Accessibility · methodology & honest limits

Accessibility is
information equity.

In institutional finance, accessibility isn't a compliance checkbox — it's risk mitigation. If a person can't interpret critical data under pressure, that's a failure of the system, not the user. Here's how I design for it in high-density financial UX — and, just as importantly, an honest account of what I've measured versus what still needs real user testing.

WCAG 2.1 AA
Target
conformance
12
Token contrast
pairs measured
960+
Data points on one
audited surface
Honest
Measured vs
needs-testing
The hard part

Density is where accessibility gets serious.

A marketing page is easy to make accessible. A trading grid with hundreds of live values, where a colour is the difference between buy and sell, is where the real work is. Three problems I design against.

High-density information

Visualising 960+ concurrent data points (as in the TradeX risk matrix) breaks naïve accessibility patterns. The design has to hold structure at scale.

  • Reflow to 400% zoom without loss of function or horizontal scroll WCAG 1.4.10
  • Semantic grid navigation — ARIA 1.2 Grid patterns so a screen reader moves row-by-row, column-by-column, keeping context inside a massive dataset

Keyboard-first workflows

Professional traders operate entirely by keyboard for speed. That serves mobility-impaired users and terminal power-users with the same design — a rare case where accessibility and performance point the same way.

  • Focus management — logical tab order, visible focus, no focus traps in modal workflows WCAG 2.4.7
  • Skip-to-action — hidden skip links and per-module bypass blocks for rapid access to execution panels WCAG 2.4.1

Colour-independent data

Markets run on red/green. But roughly 1 in 12 men has a colour-vision deficiency, and monochromacy exists. Direction can never live in colour alone WCAG 1.4.1.

  • Redundant encoding — direction conveyed via ▲/▼ symbols + explicit +/− signs, not colour alone
  • 3:1 Luminance contrast ratio — ensuring market data direction remains legible under total achromatopsia
Interactive spec · WCAG 1.4.1

Colour-Independent Data Demonstration

Icon (▲/▼) + explicit sign + 3:1 luminance contrast so market direction survives even total colour blindness.

Double-encoded (Compliant ✓)
EUR/USD▲ +0.42%
GBP/JPY▼ −0.18%
BTC/USD▲ +2.15%
Colour only (Non-Compliant ✕)
EUR/USD0.42%
GBP/JPY0.18%
BTC/USD2.15%
The part most portfolios skip

What I've measured — and what I haven't.

A "WCAG AA" badge is easy to type and hard to earn. So instead of a badge, here's the register: what has actually been measured, and what still needs the kind of testing that a claim of "accessible" really requires. This mirrors my accessibility field note in full.

Measured / documented

What the evidence supports

  • 12 token contrast pairs measured against WCAG AA thresholds, spec-formula (not eyeballed)
  • ARIA patterns documented — grid, dialog, live-region and disclosure patterns specified per component
  • Keyboard paths walked — manual keyboard audit of focus order, visible focus, and bypass blocks
  • Automated checks run — axe / WAVE passes, run by hand during design
Not yet — and I won't pretend otherwise

What still needs doing

  • No commissioned third-party audit. Self-assessment isn't the same as an independent one.
  • No real screen-reader user testing. I smoke-test with NVDA myself — but testing with people who depend on screen readers is the evidence that actually counts, and I haven't run it.
  • No automated a11y in CI. Checks run at design time, not on every commit, so regressions can slip.
  • Coverage is partial. The measured work anchors a subset of components, not the whole system.

I'd rather show you exactly where the evidence ends than sell you a badge.

Closing the gap

How I'd take it from documented to proven.

The honest gaps above aren't a shrug — they're a plan. Here's the replication path, with the order I'd run it and a rough sense of cost.

Step 01 · low cost

Wire a11y into CI

Add automated axe checks to the build so contrast and ARIA regressions fail the pipeline, not the audit. Days, not weeks — mostly configuration.

Step 02 · moderate

Screen-reader user testing

Recruit a small panel of people who use NVDA / JAWS / VoiceOver daily and run task-based sessions on the dense surfaces. This is the evidence the register is missing.

Step 03 · investment

Commissioned audit

An independent WCAG 2.1 AA audit against the priority components, scoped to the surfaces that carry the most regulatory and financial weight first.

Found a barrier? Tell me — I'd rather know.

Accessibility is a continuous practice, not a finished state. If you hit an information barrier on anything shown here, or have a suggestion, I want to hear it.

Last review: 2026-07-03 · Ed Chen, Senior Product Designer · target WCAG 2.1 AA