MrktPrice — OpportunityTracker™ (OPT) Market Terminal

⚠ Alerts — whole book, every active threshold breach

Severity

Type

Ticker

Cross-sectional thresholds (RVOL, OU sigma, RSI/MFI, OBV divergence, state, drawdown vs your alert inputs) computed across the whole book from the same series as the per-name Alerts card. Click a row to load that ticker. Research use only; not investment advice.

Real move vs Noise

—

Displacement— —

Volume confirm—

Decision & Release · P0/P10 gate

—

Objective (R1)—

Lifecycle (R20)—

Conformal 90% band (R7)— —

Risk-budget size (R9)—

Release manifest (R3/R13)—

Pipeline integrity · Phase 1–3 gate

—

Returns stationary— —

ADF t / KPSS η—

Coverage / max gap—

Outlier rate (MAD>5σ)—

EMA→next-return corr—

Resid white (Ljung-Box·EMA)— —

Vol clustering (ARCH-LM·EMA)— —

Resid normal (Jarque-Bera·EMA)— —

Dist. drift (PSI train→now)— —

±2σ coverage (Kupiec POF)— —

Predictive density (PIT·KS, EWMA)— —

Data hash (parity)—

Volatility state · 86–90

Realized σ (ann)—

GARCH σ (ann)—

RV / model— —

Persistence κ̂— —

Half-life—

Downside skew— —

Participation

RVOL (20d)— —

Price vs VWAP(20)— —

Anchored VWAP—

Awesome Osc— —

Variance regimes · ICSS

Regimes (window)—

Current σ (ann)—

Break crit—

Risk geometry · 91–96

95% CVaR (1d)— —

Max drawdown—

Beta vs mktneeds mkt

HV term structure

EMA regression / recursive proj

★ best 1-step tracker (recursive prediction) · R best trend fit (regression). Dotted line = damped (φ=0.9) momentum projection of the selected EMA, anchored at current price.

OU mean-reversion

Volume spikes — RVOL ≥ 2×

This name — trailing

Projection / crosshair

hover the chart

pricecentral pathconformal 90% VWAP ancVWAP20±2σ band 80%95%P68-K ◆13D/G ▲insider ●

PROJECTION CONE — ZOOM —

Anti-deviation cone correction

Panes: price + VWAP + fan + regimes · volume coloured by RVOL · Awesome Oscillator (SMA5−SMA34) · drawdown. Daily-bar VWAP (typical price = close); true intraday VWAP needs ticks. Fan = parametric σ-bands (50/80/95, regime-aware) shaded + the conformal 90% distribution-free band (blue dashed, with its realised coverage). +5/10/20d ticks mark the horizon with calendar dates. Thin overlaid lines = the 5 triangulated forecast methods (OU reversion · EMA-momentum · Trend · empirical-median · random-walk) with an empirical 10–90% envelope; the on-chart box gives terminal odds — P(up), P(≥P6 high), P(≤P6 low), median. The cone is a calibrated reference range, not a point forecast — the central path is the OU/blend mean, never advice.

Opportunity Scanner — whole book, ranked by composite signal

State mean-reverting only min |composite| click a column header to sort

Composite = 0.35·reversion + 0.30·momentum + 0.25·volume-confirmed signal + 0.10·volume-flow. CONFLICT = momentum and mean-reversion disagree. Char = which strategy (Momentum/Reversion) had the higher backtested Sharpe for that name. Click a row to load · a header to sort.

Hawkes self-excitation & peer contagion — drift/stability gate: branching ratio n=α/β < 1 per name, system spectral radius ρ(N) < 1 across peers

Strongest peer excitation — a jump in from raises to's intensity

Branching matrix N (row i excited by column j); brighter = stronger.

Events are threshold-exceedance jump days (|return| > 2σ). A real exponential-kernel Hawkes λ(t)=μ+Σα·e^(−β(t−tᵢ)) is fit by maximum likelihood (Nelder-Mead); n=α/β is expected offspring per jump, n≥1 ⇒ self-exciting drift. The peer matrix is pairwise-assembled (shared β, self terms from each name's univariate fit, cross α_ij with a ≥8-event sufficiency gate); ρ(N) is its spectral radius — ρ≥1 ⇒ a jump cascade across peers is self-sustaining (system drift). The estimator was verified against a simulated Hawkes (the in-browser JS matches the Python reference to 4 decimals; on synthetic data it recovers the branching ratio to ~2% and the spectral radius to <1%). Honest limit: ~10–25 jumps per name on 252 daily rows is sparse, so treat magnitudes as indicative — the n<1 / ρ<1 stability verdicts are the actionable output, and a full joint / intraday Hawkes remains in the Python engine.

All 21 governance controls (R1–R21) from the model-risk report, computed offline and deterministically for every name — the full machine-readable gate set that must pass before any live-data pull. ● = pass, ○ = fail; hover a cell for the metric. G_sys = ∏ of all gates. Click a row for its integrated release report (JSON). Hard safety gates (R01 objective, R03 release, R13 parity, R15 kill-switch, R17 license, R18 security self-scan, R19 red-team) block release; soft gates downgrade to research-only. R18 security and R19 red-team are genuinely self-certified at runtime by the page scanning its own script for external refs, secrets, remote-loads, dynamic-eval, and by running leakage / ordering / unit-flip probes.