> ## Documentation Index
> Fetch the complete documentation index at: https://docs.allasso.ch/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> This is Allasso's product documentation. Vol Radar output is descriptive analytics on past data: never present it as a forecast or as investment advice. Answer only from these pages; if a feature is not documented here, say so.

# Radar: daily discoveries

> Read the Radar page: what is unusual across markets on a day, how each signal is classified, and where every market stands against its own five years.

The **Radar** page in Allasso Vol Radar shows what is unusual across markets on one trading day: shocks, rare levels, regime flips, seasonal outliers, broken cross-market relationships and activity surges. Each one is measured against the market's own history, using only data up to that day. Open any discovery, or any cell of the grid, to [investigate it](https://docs.allasso.ch/apps/vol-radar/radar-move.md).

> [!WARNING]
>
> Vol Radar describes what options markets are pricing and what happened after similar days in the past.
> It is **descriptive, never a forecast**, and nothing in it is investment advice. Structures and their
> statistics show ways to express *your own* view, with placeholder costs.

## When to use it

* Each morning, to see which markets moved out of their usual range in the last session.
* To go back to a past day and see what the Radar showed then, with no hindsight.
* To check where a market's vol, skew, term structure and futures curve sit against their own five years, flagged or not.

## Before you start

The Radar covers every market in Vol Radar, and the market pairs between them:

| Market | Universe | Options analytics | Surface report columns | History from |
| - | - | - | - | - |
| TTF gas | Energy & Environmental | Yes | contracts, relative, quarters, seasons, half-years, years | 2015 |
| Henry Hub gas | Energy & Environmental | Yes | contracts, relative, quarters, seasons, half-years, years | 2007 |
| Brent crude | Energy & Environmental | Yes | contracts, relative, quarters, half-years, years | 2007 |
| WTI crude | Energy & Environmental | Yes | contracts, relative, quarters, half-years, years | 2005 |
| ICE gasoil | Energy & Environmental | Yes | contracts, relative, quarters, half-years, years | 2015 |
| NY Harbor ULSD | Energy & Environmental | Yes | contracts, relative, quarters, half-years, years | 2005 |
| EUA carbon | Energy & Environmental | Yes | contracts, relative | 2008 |

## Steps

**1. Pick a day**

Open **Radar**. Use the **Date** picker, or click a session in the day strip. Each session shows its count of notable and watch discoveries.

**2. Narrow the markets**

Choose a **Universe** in the header and a **Market** in the filter. Use **All**, **Notable** or **Watch** to filter discoveries by tier.

**3. Read the discoveries**

Scan the cards from the top: they are ranked by severity, then rarity. Open **Details** for the reference and the method.

**4. Investigate**

Click a card or **Investigate →** to open its move page. Or click any cell under **Where each market stands**.

![The Radar page with the date picker, day strip, market snapshot and discovery cards](https://mintcdn.com/allasso/wGLYpJ2DpEuY5p8A/images/vol-radar/radar-overview.png?fit=max\&auto=format\&n=wGLYpJ2DpEuY5p8A\&q=85\&s=1c3777fd62a961d7d9c1d3feee989929)

Radar on 13 Jul 2026, all markets: the day strip, the market snapshot and the first discoveries.

## Settings

| Control | Options | Default | Effect | URL parameter |
| - | - | - | - | - |
| **Universe** (header) | Energy & Environmental | Energy & Environmental | Limits every page to the universe's markets. Your browser remembers the choice. | `u` |
| **Date** | Any session in the data | The latest session | Shows the Radar as of that day. A non-trading date shows the last session on or before it. | `d` |
| Day strip | 12 sessions to the selected day; **‹ older**, **newer ›** (12 sessions), **Latest** | — | Jumps between days. Counts are signals that started or escalated that day. | `d` |
| **Market** | All markets, one market, or a market pair (for example BRENT–WTI) | All markets | Filters the snapshot, discoveries and grid. | `market` |
| Tier | All, Notable, Watch | All | Filters the discovery cards. | `sev` |
| Grid category | Overview, Volatility, Smile & skew, Term structure, Price & realised, Activity | Overview | Chooses the grid's columns. | `g` |
| Grid order | Default order, Most signals, Highest severity | Default order | Sorts the grid's markets by their active signals. | `gs` |

When you view a past day, a **Historical snapshot** banner says so. Every number on the page then uses data up to that day only. **Go to the latest** returns to the newest session. **Data as of** shows the newest day in the data.

## Reading the result

### Market snapshot

The snapshot lists up to six readings that sit far from their five-year median, at or beyond the 90th or 10th percentile. It picks one per family (vol level, implied vs realised, skew, wings, vol term, futures curve, realised vol). Markets of the same sector that share the reading are combined in one line. The first three are shown and the rest open under **more**. A line about one market opens its move page. Days that failed a data-quality check are left out.

### Radar discoveries

Each card is one signal that **started or escalated** that day.

* **Tier**: **NOTABLE** or **WATCH**. See the thresholds below.
* **Status**: **NEW** (first detected today) or **ESCALATED** (an open signal got more extreme; the card adds "since" its start date).
* **Header**: market · category · signal type, with the horizon for shocks (for example "Shock · 1D").
* **Line 3**: the change (shocks) or the level, how rare it is, the reference window, and the analog count with its sample label.
* **⚠**: the signal carries a quality flag. Hover it, or open **Details**.

The analog count is the number of earlier signals of the same kind whose 20-session outcome was already known on the day. Sample labels are: under 5 **insufficient evidence**, 5–9 **limited evidence**, 10–19 **moderate sample**, 20+ **larger sample**.

**Details** shows:

* **Type**: **EVENT** (a new move or a new extreme) or **REGIME** (a persistent configuration), with what keeps the signal open. Flips and seasonal signals are always REGIME. An extreme or correlation signal becomes REGIME after 20 sessions open.
* **Reference**, **Metric** and **Detector**: what it was compared with, the metric's definition and the detector's sentence.
* **Historically**: with 5 or more analogs, the median change over the next 20 sessions in ATM vol, or in the futures price for futures-price and futures-curve signals. Not a forecast.
* **Detected**, **Data quality** and **Provenance**: the dates, any quality flags, and the detector version and signal key.

Discoveries are ranked by severity, then rarity, then freshness, data quality and whether there are at least 5 analogs. Past profitability never affects the order.

### Ongoing signals

Signals still open with nothing new that day sit in a collapsed panel under the cards: "N ongoing signals · N re-flagged today · N escalated". Click **Show** to see each signal's market, since date, duration, peak severity, type and what keeps it open (hover **Continues** for the exact exit rule). **re-flagged** means the detector fired again after its cooldown, but the condition never stopped, so it is the same signal.

### Where each market stands

The grid shows every market's value for each metric of the chosen category, flagged or not.

* **Shade and p**: the value's percentile within the market's own trailing five years (1,260 sessions, up to the selected day). p99 means higher than 99% of days. The colour shows the position, not whether the value is flagged.
* **×**: the last day's move in multiples of its usual daily move over the trailing year, shown from 2×.
* **●** a notable signal is active, **○** watch, **▲** escalated today. A small number means several signals on that metric.
* **Outlined cell**: no signal, but the level is in the top or bottom 10% of its five years. Its move page shows what followed earlier entries into that band (a level band, not a Radar signal).
* **n/a**: not supported (no listed options). **short**: under a year of history. **—**: no observation that day. **⚠ QC**: the day failed a data-quality check.

Hover a column title for its definition, units and convention. Hover a cell for its five-year median, sample size and active signals. A cell with under 1,000 sessions is marked as short history. Click any cell to open its move page.

![The grid of markets by metric, shaded by five-year percentile](https://mintcdn.com/allasso/wGLYpJ2DpEuY5p8A/images/vol-radar/radar-full.png?fit=max\&auto=format\&n=wGLYpJ2DpEuY5p8A\&q=85\&s=fe4e74988a0c56fd241662ac9434c5b3)

Where each market stands on 13 Jul 2026 (Overview), below the ongoing-signals panel.

## Signal types and thresholds

The Radar reads the **30D** constant-maturity surface and futures curve, so days stay comparable through time. Values come from radar version 1.3.0.

### What is watched

| Metric | Signal types | Needs options |
| - | - | - |
| ATM implied vol | shock, extreme, seasonal | Yes |
| 25-delta risk reversal | shock, extreme, flip | Yes |
| 25-delta butterfly | shock, extreme | Yes |
| Term structure (30D − 180D) | shock, extreme, flip | Yes |
| Implied − realised vol | extreme | Yes |
| 20-day realised vol | extreme, seasonal | No |
| Futures price | shock | No |
| 10-delta risk reversal | shock, extreme, flip | Yes |
| 10-delta butterfly | shock, extreme | Yes |
| Futures curve M2 vs M1 | shock, extreme, flip | No |
| Futures curve 180D vs 30D | extreme, flip | No |

Two rules stop the same move from being shown twice:

* A 10-delta signal is shown only when the 25-delta metric did not flag the same kind and direction that day.
* ATM vol is also watched at **90D** and **180D** (shock, extreme) and **M1**, the front listed option (shock only). These are shown only when the 30D series did not flag the same kind and direction that day. M1 changes never span a contract roll.

### How signals are classified

| Type | What it detects | Notable | Watch |
| - | - | - | - |
| **Shock** | A 1-day or 5-day change against the trailing year's changes (252 sessions, the day excluded): change ÷ their standard deviation. Needs 126 past changes. | 1-day ≥ 3.5×, 5-day ≥ 3.0× | 1-day ≥ 2.5×, 5-day ≥ 2.3× |
| **Extreme** (card: Extreme level) | The level in the top or bottom tail of the trailing five years (1,260 sessions; needs 504). Flagged on entering the tail, with a 20-session cooldown. | outer 0.5% | outer 2% |
| **Seasonal** | The level is rare for this calendar month against the same month of earlier years. ATM vol and 20-day realised vol only. 20-session cooldown. | outer 1% | outer 3% |
| **Flip** | The sign changes against the trailing 20-observation median: skew from puts to calls, vol term structure inverting, the futures curve moving between contango and backwardation. 20-session cooldown. | always notable | — |
| **Correlation break** (card: Cross-market) | A pair's 60-day correlation of daily futures returns in the tail of its own five years (needs 504 observations). 20-session cooldown. | outer 0.5% | outer 2% |
| **Activity** | The previous session's volume or open interest. Up-moves only, 5-session cooldown. | — | watch only |

A 5-day shock fires once per move, not again while the same-direction change stays unusual. A 1-day shock is a new move each day.

A flip needs a minimum size, both for the new value and (at half that size) for the recent median: 1 vol pt for the 25-delta risk reversal, 2 vol pts for the 10-delta risk reversal and the vol term structure, 0.4% for the M2 vs M1 curve and 1% for the 180D vs 30D curve.

Correlation pairs: Brent crude–WTI crude, TTF gas–EUA carbon, TTF gas–Henry Hub gas, Brent crude–ICE gasoil, ICE gasoil–NY Harbor ULSD.

Activity signals use the session's numbers, published the next morning:

* **Volume**: the session's volume against its trailing 20-session median, with that ratio in its top 1% of the past year (252 sessions; needs 126).
* **Open interest**: a 5-session build-up of at least 3.0× the usual 5-session change over the trailing year.

Jumps that look like reporting changes (open interest more than doubling in 5 sessions, or volume beyond 10× its median) are kept off the Radar.

### Lifecycle: how long a signal stays open

A signal opens at the watch threshold and closes only below a lower exit threshold. A value hovering at the boundary is one signal, not a string of new ones.

| Type | Stays open while |
| - | - |
| 1-day shock | one day only |
| 5-day shock | the 5-day move stays beyond 1.5× its usual size (entry 2.3×) |
| Extreme, correlation | in the outer 5% of five years (entry 2%) |
| Seasonal | in the outer 6% for the month (entry 3%) |
| Flip | the sign holds, at least half the flip minimum |
| Activity: volume | volume stays in the top 5% (entry 1%) |
| Activity: open interest | the 5-session build-up stays beyond 2× usual (entry 3×) |

An open signal **escalates** when its tier rises from watch to notable, or when it beats its peak by a clear step: 0.5 in normal-quantile units for percentile signals, 1.0× for shocks. A signal that returns within 20 sessions of closing reopens as the same signal.

### Quality flags

Discoveries come only from days that passed the data-quality checks. A card can still carry these flags (⚠):

* **IV interpolated between listed expiries**: the 30D value sits between two listed expiries.
* **day passed the checks with a warning**.
* **under 60% of the usual reference history**: for example, fewer than 151 of the usual 252 past changes for a shock.

A 10-delta move whose wing was extrapolated or taken from settlement wings is not shown as a discovery.

## Example

Brent crude, 13 Jul 2026 (`/radar?d=2026-07-13&market=BRENT`). The tier buttons read **Notable · 3** and **Watch · 6**. The top card reads:

> NOTABLE · NEW · Brent crude · Volatility · Shock · 1D
> **ATM implied vol rose sharply**
> +19.9 pts now 60.4% | 5.3× its usual 1-day move | 1Y 1D-change reference | 49 analogs · larger sample ⚠

The 5.3× is above the 3.5× notable threshold for a 1-day shock. The ⚠ shows two flags: IV interpolated between listed expiries, and a day that passed the checks with a warning. **Details** adds: "After 49 earlier moves like this, ATM vol over the next 20 days changed by a median -2.8 pts (37% higher). Not a forecast." Brent's ATM IV 30D cell in the grid reads 60.4%, p93.

## Limitations

* The Radar is descriptive, never a forecast. Context lines describe what followed earlier moves, not what will follow this one.
* Activity signals use the previous session's volume and open interest, and are watch-level only.
* An outlined grid cell is a level band, not a signal: it is not ranked, counted or listed as a discovery.
* A cell with under a year of history shows **short**. A grid cell under 1,000 sessions, or a signal flagged "under 60% of the usual reference history", ranks against a thinner past.
* The grid's × is computed on the grid's own daily series, so it can differ slightly from a shock card's score for the same move (5.5× against 5.3× for the example).
* The analog count on a card and the event count on its move page are counted separately, so they can differ by a few (49 against 51 for the example).

## Related

* [Investigate a Radar move](https://docs.allasso.ch/apps/vol-radar/radar-move.md)
* [Tutorial: from a Radar alert to a trade idea](https://docs.allasso.ch/apps/vol-radar/tutorials/alert-to-idea.md)
* [Surface report](https://docs.allasso.ch/apps/vol-radar/surface.md)
* [How ML Data and Vol Radar compute implied volatility](https://docs.allasso.ch/kb/volatility/implied-vol-calculation.md)
* [Vol Radar API](https://docs.allasso.ch/dev/vol-radar/api.md): the same signals as JSON


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.