Quiet Signal

Role

Product designer. I designed the morning decision hierarchy, desk drill-down, warning-policy controls, and live before-and-after comparison.

Workflow

Risk limits: identify the desk needing a decision before the open, then inspect its distance to the limit and supporting evidence.

Data coverage

12 trading desks, 30 trading days, gross and net exposure, VaR, daily P&L, and 3 limit breaches.

Verification

Both views reconcile to 1 dataset. The comparison counts each first screen; the reader study in section 08 sets its pass line before any session.

Related products

Mercury makes financial position readable before detailed account inspection. Nasdaq Risk Platform monitors positions against limits. Quiet Signal combines a focused overview with a traceable desk decision.

The after view is scoped to what a risk manager has to settle before the open, in the order they settle it.

1. Which desks need a decision today?
2. How close is each desk to its limit, and which way is it moving?
3. Has a limit been crossed in the last 30 trading days, and where?

Every other question the old monitor answered, from VaR to net exposure, is still answered in the after tool’s other 5 tabs.

The sentence is a traceable decision

Read the decision flow

1. Gross exposure divided by the effective limit gives utilization
2. The warning line selects desks
3. Sort selected desks by utilization
4. Name desks and 5-day movement
5. Open the desk evidence

Fig. 01 Both tools running, with the decision log

Open prototype ←

1 dataset, 2 working tools. Switch between wipe, before, and after, and step through the 9 decisions while both tools highlight what each one changed.

Authored inventory, counted from each first screen

These are not user outcomes. I designed both screens, and each tool counts its own tiles, charts, encoding colours, and sentences as drawn and reports them to the comparison, so the figures can be checked by opening it. The aim is fewer signals competing for the same glance, not a smaller number for its own sake. Whether that helps a reader is what section 08 tests.

Fig. 02 Before: Risk Limits Monitor

Open tool ←

The answer is on screen, at the same weight as 14 tiles, 12 gauges, and 12 trend lines.

Fig. 03 After: Quiet Signal

Open tool ←

The decision comes first, in words, and the chart shows distance to the line.

Fig. 04 Before: the alert queue

View full size ←

4 alert rules, all still firing: 181 alerts in 30 days for 2 desks that need a decision.

Fig. 05 After: desk detail

View full size ←

The 30-day line, VaR, stress, and positions return 1 click down, for 1 desk at a time.

Fig. 06 After: rule replay

View full size ←

What the headline said on every day, rebuilt from the same rule, so a reviewer can check it.

1. Cut 11 of 14 KPI tiles

   VaR, P&L, headroom, and 8 more tiles left the first screen. 3 remain: desks over the warning line, breaches in 30 days, and firm-wide utilization.

   Trade-off accepted: VaR and P&L are no longer readable at a glance. They moved to the Desks tab and the desk detail.

2. Replace 12 desk colours with 1 signal hue

   Colour now carries 1 meaning: past the warning line. Every other bar is grey.

   Trade-off accepted: desks lose their colour identity, so identity moves to the label, and a reader trained on the old legend has to relearn.

3. Write the headline as a sentence

   The tool opens with the decision in words, before any chart.

   Trade-off accepted: a sentence is an interpretation and can be wrong. It is built by 1 stated rule and quotes its numbers, and the rule replay shows what it would show on each day under the selected warning threshold.

4. Sort desks by utilization, not alphabetically

   The desk that needs attention is always the top row.

   Trade-off accepted: row positions change from day to day, so muscle memory for where a desk lives is gone.

5. Draw the limit, not only the value

   The 90% warning line and the 100% limit are on every chart, so the reader sees distance to the line rather than a bare percentage.

   Trade-off accepted: the warning policy is now part of the picture. The Limits and policy tab exists because the line will move, and moving it rewrites every view.

6. Replace 12 trend lines with a 5-day change

   Each desk carries its 5-day move in points. The 30-day line lives in desk detail.

   Trade-off accepted: the long-range shape is hidden from the first screen. A slow 30-day climb and a 5-day spike can print the same number.

7. Fold gross and net into utilization of limit

   1 ratio per desk replaces 2 exposure figures and a headroom tile.

   Trade-off accepted: net exposure leaves the first screen, so a heavily hedged desk looks the same as an unhedged one.

8. Show breaches as 3 dated marks, not a 360-cell grid

   The heat grid became 1 timeline with the 3 days a limit was crossed.

   Trade-off accepted: near-misses stay off the first screen. Equity Derivatives touched 99.4% in the window and appears only when the Breaches tab’s near-miss toggle is on.

9. Move everything else behind tabs

   The alert queue, positions, grid, and reports became 5 tabs, each answering 1 question.

   Trade-off accepted: an auditor or a quant now pays 1 click for the full picture. Nothing was deleted from the data.

Subtraction on a risk surface is only safe if the reader can reverse it. The after tool is tiered in 3 levels, glance, explain, and audit, and each number that left the first screen has a named tab to live in.

The morning view decides which desk to call. When a breach still happens, the review that explains it is its own workspace: What Moved First reads a stressed hour after the fact, with a direction check before any finding is recorded.

Glance

Morning read: the headline sentence, 3 tiles, distance to limit, and the breach strip. Enough to decide which desk to call first.

Explain

Desks, desk detail, and breaches: every desk with its line, 5-day change, and 30-day shape; VaR, stress, and positions per desk; the breach register and near-misses.

Audit

Rule replay and the full desk table: what the headline would show on each day under the selected warning rule, and every figure the old monitor showed, for all 12 desks.

Policy

Limits and policy: the rule in 1 paragraph, the warning line as a control with a live preview, and the limit change log.

Accessible route

Tabs and the divider work from the keyboard, state changes are announced, every chart carries a text description, and each chart has a table behind it.

Subtraction keeps an evidence route

Read the decision flow

1. Morning read names the attention decision
2. Desk detail restores risk and position context
3. Breaches retain dated history and near misses
4. Policy controls expose the warning rule
5. Replay shows the result under the selected rule

A quieter view changes the work of everyone downstream of it, not only the person looking at it first.

Head of trading

Needs to know which desk to call before the open, and why.

Chief risk officer

Needs the firm-wide picture and confidence that nothing is hidden by the summary.

Risk operations

Need the full table to reconcile limits and prepare the breach report.

Traders

Need to see their own distance to the line without reading 11 other desks.

Audit and model risk

Need a record of what the dashboard chose to foreground, and the rule it used.

3 choices shaped the study more than any single removal.

Ranking over completeness

The after view ranks desks for 1 decision. A reader with a different question, such as which desk is most hedged, gets less from the first screen than before.

A sentence over a neutral view

Writing the headline makes the dashboard take a position. That is the point, and it is also a new failure mode, so the sentence only quotes numbers that the chart directly under it shows.

1 dataset over 2 builds

All 3 artifacts are built from 1 data file. It made the legacy tool harder to fake, and it means no removal can hide a changed number.

The tools make an argument. They do not yet have evidence from users. This is the study I would run before calling the after tool better, with the pass line set before any session.

Participants

5 to 8 people who read a limits monitor every morning: desk heads, risk managers, and risk operations analysts.

Task

Open every desk that needs a decision today. Each person uses both tools on 2 different trading days, day 30 and day 17, with tool and day order counterbalanced. Both tools can open on any study day, so no one sees the same answer twice.

Measures

Time to the last correct desk, desks missed, desks opened in error, and stated confidence on a 5-point scale.

Pass line

The after tool earns its place only if it is faster with no additional misses. If readers miss a near-miss desk more often, entry 8 in the log is reversed and near-misses return to the first screen.