Cluster Density

Who it’s for

Users deciding where to look, before deciding what to look at

Assumes

You have read the discovery vocabulary

A matrix of the market, region against asset class, showing where setup activity is concentrated. The screen you use before the Radar, not after it.

---

What is this?​

A grid. Rows and columns are market slices (region, asset class, and where relevant venue group and universe); each cell reports how much setup activity that slice currently carries, and how trustworthy the reading is.

Why does it exist?​

Because "where should I be looking" is a real question and every other discovery screen assumes you have already answered it. A cell with high cluster density and good coverage is a slice worth scanning in detail. A cell with high density and poor coverage is a slice where the number is being produced by too few instruments to mean much. And the screen tells you which you are looking at.

How does it work?​

Each cell carries metrics that are server-owned, the client renders what it is given and does not recompute availability or state, so what you see is what the service determined.

Metric	What it tells you
instrument_count	Instruments in the slice
with_bars	How many had usable price history
skipped_count	How many were skipped
coverage	The proportion actually evaluated
n_pub	Published opportunities
n_ready	Opportunities at READY
n_triggered	Opportunities at TRIGGERED
median_setup_quality	Median Setup Quality across the slice
median_liquidity_score	Median liquidity component
share_good	Proportion of setups at GOOD or better
cluster_density	The headline concentration figure
pulse_volatility	Volatility pulse for the slice
pulse_breadth	Breadth pulse for the slice
universe_note / venue_note	Caveats about what the slice actually contains

Each cell also has a state and, when it is not available, a reason. A cell that cannot be computed says so rather than rendering a zero, which is the same principle the Universe Scanner applies to absent metrics, and for the same reason.

Every cell is stamped with the snapshot ID it came from.

How do you use it?​

1. Read the matrix for concentration.
2. Check coverage before you believe density. A slice with 20% coverage has a median computed from a fifth of its instruments.
3. Read n_ready against n_triggered, a slice full of triggered setups is later in its move than one full of ready ones.
4. Read the pulses for context: high volatility with low breadth is a different market from high volatility with high breadth.
5. Drill into the slice with the Radar or the Trading Map.

---

Example​

What to take from it: coverage is high, so the median is trustworthy. But breadth is narrow while volatility is elevated, a slice where a handful of instruments are moving a lot and the rest are not. That is a context in which a breakout reading deserves more scepticism, not less.

---

Limitations​

Density is not opportunity. A concentrated slice is a slice where a detector's conditions are commonly met. Whether that is tradeable is a separate question the screen does not address.

Coverage caps meaning. Every aggregate is computed over with_bars, not instrument_count. Read them together.

Medians hide shape. A median Setup Quality of 58 is consistent with a tight cluster around 58 and with a bimodal split. share_good partially discloses this; it does not fully.

Slices are the platform's, not yours. The matrix axes are defined by the service. A slice you want that is not an axis cannot be requested.

Snapshot, not live.

---

Next​

• Analog Events, the historical-frequency companion
• Trading Map, drill into a slice
• Opportunity Radar

---