Diagnostic Toolkit
Recent Picks
| Time | Market | Source | Layer | Action | Current | Entry | Exit | Δ% | Invested | P&L | Potential | Status | Resolves |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Loading... | |||||||||||||
Live Crypto Prices (Coinbase)
Live Stock Prices (Yahoo Finance)
Recent News Matches
About the News trigger
Headlines that matched an entry in the unified catalog (Polymarket and Kalshi) above the similarity threshold and fired the Screener → Strategist → Auditor pipeline. The Strategist can then emit correlated picks on any source — PM, Kalshi, crypto, or stocks — so a single headline can cascade across every market we monitor. This is the pre-movement trigger surface — markets reprice after these headlines land.
Precision by Similarity Bucket
Each row is all news matches at that rounded similarity bucket, joined through the resulting chains to their picks. Win rate is correct / resolved — use this to decide whether to raise news_min_similarity before re-enabling the feed.
| Similarity | Matches | Chains | Picks | Resolved | Correct | Win Rate |
|---|---|---|---|---|---|---|
| Loading precision data… | ||||||
| When | Headline | Feed | Matched Market | Similarity | Picks |
|---|
Recent Whale Trades
Large trades by known whale addresses on Polymarket. When a whale places a trade above the configured notional threshold on a catalog market, the trade fires the Screener → Strategist → Auditor cascade as a pre-movement trigger — the whale acting IS the signal.
| When | Trader | Market | Side | Size | Notional | Picks |
|---|
Settings
Live config.yaml tunables. Saves write to disk, re-validate via load_config, and roll back on failure. API keys and DASHBOARD_TOKEN live in .env and are not editable here.
Audit Sweep — Screener Backstop
Periodic GPT-4o-mini sweep over the full catalog looking for L2/L3 connections the hot-path screener missed. Each finding names a recent shift (chain_id) and a specific catalog slug the sweep flagged as a missed link. Findings are alert-only — nothing trades automatically from here. Dismissing a row hides it from future views.
Recent Findings
| Found | Chain | Matched Market | Rationale | Actions |
|---|---|---|---|---|
| Loading… | ||||
Dismiss removes a finding from the default view. Set a dashboard token in .env to enable the action buttons.
Source History
Bulk historical price/probability rows pulled by the source_history extractors, plus persisted correlation findings from the miner. Use this to inspect what the engine has captured for replay/backtesting beyond the live price_ticks log.
Stored Rows by Source
—One row per source. Empty means nothing has been pulled yet — see CLAUDE.md Dashboard Admin Routes for the curl invocations to populate it.
| Source | Rows | Tickers | First | Last |
|---|---|---|---|---|
| Loading… | ||||
Series Drill-down
Click a row in the rollup above to autofill, then Load. Tail returns the most recent 500 rows for that pair.
| Timestamp (UTC) | Price | Liquidity |
|---|---|---|
| Pick a series above and click Load. | ||
Top Correlations
—Pearson correlation of bucket-over-bucket returns between every stored series pair. Run python -m ripple.source_history_correlate --no-dry-run (or the auth-gated /api/source-history/correlate endpoint) to refresh. Sorted by |r| descending — sign matters.
| Pair | r | n | Window | Span | Computed |
|---|---|---|---|---|---|
| Loading… | |||||
Roadmap
Top: incoming suggestions (gitignored) — edit, promote, or delete. Bottom: the tracked ROADMAP.md for context. Promote moves a suggestion into the latest version section and auto-commits + pushes.
Incoming Suggestions
—ROADMAP.md (read-only)
—Bankroll
Real trading-account accounting. Every pick debits at entry, credits at resolution. API spend and trading fees debit the same pool. The 50% cap limits how much can be "in play" across unresolved picks at any time.
| When | Type | Amount | Fee | Balance | Details |
|---|
Historical Performance
Summary cards
All Picks
| Time | Market | Source | Layer | Action | Current | Entry | Exit | Δ% | Invested | P&L | Potential | Status | Resolves |
|---|
Exit Reason Distribution
P&L by Exit Reason
Stop-Loss Post-Mortem
▶ Shift Detection & Thresholds
Shift Detection
Threshold Settings
python -m ripple.tuner to optimize
▶ Advanced / Tuning
Gate ROI
| Verdict | Chains | Picks | Resolved | Win Rate | Net P&L | Invested |
|---|
PASS = Gemini said YES + the Strategist ran. BLOCK = Gemini said NO (no downstream picks to score). BYPASSED = news/social pre-movement triggers (embedding similarity IS the gate).
Filtered Shifts (24h)
| Source | Drop Reason | Count |
|---|
Chain Cost Ledger
| Chain ID | Gemini | Screener | Strategist | Total | Calls |
|---|
Escalation Breakdown (30d)
| Model | Calls | Cost |
|---|
| Reason | Calls | Total | Mean |
|---|
Market
Trigger Shift
▶
Ripple Analysis
L1
L2
L3
Pre-movement chain: this shift came from a news headline or social post — the trigger market hasn't repriced yet, so L1 is directly tradable alongside L2 and L3. You'll see L1 picks in the Watched Markets and Pick Results tables below.
▶
Watched Markets
| Slug | Layer | Action | Yes Price | Trend | Liquidity | Expires | Status |
|---|
▶
Pick Results
| Time | Market | Source | Layer | Action | Current | Entry | Exit | Δ% | Invested | P&L | Potential | Status | Resolves |
|---|
Open Picks
Unresolved picks across all sources with current unrealized P&L. Resolved picks live in the History tab.
| Time | Market | Source | Layer | Action | Current | Entry | Exit | Δ% | Invested | P&L | Potential | Status | Resolves |
|---|
Reasoning
Per-chain Strategist trace: sentry summary, screener take, cascade map, layer-by-layer thesis, and the router telemetry (fast-path vs. escalation, risk score, liquidity at call, tokens, cost) for every Claude call stamped to this chain. Click a chain to expand.
Recent Chains
—History
| Time | Market | Shift | Risk | L1 | L2 | L3 | Audit |
|---|
Recent Social Matches
About the Social trigger
Bluesky posts from the configured allowlist that matched an entry in the unified catalog (Polymarket and Kalshi) above the similarity threshold and fired the Screener → Strategist → Auditor pipeline. The Strategist can then emit correlated picks on any source — PM, Kalshi, crypto, or stocks — so a single post can cascade across every market we monitor. Like news, social is a pre-movement trigger surface — markets reprice after these posts land.