Daily Recap

The end-of-day arc, what mattered, the cross-asset verdict, and tomorrow’s setup.

The Daily Recap is auramarkets’ single most useful page if you missed the trading session. It’s the AI-written summary layered on top of the day’s quant score arc.

Generated automatically at 4:30 PM ET on weekdays, then archived. You can scroll back through every recap with the date selector at the top.

How to open it

• From Macro: click Daily Recap in the left rail.
• Or jump to it from the Market Pulse sidebar — the recap badge appears once it’s published.

The header

• Day type label — Trending Up/Down, Reversal (Recovery/Failed), Range-Bound, Breakout, or Mixed. Color-coded. → What each label means
• Score and sentiment word — e.g., “+18 / cautious” or “−45 / risk-off”
• Title — a one-line characterization of the day, written by the AI (e.g., “Risk-On Breadth Masks Bond Market’s Recession Warning”)

The arc bar

Right under the header — the gradient bar showing how the quant score moved through the session. Color encodes risk-on/off at each moment.

→ How to read the arc bar

The state strip

The hero disambiguates three constructs that live on different clocks:

• Today’s session — the intraday shape classifier (Range-Bound, Trending, Reversal…): how the day traded.
• Backdrop (multi-day) — the regime engine’s standing state (Risk On, Risk Off, Rotational…): the environment the day traded in. A range-bound session inside a risk-on regime is not a contradiction — they answer different questions.
• Day score — the quant score’s net of the day’s signals, shown on a band gauge (bearish / cautious / mixed / cautious / bullish at −30/−10/+10/+30) so a reading like +15 · cautious is self-explanatory: mildly positive, not yet bullish. A relationship sentence underneath ties all three together.

The same strip appears in the Market Pulse sidebar and on Today’s regime card, so the three concepts read identically everywhere.

Evidence is clickable

Tickers in the recap prose (SPY, QQQ, a single name with its move) open the standard stock pop-out in place, and entity references link to their source surface: a theme like Gold Miners opens its chart, an AI value-chain segment like Memory/Storage lands on AI Watch’s chain table with that segment expanded, and metric phrases (DDR5 spot, hyperscaler G-spread, RPO) jump to the exact panel that tracks them. Only verified references link — ambiguous words stay plain text. Hovering a linked ticker briefly previews a mini price chart.

The sections

The recap is forensic: everything it asserts is measured, not remembered. Each one follows the same structure, in this order:

Section	What’s in it
The Day’s Arc	The session as a story: how it opened, where it inflected, how it closed. The “what happened” narrative.
Regime Path (engine)	The day as the regime engine saw it — intraday state pills showing every flip (e.g., RANGE-BOUND → ROTATIONAL → RISK OFF) and when it happened. Computed, not narrated.
Signal vs Noise	Separates signal from noise by measuring each event’s actual tape reaction — the events that got headlines but moved nothing get called out as noise.
Cross-Asset Check	What credit, vol, gold, FX, and rates were saying — specifically which engine components dissented from the equity read.
Structural Shifts	Concentration, factor rotations, sector dispersion, breadth health.
Equity Internals	Breadth, factor excess, sector leadership, vol term structure detail.
The Score Story	A walk-through of how the quant score itself moved — peaks, troughs, what they coincided with.
Tomorrow’s Setup	Bias, catalysts, regime-shift triggers, historical edge from similar days — and key levels cited from computed numbers (52-week high/low, 50-day, 200-day, prior day’s range), never invented.
Bottom Line	One-paragraph summary if you’re skimming.

How it’s generated

• A deterministic engine (quant score, the regime engine’s intraday path, breadth, factor excess, divergences, per-event tape reactions) produces the structured features.
• Claude writes the prose by reading those features plus the day’s quant timeline, the Pulse events, historical context for similar days, and a computed AI-complex evidence block from AI Watch — group moves, value-chain segment outliers, the CDS gap, hyperscaler bond spreads, hardware spot — so on days the AI trade drives the tape, the recap can attribute the move (“Memory −3% with DDR5 spot rolling over”) instead of hand-waving “semis weak”.
• The AI never invents numbers — it only narrates the structured data. Sections stay in their lane by rule: a rates or credit driver appears in an equities read only when tied to named equity moves, and a theme already covered by another section is referenced, never restated.

Browsing past recaps

The date selector at the top lets you scroll backward through every archived recap. Useful for:

• Reviewing your trades — open the recap from the day you took a position
• Pattern recognition — read several “Reversal (Failed)” days in a row to internalize what they look like
• Learning the vocabulary — every recap uses the same terms; reading 5–10 will make the Concepts glossary click