How to Check Line Movement Using Recorded Odds

To read line movement, compare complete odds for the same market at different times, use a consistent odds format, and remove the margin separately at each checkpoint. This reveals a change in the no-vig estimate derived from this bookmaker’s quoted pair—not a market-wide consensus or the true probability. Odds history alone cannot establish why prices changed, identify sharp action, or predict the result.

The example below uses three recorded DraftKings moneyline pairs for Carolina Panthers @ Atlanta Falcons. It separates what the archive actually recorded from explanations the data cannot support.

What must match before you compare odds?

The event, bookmaker, market, period, settlement rules, and outcomes must match. For spreads and totals, the threshold must also match; every timestamp needs an explicit timezone.

Check these details before calculating:

• Event: the same teams and game date.
• Bookmaker: one operator, rather than prices stitched together from different sportsbooks.
• Market and period: for example, full-game moneyline, not first-half moneyline.
• Settlement rules: the same conditions for deciding a winning outcome.
• Outcomes: the complete set of mutually exclusive outcomes from one snapshot.
• Threshold: the same spread or total when comparing prices.
• Time: the exact recorded timestamp, expressed in one timezone—UTC here.

Different bookmakers can offer different prices simultaneously. Combining their quotes would show a difference between operators, not necessarily movement at one operator. Mixing periods or settlement rules changes the bet itself, while mismatched timezones can put observations in the wrong order.

In this moneyline example, the archive’s line field is empty: there is no spread or total threshold.

What does the Panthers–Falcons archive show?

The selected records show Carolina becoming the favorite between the first and final recorded checkpoints. They cover DraftKings’ two-outcome NFL moneyline for Carolina Panthers @ Atlanta Falcons—not every intervening price change. The CSV has no period or settlement field, so it does not confirm those terms.

The official Falcons game center confirms the matchup and September 20, 2026 date. The archive lists the scheduled start as 2026-09-20 17:00 UTC.

Our example, table, and calculations use The Odds Gap’s third-party records tagged book=draftkings: the June and final archive points come from its NFL moneyline history CSV, and the intermediate point comes from its snapshot CSV. These are not an official DraftKings export or proof that a particular reader could place a bet at those prices.

Atlanta means Atlanta Falcons and Carolina means Carolina Panthers throughout the tables. All timestamps are UTC.

The first recorded quote is not a proven DraftKings opening line, and the final recorded quote is not the literal last market tick. The archive’s earliest_ts identifies its first captured quote. Its close_ts identifies its final captured pregame quote—here, about an hour before the scheduled kickoff.

The detailed export timestamps the final pair at 16:00:32.673571; the aggregate history rounds that timestamp to seconds. Other snapshots exist between these three selections, so the table does not establish a continuous, one-direction move.

Sampling intervals and possible gaps limit what these three checkpoints reveal: they do not show every price change or establish a continuous, one-directional move between observations.

What changes after removing the margin?

Atlanta’s proportional no-vig probability falls from 52.1739% to 41.7007% between the first and final selected records; Carolina’s rises from 47.8261% to 58.2993%. These estimates come from the selected archived DraftKings pairs using one de-vig method—not a market-wide consensus or measurements of true win probabilities.

For each side, first calculate the raw implied probability:

q = 1 / decimal odds

Then calculate the complete market:

• Book sum: S = q_Atlanta + q_Carolina
• Overround, in percentage points: 100 × (S − 1)
• Proportional no-vig probability, in percent: 100 × q_side / S

The following calculations use the quoted American odds above, without first rounding their decimal equivalents. Decimal odds are displayed to six decimal places and percentages to four; using the displayed decimal odds can produce small differences in the last digits. The calculation treats each listed pair as the complete two-outcome market.

For the first pair, the unrounded inputs can be obtained directly:

• Atlanta −120: q = 120 / (120 + 100)
• Carolina +100: q = 100 / (100 + 100)

Their book sum is 104.5455%, and the excess above 100% is 4.5455 percentage points. Those are different quantities. Overround also does not equal the bookmaker’s actual profit.

Using unrounded estimates, the first-to-final change is −10.4732 percentage points for Atlanta and +10.4732 percentage points for Carolina. That describes a change in the estimates extracted from these quotes, not proof that the teams’ true probabilities changed by exactly that amount.

Keep de-vig methods separate: the provider’s devig_fair_prob field uses the power method. This article uses proportional, or multiplicative, de-vig. Do not mix the provider’s ready-made values with the calculations in this table.

How can you reproduce the calculation with ToolsGambling?

1. Start with the three complete pairs and UTC times in the table above. Use Atlanta −120 / Carolina +100, then Atlanta +120 / Carolina −142, and finally Atlanta +130 / Carolina −155. Keep both sides of each checkpoint together; the article contains all the prices needed for the calculation.
2. Open the ToolsGambling margin calculator. Select American odds, two outcomes, and Multiplicative for proportional de-vig.
3. Calculate each pair separately. Enter Atlanta −120 and Carolina +100 and check the 4.55% margin. Then replace both prices with each subsequent pair. The calculator shows the margin and fair no-vig odds, not probability percentages.
4. Compare with the article’s proportional no-vig percentage table. Use the original odds pair and the earlier formula: calculate each raw implied probability as q = 1 / decimal odds, then S = q_home + q_away and the no-vig percentage as 100 * q_side / S. Compare the result with the table. Retain precision through the calculation and round only the final result; using rounded odds may produce a small discrepancy.
5. Record the context. Save the source, event, bookmaker, market, UTC time, original pair, and de-vig method. The CSV does not verify the period or settlement rules; mark them unconfirmed unless you have independently checked the operator’s rules.
6. Optionally audit the source records. The history CSV linked above contains the June and final archive points. The free snapshot CSV covers September 15–21, 2026, including the September 15 intermediate point but not June. Useful verification fields include book=draftkings, market=ml, side, earliest_ts, and close_ts, plus snapshot_ts and american_odds in the detailed snapshots.

ToolsGambling English margin calculator with Atlanta −120 and Carolina +100, American odds, two outcomes, and proportional de-vig

Our screenshot of the local English ToolsGambling calculator using Atlanta −120 / Carolina +100 and the multiplicative method. This is not a bookmaker screenshot or an archive capture.

The calculator cannot load the historical CSV or reconstruct missing quotes. A correct calculation still depends on selecting comparable source records.

What can—and cannot—you conclude from the movement?

You can conclude that the selected recorded prices imply a shift toward Carolina after proportional de-vig. You cannot establish the cause from those prices alone.

The archive does not disclose betting volume, bettor identities, roster information, internal bookmaker decisions, or price availability for an individual customer. News, injuries, and sharp action remain hypotheses unless supported by separate, appropriately timestamped evidence.

A single bookmaker’s quotes also cannot establish a steam move, which involves rapid, synchronized movement across multiple operators. Three selected observations do not prove a monotonic path between them.

The eventual game result is a separate fact. Even if it agrees with the direction of the price change, it does not prove the line “knew” the winner. Our guide to who sets sports betting odds explains general pricing mechanisms, not a verified cause of this particular move.

How does the check change for spreads, totals, and 1X2?

The principle stays the same: compare the complete market under unchanged conditions. What changes is the number of outcomes and whether the bet has a threshold.

• 1X2: include home win, draw, and away win. Omitting the draw produces an incomplete book sum and an invalid margin calculation.
• Spreads: compare prices at the same spread. A price change and a change in the spread itself are separate observations.
• Totals: include both sides at the same total. Different total thresholds are not simply different prices for the same bet.
• Live betting: analyze it separately from pregame movement. The game state is changing, so this pregame comparison cannot be transferred directly.

For your own future observations, ToolsGambling’s odds-movement tool provides a current live snapshot and a tracker for manually added prices from the point you add them. It is not a public historical DraftKings archive and does not retrieve this past sequence.

How is market history different from your own CLV?

Market history describes recorded quotes; closing line value, or CLV, compares your accepted price with the closing price of a comparable market. There is no personal bet in this example, so it does not calculate personal CLV.

That comparison requires your accepted odds and a defensible closing reference. The archive’s final observation about an hour before kickoff cannot be presented without qualification as the definitive closing line. See the betting journal and CLV guide for recording that comparison.

Neither a line’s direction nor one positive CLV observation proves profitability. The useful outcome here is narrower: you know which quotes were compared, when they were recorded, and how their comparable no-vig estimates changed.