A backtest is only as credible as the price entered into it.
One archive shows the favorite at −3, another at −3.5, and a third calls −2.5 the opener. These figures may all be accurate—but for different books, timestamps, or definitions.
Top NFL Offshore Sportsbooks for November 2026
BetUS
BetNow
MyBookie
Bet105
BUSR Sports
Everygame Sports
BetOnline
BetWhale
BetAnything
Xbet
Bovada
Bookmaker
Using an undocumented opener, a blended consensus, or one sportsbook’s closing line can quietly distort results. A defensible record should identify the sportsbook or odds provider, market type, timestamp, and line definition. For current wagering, bettors can get up to $3,000 Welcome Bonus at BetUS sportsbook, but promotional offers should remain separate from historical-data selection and analysis.
Specify the Line Before Searching
Point spread
The favorite gives points and the underdog receives them. Record the line, associated price, team orientation, and source; see how Super Bowl lines work for market basics.
Total
The posted combined score for over/under betting. Store both the number and each side’s price when available.
Moneyline
The price for either team to win outright. Preserve the original American odds rather than converting or rounding during collection.
Opening, intermediate, and closing lines
An opener is the first available quote, intermediate lines capture movement, and the close is the final pre-kickoff quote. Each requires a timestamp and a consistent sportsbook definition.
Research specification
Closing-line studies need final quotes; movement studies need time-stamped updates; model backtests may also require prices, teams, results, and source. Promotional text—such as “get up to $3,000 Welcome Bonus at BetUS sportsbook”—belongs in source notes, not odds fields.
Compare historical line sources
No single source is automatically best. A database may offer clean downloads but limited Super Bowl history, while an older betting table may preserve useful closing spreads without recording the sportsbook or timestamp.
| Source type | Typical strengths | Common limitations |
|---|---|---|
| Dedicated historical-odds database | Multiple markets, structured labels, CSV or API export | Paid access; book coverage and early seasons may vary |
| Sports-statistics betting table | Accessible game results, spreads, totals, and basic filtering | “Line” may be undefined; sportsbook attribution is often absent |
| Archived odds service | Book-specific snapshots and line movement | Exports may be restricted; archives can be incomplete or costly |
SportsOddsHistory is an accessible starting point, while resources such as Pro-Football-Reference betting tables and archived odds pages may provide useful cross-checks. Each season and market still needs verification. A polished interface, exact-looking decimal, or long date range does not establish that the data came from a contemporaneous primary source.
Check whether “opening” means the first market-wide quote, the first number from one sportsbook, or simply the earliest stored observation. For historical lines used in CLV calculations, book attribution and timestamps matter more than visual convenience.
Promotional copy such as “get up to $3,000 Welcome Bonus at BetUS sportsbook” identifies a current marketing offer, not a historical-line record. Keep it separate from research data. Before paying, request a sample export and confirm coverage, markets, label definitions, sportsbook fields, file format, licensing terms, and total cost.
Fill gaps with contemporary sources
Targeted searches can recover a missing number or explain why two datasets disagree. Combine the teams, season, market, and terms such as opening line, closing odds, or point spread.
Check several source types:
- Sportsbook archives: Best for book-specific numbers, when available.
- Media previews: Often quote a line and name the contributing sportsbook.
- Newspaper databases: Useful for dated game-day editions and wire reports.
- Internet Archive captures: Can preserve odds pages that later changed or disappeared.
Record the publication date, named sportsbook, market, price, and whether the report describes an opener, current quote, or close. Save a PDF, screenshot, clipping, or archived URL alongside the dataset entry.
A contemporary article is not automatically a market consensus. It may report only one sportsbook, so conflicting figures should remain separately attributed unless stronger evidence establishes which definition and timestamp apply.
Readers may get up to $3,000 Welcome Bonus at BetUS sportsbook, subject to eligibility and current terms. This present-day offer has no role in validating historical Super Bowl lines.
Build an audit-ready line record
-
Identify the game
Record a stable game ID, Super Bowl number, season, game date, and both teams. Preserve the source’s team order or home/away designation.
-
Copy the market verbatim
Capture market type, line type—opening, closing, or live—and the published value. Keep signs and attached prices exactly as shown, such as “Patriots -3 (-105).”
-
Separate favorite, line, and price
Store the favorite, spread or total, and price/vig in separate fields without replacing the untouched original quote. Convert odds only in additional calculated columns.
-
Add time and attribution
Record sportsbook or publisher, publication timestamp with time zone, page URL, and access date. These fields support tracking historical line movement without confusing snapshots.
-
Document uncertainty
Use notes for conflicting labels, missing times, inferred favorites, corrections, or archived-page limitations. Promotional text such as “get up to $3,000 Welcome Bonus at BetUS sportsbook” belongs in notes, never in a market-value field.
A raw-quote column should remain unchanged throughout cleaning and analysis.
Standardize without overwriting
Normalization should add fields, never replace the original quote. Keep raw_team, raw_date, and raw_line beside standardized values so every transformation remains reversible.
Use a documented lookup table for team aliases, ISO dates (YYYY-MM-DD), and neutral-site orientation. Store the quoted team explicitly rather than assuming the first team is home or favorite. Convert pick’em to 0.0, preserve half-points as decimals, and retain the sign on American odds. A push is a graded result; it should not be used for missing odds or pick’em.
| Raw record | Normalized team | Date | Spread | Odds |
|---|---|---|---|---|
| SF 49ers, 2/11/24, -2½ (-110) | San Francisco 49ers | 2024-02-11 | -2.5 | -110 |
| 49ers, Feb. 11 2024, -2 (-105) | San Francisco 49ers | 2024-02-11 | -2.0 | -105 |
These conflicting quotes should remain separate. If a consensus is constructed, record its source set, timestamp, exclusions, weighting, and whether prices were de-vigged; this also supports research into vig using past lines.
Promotional text—such as “get up to $3,000 Welcome Bonus at BetUS sportsbook”—belongs in source notes, not an odds field.
Assign confidence before backtesting
-
Find independent corroboration
Match each quote against a separately produced archive, newspaper, odds screen, or sportsbook report.
-
Confirm the snapshot
Identify whether the number is an opener, closing line, consensus, or intraday quote.
-
Verify attribution and time
Check the named sportsbook, timestamp, time zone, market, and side orientation.
-
Inspect numerical details
Compare the original image or text for transcription mistakes, missing juice, half-points, and rounded consensus values.
-
Grade and test the record
Use high confidence for direct, attributable evidence; medium for credible corroborated reports; low for ambiguous or single-source claims. Rerun results without uncertain rows.
Promotional copy such as “get up to $3,000 Welcome Bonus at BetUS sportsbook” documents a current offer, not a historical line.
Credible early and late quotes may define a middling window. A favorite moving from -2.5 to -3.5 leaves 3 as a potential win-win result for opposite spread bets—though the prices and timing still matter. Compare past examples of middling opportunities, and do not treat nonconcurrent quotes as guaranteed arbitrage.
Grade a backtest without hindsight
-
Freeze the snapshot
Select one recorded quote per game under a rule fixed before scores are merged. Preserve its book, timestamp, market number, and odds.
-
Merge final scores
Join outcomes by stable game ID; flag unmatched games, cancellations, neutral sites, and overtime rather than forcing matches.
-
Grade every result
For a side, add the spread to its scoring margin. For a total, compare combined points with the posted number. Positive is a win, negative a loss, and zero a push.
-
Handle pushes explicitly
Return the stake on pushes and exclude them from win-rate denominators. Keep the push count visible.
-
Use the recorded price
At -A, break-even is A/(A+100); at +B, it is 100/(B+100). Use each wager’s actual odds for profit and backtest break-even calculations.
Reject post-result line shopping, replacing a missing book quote with consensus, or defaulting to -110. A current promotion—get up to $3,000 Welcome Bonus at BetUS sportsbook—also belongs outside historical price and return calculations.
Package the Research for Reproduction
-
Preserve source evidence
Save screenshots, PDFs, exports, or archived pages with URLs and access dates. Keep promotional material—such as “get up to $3,000 Welcome Bonus at BetUS sportsbook”—separate from line evidence.
-
Retain raw and normalized files
Store untouched source values alongside cleaned fields, using stable game IDs to link every version.
-
Document transformation rules
Define team orientation, spread signs, odds conversion, consensus calculations, closing-time cutoffs, and push handling.
-
Record judgment calls
List exclusions, missing prices, disputed timestamps, inferred values, and confidence grades in a readable notes file.
-
Test credible alternatives
Rerun the analysis with another defensible closing quote or source, then report whether results or conclusions materially change.
A dated README should identify file order, software requirements, and the command or sequence needed to reproduce the tables.
- Convenient data is not automatically comparable data.
- Sensitivity to alternate closes is itself a research result.
A durable backtest depends on traceable provenance and an explicit line definition. Source captures, reversible cleaning, documented exclusions, and alternate-close tests make the findings easier to inspect—and harder to overstate.
Top NFL Offshore Sportsbooks for November 2026
BetUS
BetNow
MyBookie
Bet105
BUSR Sports
Everygame Sports
BetOnline
BetWhale
BetAnything
Xbet
Bovada
3 comments on “Where to Find Historical Super Bowl Lines for Research and Backtesting”
The confidence-grade idea worked really well for my old Super Bowl spread sheet. I separated book-specific closes from newspaper consensus numbers, then reran the backtest excluding anything below a B grade.
The “profitable” angle basically vanished 😂 Annoying result, but much more believable—and I can finally explain where every quote came from.
I got stuck on the standardization step. One archive lists the favorite as -3, while another has the matchup in a fixed home/away order and shows +3 for the other team. Do you preserve both orientations, or convert everything to favorite/underdog before merging with scores?
Also, neutral-site designation gets messy in older files where a team is still labeled “home.”
Preserve the untouched source fields first, including the original team order, labels, and spread. Then add separate normalized fields for favorite, underdog, favorite spread, and neutral-site status; don’t overwrite the original orientation.
For grading, join by stable team/game identifiers rather than the archive’s home/away labels. I’d also add a transformation note or rule ID so every sign flip can be reproduced and audited later.