BRAHMANDA EVENT RISK TERMINAL · XAU
Analyzed0
Critical0
Bimodal0
GVZ—
CONNECTING
Event Feed loading…
Connecting to event stream…
Analogue Memory
Analyze an event — its nearest historical analogues appear here,
weighted by embedding similarity.
Precedent fan — what history did next
Retrieval & Prior Parameters w = simα · q0.2 · dtwγ
δ tail threshold10%
τ similarity0.75
DTW radius2
K min analogues25
α sim exponent0.5
γ dtw exponent0.3
retrieve at τ, relax if <K · weight w = sim^α · q^0.2 · dtw^γ · mix N(μᵢ,σᵢ)
Forecast Cone — posterior at 7d / 15d / 30d
Analogue Flow & Return Distribution event → analogue type → 15d outcome
Selected event → analogues → 15d outcome
Analyze an event to see where its analogues' outcomes flow.
Return distribution by horizon (ridgeline)
Analyze an event to see the 7d / 15d / 30d forward-return distributions.
KDE Posterior — 15d forward return
The posterior distribution renders here after Run posterior.
Precomputed density grids are used when available.
Risk Decision
—
AWAITING
Analyze an event to produce a decision. The score combines severity, analogue confidence, directional consistency and posterior tail mass.
Implied Volatility GVZ · CBOE
GVZ history loads with the terminal.
🧠 AI Analyst Synthesis
Analyze an event above, then generate a narrative synthesis of everything this terminal found.
This calls Claude itself to read the real event, analogue, outcome-distribution and forecast numbers above and write a plain-English analyst note — base event → analogues → forecast → decision — grounded strictly in those numbers, not a canned template. It only works when this page is opened as a Claude Artifact on claude.ai.
🪐 3D Embedding Map — Gold News Corpus
Loading…
Loading embeddings…
Selected event Its retrieved analogues Select an event above to light it up here.
Validation — Walk-forward Backtest test corpus · live KDE (redflag_kit)
Starting…
Run a walk-forward backtest to evaluate 90% interval coverage, directional win rate, cumulative return, Sharpe, and red-flag utility after costs.
⚡ Strategy Backtest — Event-Conditioned Signal P&L
KDE median signal → Long / Short / Flat · cost-adjusted · click any dot for detail
—
SHARPE
—
CUM. RETURN
—
MAX DRAWDOWN
—
WIN RATE
—
TRADES
—
FLAG Δ MEAN R
Red Flag Event Normal Event Loads automatically on page open; runs a backtest if none exist yet.
Signal
KDE Median
Realised r+H
Red Flag
P&L (after cost)
📈 Spot & Curve
Loading…
Spot & Futures Curve (M1 / M2)
Calendar Spread
Needs a /curve endpoint — not implemented in redflag_kit. Ask to add it if you want this live.