Explore market and public databases, compare forecast ranges, and ask Scout about the saved evidence.
Scout sits below each completed forecast. Explore the median, historical highs and lows, or source and model evidence. Add chart notes and reopen saved conversations in Requests.
Define a trading rule, size positions and review costs, equity and drawdown. Export SPY share research or FTMO lot signals with timestamp receipts. Historical simulations remain separate from a live track record.
Portfolio evidence guideExplore BigQuery public datasets, World Bank indicators and U.S. Treasury series. Download observations or forecast a selected series.
Search an economic indicator to select its series.
Download observed history from Alpaca, Dukascopy, or supported market sources.
Upload predictions and inspect their ranges and anomalies.
Compare models and forecast quantiles with the observed history.
Filter markets by price relative to a forecast quantile.
Review Kalshi and Polymarket game forecasts before kickoff.
Connect your AWS account to run SageMaker jobs.
Reference setup for naive, moving-average, and classical models before introducing advanced architectures.
Open noteDesign for direct multi-horizon outputs with explainability hooks suitable for operations and revenue teams.
Open noteWalk-forward validation standards with leakage checks, confidence calibration, and error decomposition.
Open noteHow SageMaker Canvas evaluates time-series forecasts and how Quantura preserves P10/P50/P90 as a distribution.
Open noteHow to derive acceleration, surprise, and regime features from forecast trajectories.
Open notePipeline standards for training automation, validation checks, deployment control, and rollback.
Open noteData quality, feature integrity, drift/skew dashboards, and policy-based retraining triggers.
Open noteTracking conventions that connect datasets, code versions, model artifacts, and production outcomes.
Open noteStart with naive and seasonal-naive baselines, then move into moving averages and ARIMA/ETS. This gives a stable benchmark before adding modern architectures that can hide failure modes.
Train models to emit full forward paths, not single points. Pair sequence encoders with interpretable attention and feature importance so commercial stakeholders can audit forecast drivers.
Use rolling windows with strict time ordering, then report point and interval metrics side by side. Validate that feature snapshots and joins do not leak future information into training or evaluation windows.
SageMaker Canvas reports advanced time-series metrics including Average wQL, WAPE, RMSE, MAPE, and MASE. These measurements describe different error characteristics; no single score replaces inspection of the forecast horizon and quantile calibration.
Quantura treats P10 as a lower scenario, P50 as the median/base scenario, and P90 as an upper scenario. It never collapses the distribution into a guaranteed point target. Uploaded rows are date-aware and retain the model's supplied boundaries.
Convert forecast curves into actionable features: first and second derivative (delta, acceleration), surprise versus realized values, and regime filters that suppress low-confidence transitions.
Treat model updates as software releases with deterministic builds, test gates, staged promotion, and automated rollback triggers for unstable live performance.
Monitor feature distributions, input quality, prediction drift, and business KPIs together. Trigger retraining only when quality gates or drift thresholds are breached.
Every model should be reproducible from data snapshot to deployment hash. Link experiment IDs to model registry entries and keep inference lineage for governance and postmortem debugging.
Quantura Forecasts begins with a formal unresolved question, a defined evidence cutoff, and an objective resolution rule. A possible future headline is a human-readable scenario—not a claim that the event occurred. Published probabilities are preserved as an append-only trajectory and resolved probabilities are evaluated against observed event frequencies.
Reviewers validate the proposition, deadline, primary resolution source, political-safety rules, and point-in-time evidence.
New information creates a probability revision. It never replaces the initial snapshot or earlier revisions.
Category-specific structured resolvers use authoritative evidence. Conflicts become disputed records instead of forced outcomes.
Versioned endpoints use hashed, scoped, revocable API keys, cursor pagination, per-key rate limits, and redacted projections that exclude private strategy fields.
API resources are served under /api/v1. Available scopes separate current records, probability history, resolutions, bulk trajectories, and administration. Authentication credentials belong in request headers and are never accepted in URLs.
Keep data acquisition, AWS model building, prediction review, and notifications as explicit stages with auditable handoffs.
AWS compute charges remain separate from Quantura platform charges and are billed by AWS to the connected account when resources run there.