v5.7.5 Executive brief available
Intelligence is the edge. 0 trades

Where Alpha is Forged.

Valtara synthesizes eight layers of institutional-grade data into a calibrated, traceable market brief for XAU/USD — consumable by human trader or autonomous agent.

Reasoning Pipeline Macro Overlay News Stream
Live
XAU/USD
SYNTHESIZING
Data Inputs
Market Data 22
Macro Signals 9
Positioning 52w
Sentiment live
News AI
Events high
Price Structure 7 TF
Synthesis Hub
8 Agents · Parallel
Master
GPT-5.4
Advisory Output
Directional Bias
5-bucket probability
Setup Brief
Waiting for signal…
Scenario Map 3 paths
Primary · Alt · Tail
Auditable
JSONL pipeline log

Powered by institutional-grade data & AI providers

FRED CFTC Yfinance ForexFactory Perplexity OpenAI MetaTrader 5
Vision & Mission

Public infrastructure for institutional gold intelligence.

Valtara's north star is to make the analytical depth that previously required a Bloomberg Terminal, a quant team, and a Reuters feed accessible, transparent, and composable — for retail traders, developers, and small institutions alike.

"Menjadikan intelligensia pasar emas yang sebelumnya hanya dapat diakses oleh institusi menjadi infrastruktur publik yang transparent, affordable, dan composable — dapat dikonsumsi secara manual oleh trader individu maupun secara otonom oleh ekosistem AI agent generasi berikutnya."
Five Operational Missions
M1
Multi-Source Synthesis
Integrasikan 8 lapisan data (macro, positioning, flow, technical, risk-off, calendar, news, live tick) ke dalam satu pipeline analitik yang koheren.
M2
Probabilistic Reasoning
Menghasilkan directional bias dalam distribusi probabilitas 5-bucket, bukan verdict biner bullish/bearish.
M3
Evidence-Based Calibration
Feedback loop berbasis post-hoc labeling seluruh keputusan historis, dengan rekalibrasi bobot Bayesian terjadwal.
M4
Full Transparency
Setiap skor, tilt, dan rekomendasi dapat di-trace ke raw data source, waktu pengambilan, dan rantai penalaran AI.
M5
Dual-Mode Interoperability
Antarmuka simetris untuk konsumsi manusia (dashboard) dan mesin (API/MCP) tanpa kompromi kualitas informasi.
Unified Data Connectivity

One domain. Zero ambiguity. Every signal fused.

Stop stitching dashboards, CSV exports, and three news terminals together. Valtara ingests primary macro, positioning, and market-microstructure streams into a single synchronized pipeline — so every signal ships with a traceable source.

Active Data Streams

8 sources
FRED
FRED Macro
9 factors · 90d history
COT
CFTC COT
52w · commercials & specs
MT5
MetaTrader 5 RDP
XAU/USD · DXY LIVE TICK
Architecture

Four layers. One coherent brief.

Each layer is independently testable and replaceable — yet flows upward into a unified, auditable analytical output.

Layer 01 Foundation

Data Fusion

Acquires, cleans, and time-synchronizes primary data streams into a single coherent pipeline. Every datapoint carries its source identifier and timestamp, enabling full downstream auditability.

FRED
9 macro factors · 90d
CFTC COT
52w positioning
MT5 RDP
XAU / DXY live
Yfinance
VIX · GSR · OI
Layer 02 Quantification

Analytical Reasoning

Transforms raw streams into structured analytical constructs — bias score, limit order zones, active setups, scenario paths, and deterministic execution signals with explicit invalidation levels.

Bias Score
0–100 · 5 component
LOZ
Tier A · B · C
Scenario Map
Primary · Alt · Tail
Signal
BUY · SELL · WAIT
Layer 03 Reasoning

Multi-Agent Synthesis

Eight specialist agents — each accountable for one analytical domain — run in parallel and feed a single Opus-tier Master that returns a calibrated probability distribution with disagreement entropy for conviction scoring.

Specialists
8 domain agents
Red Team
Adversarial check
Master
GPT-5.4 Opus-tier
Output
prob_dist · 5 bucket
Layer 04 Delivery

Interface & Infrastructure

Symmetric consumption surface — an interactive dashboard for humans, a REST API and Model Context Protocol server for autonomous agents. Event-driven, WebSocket-pushed, and fully audited in append-only JSONL.

REST API
OpenAPI spec
MCP Server
Agent skill
WebSocket
Push < 100ms
Audit Log
JSONL append
CAPABILITIES

Five pillars of defensible intelligence

What separates Valtara from indicator stacks and black-box AI — purpose-built for depth, calibration, and transparency.

01 / Memory

Episodic Memory Layer

Every bias decision and trade setup is snapshotted into a decisions store, then retrospectively labeled against price realization — building a corpus of lived experience the Master Synthesis can retrieve against.

Stored Decisions
Labeled Outcomes · 30d +2,847
02 / Infrastructure

Event-Driven Pipeline

Content-hash deduplication, cooldown protection, and WebSocket push keep the engine responsive without wasteful recomputes. Eight specialist agents run in parallel, a Master Synthesis fires only when signal actually shifts.

Dispatcher Latency
WS Push · Tick→UI
< 100 ms
03 / Calibration

Bayesian Self-Recalibration

Brier scores per component feed weekly weight updates — the blend of macro, positioning, flow, and risk-off signals tunes itself toward whichever layer is currently predictive, with bounded learning rate and audit trail.

Current Calibration
Brier Score · 7d rolling
Current 0.18 · target < 0.22 · auto-retune Sat 06:00 WIB
04 / Interop

API & MCP Interface

Symmetric endpoints for humans and agents. Install Valtara as a skill inside Hermes, OpenClaw, or Claude Computer Use — your autonomous executor gets domain intelligence without rebuilding the pipeline.

MCP Skill Example
GET /api/v1/bias/latest
{"score":62,"verdict":"BULLISH"}
05 / Traceability

Transparent Reasoning

Every score, tilt, and narrative line is traceable to its primary source — FRED series id, COT report date, SMC anchor, news timestamp. No black box, no unaccountable conviction.

Audit Chain
Ecosystem Position

Brain for humans. Brain for agents.

Valtara is not a competitor to autonomous agent frameworks — it is their domain-specific cognitive layer. The same intelligence feeds three consumption modes, cleanly separated.

Mode A

Manual Trader

Open the Valtara dashboard, read the bias and scenario brief, pull the trigger yourself on MT5 or your broker of choice. Zero dependency on external agents.

Dashboard · WebSocket live
Recommended
Mode B

Webhook Bridge

Valtara pushes calibrated setups to your MT5 Expert Advisor or custom webhook endpoint. Semi-automated execution with human oversight — the pragmatic middle ground.

Webhook · WS event stream
Mode C

Autonomous Agent

Install Valtara as a skill inside Hermes, OpenClaw, or Claude Computer Use. Your agent consumes bias, scenarios, and setups through MCP — then reasons and executes on its own.

REST API · MCP skill
Why collaboration, not competition

Generic agents bring autonomy. Valtara brings domain depth.

Hermes and OpenClaw excel at generic planning, tool use, and execution. But gold market intelligence — eight-layer data fusion, calibrated probability, and traceable reasoning — doesn't emerge spontaneously from a ReAct loop. Valtara plugs the gap.

Hermes
OpenClaw
Claude CU
Evolution Roadmap

Three phases. One coherent thesis.

Each phase deepens capability without altering the product's core positioning — analytical pipeline, then memory-augmented advisor, then composable intelligence infrastructure.

Active

Memory-Augmented Advisor

Current focus · v5.7.5 → v6.x

Adds episodic and semantic memory: decisions store, outcomes store, embeddings-based analog retrieval, pattern library clustering, and Bayesian weight recalibration from Brier scores.

  • Decisions & Outcomes DB
  • Semantic retrieval (RAG)
  • Pattern library clustering
  • Bayesian recalibration
Next

API-First Hardening

Formalizes third-party consumption

API versioning, authentication and rate limiting per tier, formal Model Context Protocol server wrapper, OpenAPI schema, webhook event push, and a developer documentation portal with sandbox credentials.

  • /api/v1 versioning
  • MCP server wrapper
  • API key + rate limit
  • Developer sandbox
Future

Ecosystem Play

Distribution & partnerships

Publish skill packages on ClawHub and Hermes Skill Marketplace, revenue-share partnerships with futures brokers, white-label programs for regional prop firms, and disciplined asset expansion driven by empirical demand.

  • · Skill marketplace listing
  • · Broker revenue share
  • · White-label program
  • · Multi-asset (demand-gated)
Access Tiers

Priced for every mode of consumption.

From curious retail explorer to agent integrator to institutional desk — one brain, four access levels.

Tier 01

Explorer

$9 /mo
≈ Rp 145.000 / bulan

For retail traders exploring institutional-grade analysis for the first time.

  • Dashboard access (delayed 15 min)
  • Daily Bias + scenario brief
  • News & calendar feed
  • · Realtime WebSocket
  • · API / MCP access
Most Popular
Tier 02

Trader

$29 /mo
≈ Rp 465.000 / bulan

For active retail traders and independent prop traders running daily sessions.

  • Everything in Explorer
  • Realtime dashboard + WS push
  • Setup Archive & hit-rate tracker
  • Scenario Map overlay
  • Memory-augmented context
Tier 03

Developer

from $99 /mo
≈ Rp 1.590.000 / bulan

For agent integrators, systematic traders, and developers building autonomous bots.

  • Everything in Trader
  • REST API · /api/v1
  • MCP server skill
  • Webhook event push
  • Sandbox credentials
Tier 04

Institutional

Custom
Contract pricing

For prop trading firms, family offices, and small hedge funds needing SLA-backed delivery.

  • Everything in Developer
  • Dedicated SLA
  • Priority support
  • Custom endpoints
  • White-label optional

Kurs konversi ≈ Rp 16.100 / USD (indikatif). Valtara adalah advisory infrastructure dan bukan auto-trader; pengguna bertanggung jawab penuh atas setiap keputusan eksekusi.