Post-Run Expectations (CFBD → Odds API, Week N)

Source: docs/runbook.md

✅ What should exist in the DB (example: 2025, week 8)

Sources present: oddsapi for spread, total, moneyline.
Counts (order of magnitude):
Expect ~2.6k rows per type (spread/total/moneyline) for a full FBS slate with ~10–12 books. Exact numbers vary by book coverage and timing.

Quick checks (paste into Supabase SQL):

-- 1) Per-type counts by source
select season, week, line_type, source, count(*) as rows
from market_lines
where season = 2025 and week = 8
group by 1,2,3,4
order by 1,2,3,4;

-- 2) Totals by line type
select
  count(*) filter (where line_type = 'moneyline') as ml_rows,
  count(*) filter (where line_type = 'spread')    as spread_rows,
  count(*) filter (where line_type = 'total')     as total_rows
from market_lines
where season = 2025 and week = 8;

-- 3) Recent sample rows (closing_line fallback to line_value)
select
  game_id, season, week, line_type,
  coalesce(closing_line, line_value) as line_val,
  book_name, source, "timestamp"
from market_lines
where season = 2025 and week = 8
order by "timestamp" desc
limit 30;

-- 4) Moneylines look like American odds (-175 / +150)
select
  game_id, coalesce(closing_line, line_value) as moneyline_price,
  book_name, source, "timestamp"
from market_lines
where season = 2025 and week = 8
  and line_type = 'moneyline'
order by "timestamp" desc
limit 30;

-- 5) Book coverage by type
select line_type, source, book_name, count(*) as rows
from market_lines
where season = 2025 and week = 8
group by 1,2,3
order by 1,2,3;

Healthy example signatures:

Per-type rows: spread≈2679, total≈2696, moneyline≈2640 (source=oddsapi)
Books represented: DK, FD, Caesars, BetMGM, Bovada, BetOnline.ag, BetRivers, Fanatics, LowVig.ag, MyBookie.ag, BetUS, etc.
Sample row values show sensible numbers (e.g., spreads -17.5, totals 53, moneylines -1000) with recent timestamps.

What the UI should show

Home & Weeks tables

Market Close columns for Spread and Total populated.
ML column populated for many games.
Source badges like (ODDSAPI) next to numbers; hovering ML shows implied probability and a tooltip with timestamp.

Game Detail page

Moneyline card appears when ML exists (Price, Implied Prob %, Source badge).
If no ML exists, the ML column shows an em dash and the Moneyline card is hidden.

What workflows/logs should show

Odds Poll (3× daily) and Nightly Ingest:

CFBD step: "Parsed ~295 games for week N (FBS filtered)"
Odds step: "Found ~50–60 events" per fetch and parsed counts for spreads/totals/moneylines.
Chunked upserts logs like: ✅ Upserted XXX market lines in chunks of 500 and [DEDUP] Removed 0 duplicate lines…

One-Week Odds Test: same as above but only the requested week.

If results look off

Zero rows: confirm filters (season, week) and that odds were polled for that week.
No ML: rerun odds poll; some books delay ML. Verify with query #4.
Badges missing: ensure API responses include marketMeta.*.source; rebuild web if you just pulled changes.
Timeouts: workflows should target a single week; chunked upserts are already enabled.

Sanity math (why ~2.6k rows/type?)

~295 FBS games × ~9–11 books × 1 row per game/book/type → ~2,600–3,200 rows per type.