Release Radar

Changelog

Continuous improvements to your trading edge. Every feature, fix, and optimization — timestamped and documented.

44

Releases

271

Features

99

Days Active

Latest
May 10v2.15.0

New Pricing, Alpha Tuning & Intelligence Overhaul

New Pricing TiersIntroducing Light (CHF 49), Pro (CHF 149), and Enterprise (CHF 499) plans. All with 15-day free trial. Bitcoin payments accepted.
Alpha Tuning for 3% DailyExit system recalibrated to hold winners longer. Tighter stop-losses, stronger profit protection, and reduced time-based exits. Target: 1-3% daily returns.
Intelligence Dashboard OverhaulCollapsible sections for easy navigation. Live economic calendar with 25+ weekly events from ForexFactory. Agent status banner for non-active users.
Agent ChatTalk to your AI agent directly from the dashboard. Ask market questions, request analysis, or trigger trades — all via natural language.
Leverage UpdateSafe preset now at 5x, Balanced at 20x, Growth at 40x. Better risk-reward balance for all trader profiles.
May 8v2.14.0

Zero-Friction Onboarding & Conversion Engine

Onboarding Overhaul4-phase implementation: balance checks, structured error messages, 12 GA4 analytics events, simplified wizard for first-time users.
40+ Email TemplatesComplete email lifecycle: welcome, getting-started guide, exchange connected, agent started, first trade celebration, trial reminders, and 6-email NLP conversion sequence.
Smart PaywallInline subscription gate (no redirect), trial countdown in sidebar, upgrade CTA for free users.
Password ResetFull self-service password reset flow via email with secure token.
March 7v1.50.0

Alpha Learning Governance, Shadow Promotion, and Canary Rollback

Learning Loop LiveClosed trades now feed Alpha Intelligence in production, persisting pattern memory, calibration state, regime/session performance, and operator-facing learning snapshots after each completed trade.
Shadow Penalty RegistryRecurring negative trade motifs are now classified into deterministic tags and accumulated into a shadow penalty registry instead of being handled as one-off forensic notes.
Proposal EnginePattern losses now generate concrete shadow proposals with named policy IDs and parameter hints, allowing Alpha to suggest policy changes without mutating live behavior impulsively.
Replay + OOS LinkageShadow proposals are evaluated against stored historical patterns and can be linked to out-of-sample freeze IDs before being considered promotion-ready.
Promotion-Ready CandidatesAlpha now distinguishes between raw ideas and promotion-ready candidates, surfacing only proposals that passed shadow replay and OOS checks into the operator prompt layer.
Canary Runtime OverridesPromotion-ready proposals can enter a bounded canary path where runtime risk configuration is overridden in a controlled way instead of rolling immediately across the whole stack.
Automatic Canary RollbackIf a canary underperforms in live operation, Alpha can now revoke it automatically, closing the governance loop from learning to validation to rollback.
Institutional Learning PathThe stack now supports a full governance chain: learn → classify → propose → replay → OOS validate → canary → rollback, turning Alpha from a rule engine into a supervised adaptive trading system.
March 6v1.49.0

Regime-Routed 4 Alpha Architecture + Canonical Risk/Exit Stack

4 Alpha OrchestratorSignal Fusion V4 now routes TRENDING, CHOPPING, ACCUMULATION, and DISTRIBUTION into dedicated alpha modules with handoff persistence, cooldown, and selection telemetry.
V3 Live + V4 Shadow BridgeThe production V3 path now records V4 strategy/alpha metadata in shadow mode so operators can compare live decisions against the new regime-routed stack before full promotion.
Canonical H4 Source of TruthMacro H4 context is now propagated consistently through validation, execution telemetry, and forensic records instead of mixing macro and technical H4 labels across modules.
Signal Direction AlignmentValidation and execution now block entries whose decision side conflicts with authoritative Signal Fusion direction, preventing LONG-vs-SHORT mismatches.
Trend Reversal GuardsNew bounce/squeeze guards block late TRENDING fallback entries into oversold/bid-heavy reversals and add an early trend-reversal loss cut when local microstructure turns against the position.
Net-of-Fees Protective ExitsProtective closes now refuse to lock in negative net PnL after round-trip costs, eliminating the prior fake-profit trailing stop failure mode.
Alpha-Aware Profit ProtectionTrend-following positions now get wider proportional giveback, while mean-reversion exits protect earlier. Exit context now carries strategy_name and alpha_id for attribution.
Preset & Dashboard AlignmentRisk presets and dashboard schema now expose orchestrator controls, CHOPPING mean-reversion thresholds, and alpha-aware exit parameters so tenants inherit the same structure by default.
March 5v1.48.0

Coinbase Intelligence Integration + Multi-Venue XED

Coinbase Public Data IntegrationAdded Coinbase Exchange public market ingestion (ticker, stats, level-2 book) alongside existing Binance feed for richer external alpha context.
XED Refactor to Multi-VenueCross-Exchange Divergence now aggregates Hyperliquid vs multiple reference venues (Binance + Coinbase) instead of a single-binance comparison path.
Venue Consensus & Lead VenueIntroduced venue_consensus_score and lead_venue metadata to improve signal quality and explainability when references disagree or align strongly.
Fail-Open ReliabilityNo hard-blocks introduced: if one venue is unavailable, XED degrades gracefully using available venues; if all references are missing, it returns neutral safely.
Signal Fusion Parser UpgradeXED parser now uses multi-venue metadata (reference_count, venue_consensus_score, lead_venue) to modulate confidence and reduce noisy directional flips.
Alpha Context EnrichmentAlpha intelligence context now carries reference_venues, lead_venue, venue_consensus_score, and Coinbase premium/volume for deeper diagnostics and post-trade audits.
Backward Compatibility PreservedLegacy BINANCE_LEADING handling remains supported while adding REF_LEADING semantics, avoiding regressions in downstream components.

Missing a feature? Let us know