSE` = SU$muted, `TRUE` = SU$orange), guide = "none") +
  scale_size_manual(values = c(`FALSE` = 2, `TRUE` = 4), guide = "none") +
  scale_x_reverse() +
  labs(
    title = sprintf("%s had the %s-best starting field position of any FBS team facing an FCS opponent in Week 1",
                    TEAM, scales::ordinal(su_rank)),
    subtitle = sprintf(
      "Average drive start, yards to goal — further right is better | %d comparable teams | Dashed line = group median",
      nrow(fp)),
    x = "Average yards to goal at drive start", y = NULL,
    caption = su_caption("2026 Week 1 | FBS teams facing FCS opponents")
  ) +
  theme_su(legend_pos = "none") +
  theme(axis.text.y = element_text(size = 8))

save_fig(p_fieldpos, "field_position_vs_peers", height = 6.5)

Exhibit 3 — What each starting position produced

raw_wk <- sprintf("~/cfb-data/raw/pbp_inseason/pbp_%d_wk%02d.parquet", SEASON, WEEK)

drives <- read_parquet(raw_wk) |>
  filter(pos_team == TEAM) |>
  group_by(drive_id) |>
  summarise(start_ytg = first(drive_start_yards_to_goal),
            pts = first(drive_pts), .groups = "drop") |>
  filter(!is.na(start_ytg), !is.na(pts))

drives |>
  mutate(bucket = case_when(
    start_ytg <= 50 ~ "1",
    start_ytg <= 65 ~ "2",
    TRUE            ~ "3"
  )) |>
  group_by(bucket) |>
  summarise(Drives = n(), Points = sum(pts), .groups = "drop") |>
  mutate(
    `Starting field position` = c(
      "New Hampshire's side of midfield",
      "Own 35 to midfield",
      "Inside own 35"
    )[as.integer(bucket)],
    `Points per drive` = Points / Drives
  ) |>
  select(`Starting field position`, Drives, Points, `Points per drive`) |>
  gt() |>
  tab_header(
    title = md("**Short fields produced the most points, but the offense scored from everywhere**"),
    subtitle = md(sprintf(
      "%s offensive drives only | %d drives, %d points | Not split by score margin or time of game",
      TEAM, nrow(drives), sum(drives$pts)))
  ) |>
  fmt_number(columns = `Points per drive`, decimals = 2) |>
  cols_align("center", columns = -`Starting field position`) |>
  cols_align("left", columns = `Starting field position`) |>
  tab_style(style = cell_text(weight = "bold"),
            locations = cells_body(columns = `Starting field position`)) |>
  data_color(columns = Points, method = "numeric", palette = SU_SEQ) |>
  tab_source_note(md(paste0(
    "**", su_caption("2026 Week 1 | Syracuse vs New Hampshire"),
    "** — excludes the fumble-return touchdown and the safety, which were not offensive drives."))) |>
  gt_theme_su()

Short fields produced the most points, but the offense scored from everywhere
Syracuse offensive drives only | 15 drives, 54 points | Not split by score margin or time of game
Starting field position	Drives	Points	Points per drive
New Hampshire's side of midfield	6	26	4.33
Own 35 to midfield	3	17	5.67
Inside own 35	6	11	1.83
Data: CollegeFootballData via cfbfastR | 2026 Week 1 | Syracuse vs New Hampshire — excludes the fumble-return touchdown and the safety, which were not offensive drives.

Figures written to disk

figs <- list.files(FIG_DIR, pattern = "\\.png$", full.names = TRUE)
if (length(figs) == 0) {
  cat("No figures written.\n")
} else {
  cat("300-dpi PNGs ready to drop into the piece:\n\n")
  for (f in figs) cat("  ", f, "  (", round(file.info(f)$size / 1024), " KB)\n", sep = "")
  cat("\nThe two gt tables render above; screenshot them from this page, or\n")
  cat("install webshot2 to export them as images automatically.\n")
}

## 300-dpi PNGs ready to drop into the piece:
## 
##   figures/2026_wk01_story/field_position_vs_peers.png  (278 KB)
## 
## The two gt tables render above; screenshot them from this page, or
## install webshot2 to export them as images automatically.