compliance.violation, compliance.resolved and the monitor events. Each is a JSON body with an event field. Webhooks explains when they fire and how to set up an endpoint.
Events
compliance.violation
Sent when at least one rule is newly breached. Each list carries only the new breaches: violations for holdings in a class outside its band, class_violations for the class totals themselves, and size_violations for holdings outside their size range. Any list can be empty.
Each breached row carries outside_by: edge says which side of the band or range the figure is on (above_max or below_min), and pct_points says how far outside it is, in percentage points of portfolio value. On a violations row it describes the holding’s class, so every row of one class repeats the same figure. outside_by has no dollar amount, share count or side to take. It’s a fact about your own rule, worked out from Allocait’s stored closing prices.
A violations row mixes two shares. current_pct and holding_pct are the holding’s own share of portfolio value (the same number), and class_pct is the share of its whole class. status, min_pct, max_pct (repeated as band_min_pct and band_max_pct) and outside_by belong to the class, so class_pct minus max_pct gives outside_by.pct_points, and current_pct minus max_pct doesn’t. priced_count and unpriced_count count the class’s holdings with and without a stored closing price, and prices_complete is true when none is missing. A class_violations row has the same counts, and its current_pct is the class’s share.
A band or size status allows a tolerance of 0.01 of a point. A figure within 0.01 of an edge counts as inside, and pct_points is reported as measured, without the tolerance taken off.
A row carries no price and no dollar figure, because the check runs on those stored prices. It keeps its status, the percentages it was judged on, and the band or size range. For dollars, run a compliance check with your own prices.
A size_violations row carries side, long or short, from the sign of the holding’s shares. A short’s current_pct is its size without the minus sign, and its range is the class’s short size range. A holding that stays outside its range but changes side between two checks is not announced again.
Illustrative example. The figures are hypothetical, not a recommendation.
compliance.resolved
Sent when one or more entries from the last check no longer appear in this one. All three lists are always present, and any can be empty:
resolved_symbolslists holdings that no longer appear inviolationsbecause their class is back inside its band, or the class no longer has a band. A holding that leftviolationswhile its class is still outside its band (sold out of it, for example) is not listed here. It is listed on the night its class is back inside, and the class is inresolved_classesthat same night.resolved_classeslists asset class ids whose total no longer appears inclass_violations.resolved_size_symbolslists holdings that no longer appear insize_violations.
Monitor events
monitor.confirmed, monitor.retraced and monitor.cleared report the daily close against a level you recorded on a monitor. monitor.confirmed means a close settled through the level. monitor.retraced means the day’s low or high went through the level but the close settled back. monitor.cleared reports that a close cleared a break that had been confirmed. Monitors explains the modes, the levels and what a break means. A NEAR monitor sends none of these events.
Allocait compares each new daily close with the level on the monitor, plus its confirmation buffer (buffer_pct, a percent). No screen or API sets a buffer, so it reads 0. A RANGE monitor uses both levels. consecutive_closes counts the closes in a row that have settled through the level, and it is 0 on a monitor.cleared event.
Each payload carries event, portfolio_id, monitor_id, monitor_name, symbol, comparison_mode, session_date, close, level_1, level_2 (null when the monitor has one level), buffer_pct, consecutive_closes and timestamp.
Illustrative example. The figures are hypothetical, not a recommendation.
session_date, close and levels and a new timestamp. Identify a repeat by event, monitor_id, symbol and session_date.
Next steps
Verify deliveries
Check the signature on each delivery.
Webhooks
When events fire, retries and endpoint setup.
Monitors
What the monitor events report.