API reference · Research

Get calibration

GET/v1/research/calibration

How the venue's opening prices compared with settlement over every settled market the filters keep, by price bucket, with intervals.

On this page

Auth. An API token: Authorization: Bearer lb_… (create one). Signed-in web sessions work too.

Limits. One history query per call (500 a day on Free; unlimited on Builder and Standard).

Parameters

Query parameters

Name	Type	Required	Description
`speaker`	string	no	A canonical speaker key.
`series`	string	no	A venue series ticker.
`category`	string	no	A category.
`phrase`	string	no	A normalized phrase key.
`venue`	string	no	`kalshi` or `polymarket`.
`since`	integer	no	ms epoch, by settlement time.
`until`	integer	no	ms epoch, by settlement time.
`min_volume`	integer	no	Leave out markets with less volume.
`basis`	string	no	`mid` (the first two-sided mid, default) or `trade` (the first trade, which the caveats explain).
`buckets`	integer	no	Number of equal-width price buckets. 2 to 20. Default `10`.
`test_fraction`	number	no	Share of markets, by settlement time, held out as the later split. Default `0.3`.

Request

Terminal

curl "https://api.langbot.ai/v1/research/calibration?speaker=powell&basis=mid" \
  -H "Authorization: Bearer $LANGBOT_API_TOKEN"

Python

from langbot import LangbotClient

async with LangbotClient() as lb:  # reads LANGBOT_API_TOKEN
    cal = await lb.calibration(speaker="powell", basis="mid")
    print(cal["n"], cal["overall"])

TypeScript

const res = await fetch("https://api.langbot.ai/v1/research/calibration?speaker=powell&basis=mid", {
  headers: { Authorization: `Bearer ${process.env.LANGBOT_API_TOKEN}` },
});
console.log(await res.json());

Response

200 OK

JSON

{
  "v": 1,
  "basis": "mid",
  "buckets": 10,
  "n": 214,
  "by_bucket": [ { "bucket": 1, "lo_c": 10, "hi_c": 20, "n": 31, "avg_price_c": 14.6, "yes_rate": 0.161, "yes_rate_ci90": [0.08, 0.29], "gap_c": 1.5, "brier": 0.134, "p_value": 0.71 } ],
  "overall": { "n": 214, "gap_c": -0.8, "gap_ci90_clustered": [-4.1, 2.6], "brier": 0.151 },
  "split": { "test_fraction": 0.3, "train": { "n": 150 }, "test": { "n": 64 } },
  "trials": { "count_30d": 4 },
  "caveats": ["mid basis: the first two-sided mid"]
}

Every figure has its sample size and a 90% interval, with Benjamini-Hochberg control across buckets. Each query is written to your trials ledger. Measurements, never signals.

Errors

Status	`code`	When
401	`auth_required`	No token was sent.
401	`token_invalid`	The token was not recognised (mistyped, revoked or expired).
422	`validation_error`	A parameter failed validation; `detail[].loc` names it.
429	`quota_exceeded`	Today's history queries are spent; `quota.resets_at` says when they reset.
429	`rate_limited`	Over your request budget; wait `Retry-After` seconds.

Every error body is {"detail": …, "code": "…"}; branch on code. All codes: Errors.