CSV and API

§1The two CSV files

One file for the board, one for a company’s history. Each opens from a CSV button on the page that shows the same figures.

A script can download the same files with your API key, sent the way the API takes it (§3.1). Each download counts against the key’s 60 requests a minute. Downloads from a signed-in browser have their own 60 a minute per account.

curl -H "Authorization: Bearer $VOLMODEL_KEY" \
  -o nvda.csv https://volmodel.com/export/t/NVDA.csv

§2Reading a file

Plain comma-separated text with one header row. Numbers are written raw so a spreadsheet can do arithmetic on them.

A move of 0.0841 means 8.41% of the share price, and a rank runs from 0 to 100; §4 gives the unit of every column. An empty cell means there is no value. We never write 0 or NULL in place of a missing number, because a zero would read as a measurement.

quality_flags holds the markers on that row, separated by semicolons, for example cover_mark_settled;iv_rank_noisy. Each one is explained under Markers, and a figure we left out says why in this column.

A text cell that would start with =, +, - or @ is written with a leading ', so a spreadsheet shows it rather than running it as a formula. Numbers are never changed this way.

The last two lines start with #. They carry the data notice and the method versions present in the file. Most tools can skip them when told to; in a spreadsheet they show as two rows at the bottom, which you can delete.

import pandas as pd

df = pd.read_csv("nvda.csv", comment="#")
df["expected_move_pct"] = df["expected_move"] * 100

Column names do not change. New columns can be added between existing ones, so read columns by name rather than by position.

§3The API

Three read-only endpoints that return JSON. Each request needs a key from an account on a paid plan.

§3.1Your key

On your account page, under API, press Create a key. The key is shown once, so copy it then: we keep only its first eight characters, which is how the page lists it afterwards. You can hold up to 3 keys and revoke any of them, and a revoked key stops working at once. Signing out of every device does not revoke keys.

Send the key in the Authorization header. The API does not read a key from the address, because addresses end up in logs and browser history.

curl -H "Authorization: Bearer $VOLMODEL_KEY" \
  https://volmodel.com/api/v1/board

import os, requests

r = requests.get(
    "https://volmodel.com/api/v1/tickers/NVDA/prints",
    headers={"Authorization": f"Bearer {os.environ['VOLMODEL_KEY']}"},
)
r.raise_for_status()
prints = r.json()["data"]

Each key may make 60 requests in each calendar minute. Every response carries RateLimit-Limit, RateLimit-Remaining and RateLimit-Reset (seconds until the minute ends). The plan is checked on every request, so if a subscription ends, its keys stop working on the next call.

§3.2Endpoints

§3.3Responses and errors

Every response is an object with four fields. data holds the rows. methodology_versions lists every method version among those rows (see Method versions); when it has more than one entry, rows computed under different versions are not strictly comparable, and each row carries its own methodology_version so you can tell them apart. attribution and terms_url carry the data notice.

{
  "data": [
    {
      "methodology_version": 3,
      "report_date": "2026-08-26",
      "session": "AMC",
      "expected_move": 0.05868320610687022,
      "em_decomp": 0.05589207148524317,
      "rm_cc": 0.08737956691786697,
      "rm_intra": 0.07109518635645107,
      "crush": -0.34680822531355954,
      "crush_atm_intra": -0.3378151378301292,
      "crush_pred": -0.3500566047904069,
      "pred_is_backfilled": false,
      "pred_session_unconfirmed": false,
      "crush_pred_naive": -0.36255849622892683,
      "crush_cal_set": "2026Q3",
      "crush_error": 0.003248379476847363,
      "quality_flags": [
        "cover_mark_settled",
        "crush_expiry_substituted",
        "decomp_macro_window",
        "open_crush_gap_otm"
      ],
      "publishable": true
    }
  ],
  "attribution": "Options data licensed from third-party providers. Delayed and derived; not investment advice.",
  "terms_url": "https://volmodel.com/terms/",
  "methodology_versions": [3]
}

A request that fails returns a status and a JSON body such as {"error": "rate_limited"}, including an unexpected failure on our side (500, internal_error):

§4Units

The CSV files and the API use the same names and the same units.

§5Every field

Grouped as the pages group them. The grey line under each says which files and endpoints carry it.

The report

symbol

Ticker.board.csv · /board

name

Company name.board.csv

report_datesession

The day of the report, and whether it comes before the open or after the close.board.csv · SYM.csv · /board · /prints

event_status

confirmed once the company has announced the date, estimated before then.board.csv · /board

as_of

The session the row’s figures were read on, at 15:46 ET. A name whose options went unread keeps its last reading, so this can be older than today.board.csv · /board

publishable

true when the company has a public page on this site. Names with too few clean reports have none, and are still covered here./prints · /snapshots

date

The session of that day’s reading./snapshots

The priced move

expected_move

The move the at-the-money straddle prices, before it is split (§2.1).every file and endpoint

em_decomp

The report’s own part of that move, our headline figure (§3).every file and endpoint

em_decomp_loem_decomp_hi

Its error bars (§3.2).board.csv · SYM.csv · /board

em_decomp_se

Its standard error.board.csv

straddle_spread_pct

The bid–ask spread of that straddle, as a share of its middle price.board.csv · SYM.csv · /board · /snapshots

near_oi

Open interest in the contracts nearest the money.board.csv

The contract

event_expiry

The expiry the priced move is read from: the first one after the report.board.csv · SYM.csv · /board

event_strike

The listed strike nearest the price at that expiry, where the straddle is priced. Empty on readings before 23 October 2026.board.csv · /board

iv_e1

The implied volatility of that expiry, the report included.board.csv · /board

spot_capture

The share price the straddle is priced over, read at 15:46 ET.board.csv · /board

straddle_mid_usdstraddle_sellable_usd

The straddle in dollars per share: the middle of the quote, and what a seller is paid at the bid.board.csv · /board

breakeven_lobreakeven_hi

The strike minus and plus what a seller is paid: where a short straddle held to expiry starts to lose.board.csv · /board

Ordinary days

clean_vol

Volatility with the report taken out: the ordinary movement priced around it.board.csv · /board · /snapshots

iv_front

Price reading of the nearest expiry.board.csv · /snapshots

iv30

Price reading at 30 days.board.csv · /board · /snapshots

iv_rankiv_percentile

Where the 30-day reading sits in its past year (§6).board.csv · /board · /snapshots

iv_rank_n

Days in that window.board.csv · /snapshots

iv_rank_noise

How many rank points the width of the quote alone could move the rank.board.csv

clean_iv_rank

The same rank, for clean_vol. Empty until a name has 120 days of our own readings, so most rows have none yet.board.csv · /snapshots

clean_iv_pctileclean_iv_rank_n

Its percentile, and the days in its window.board.csv

ts_slopeskew_25d

The slope (price reading of the report’s expiry minus the 30-day reading) and the tilt (25-delta put reading minus call reading; positive is the usual case).board.csv · /snapshots

Richness and the forecast

richness_pctile

Where the report move sits among this company’s own past reports, 0 to 100 (§5).board.csv · /board

crush_pred

Our forecast of the collapse, written the evening before and never changed (§7).board.csv · SYM.csv · /board · /prints

crush_pred_naive

The textbook forecast, before the correction crush_pred applies. It never changes. Empty (or null) when there is no forecast.board.csv · SYM.csv · /board · /prints

crush_cal_set

The quarter whose correction crush_pred used (2026Q4). Each quarter’s correction is fitted only on reports graded before that quarter began; a quarter with no correction leaves the two forecasts equal. Empty (or null) when there is no forecast.SYM.csv · /prints

pred_is_backfilled

1 (or true) when the forecast was reconstructed for a past report rather than published before it. Keep the two apart when you grade us. Empty (or null) when there is no forecast.SYM.csv · /prints

pred_session_unconfirmed

1 (or true) when we could not confirm the report came out after the forecast was read. The record keeps these out of its averages and reports them on their own line. Empty (or null) when there is no forecast.SYM.csv · /prints

After the report

rm_ccrm_intra

How far the stock moved, to the first close after the report and to 10:00 that morning (§4a).SYM.csv · /prints

rm_open

How far it moved to the first open after the report.SYM.csv

rm_cc_signedrm_open_signedrm_intra_signed

The same three moves with their direction kept.SYM.csv

rm_t1_closerm_t2rm_t3

The move to the close one, two and three sessions after.SYM.csv

crushcrush_atm_intra

How far the price reading fell across the report, at the day-after read and at 10:00 (§4).SYM.csv · /prints

crush_assoldcrush_iv30crush_assold_intra

The other ways §4 measures the same fall.SYM.csv

crush_error

Actual collapse minus our forecast.SYM.csv · /prints

crush_error_open

The same, at 10:00.SYM.csv

skew_t1skew_t1_post

The tilt the day before the report and the day after.SYM.csv

Selling and buying back

mark_openmark_intramark_close

The price to buy back the straddle that was sold, at the open, at 10:00 and at the close, as a share of the stock price the day before (§8).SYM.csv

mark_open_askmark_intra_askmark_close_ask

The same, at the asking price: what you would actually pay.SYM.csv

mark_t2mark_t3

The buy-back price at the close two and three sessions after.SYM.csv

mark_expirymark_strikemark_source

Which contract those prices are for, and whether each came from a quote or a trade.SYM.csv

k_star_eventt1_tdays

The strike the priced move was read from, and the trading days its expiry had left.SYM.csv

iv_t1iv_t1_e1iv_t1_e2

The price reading the day before, and of the first two expiries after the report; the split starts from these two and uses later ones where they are listed.SYM.csv

skew_span_t1

How far apart the 25-delta strikes were the day before, as a log ratio.SYM.csv

On every row

quality_flags

Markers on the row (Markers): a semicolon list in a file, an array in JSON.every file and endpoint

methodology_version

The method version the row was computed under (§11).every file and endpoint

§6What you may do with the data

Personal and internal use.

The figures are derived from market data we license. The terms cover personal and internal use. To use the numbers in something you publish or sell, ask us first. Every figure is delayed: we read it at 15:46 ET and publish it overnight.