Skip to content

Releases: jessn-dev/Sentient

v1.1.0-beta1

24 Jan 09:03
f2dac69

Choose a tag to compare

v1.1.0-beta1 Pre-release
Pre-release

This pre-release marks a major architectural shift for Sentient. I have moved from simple static targets to a robust, maturity-based tracking system. This version also introduces the "Unified Market Engine," significantly improving data reliability and eliminating empty dashboards.

🚀 New Features

  1. Unified Market Movers Engine
    Smart Scraper: Merged Finviz and Yahoo Finance into a single pipeline. The system now attempts to scrape real-time data from Finviz and seamlessly falls back to Yahoo if rate-limited or blocked.

Intelligent Caching: Implemented class-level caching (TTL=300s) to prevent API spam on dashboard refreshes.

  1. Flexible Prediction Timelines
    User Control: You are no longer locked into a 7-day model. When tracking a stock, you can now select a 7, 15, or 30-day forecast horizon via the new modal.

Smart Scheduling: The system automatically calculates target dates, skipping weekends to ensure the forecast lands on a valid market day.

  1. Maturity-Based Accuracy Tracking
    True Performance Metrics: The /accuracy dashboard now distinguishes between "Realized" and "Unrealized" gains.

Pending State: Predictions remain in a "⏳ Pending" state (0% score) until they reach maturity (today >= end_date).

Precision: The "Avg Precision" score now excludes pending items to prevent skewing your success rate.

  1. Deep Market Intelligence
    Institutional Data: New endpoint /market/data/{symbol} provides:

Put/Call Ratios: Gauge market sentiment.

Implied Volatility (IV): Assess expected price movement.

Top Holders: See institutional ownership stakes.

Auth Hardening: Added a pre-flight check (/auth/check) to prevent duplicate user signups.

🐛 Bug Fixes & Stability
Dependabot: Fixed critical build failure by migrating from upstash-qstash to the correct qstash PyPI package.

Rate Limits: Implemented a Retry Loop (3 attempts) with exponential backoff for the Alpaca client to prevent immediate crashes on network blips.

Yahoo Fallback: Fixed the "No Alpaca Client" warning that caused immediate rate-limiting by caching historical data requests for 1 hour.

⚠️ Known Issues (Work in Progress)
Forecast Graph: The "Analyze" graph currently shows historical data only. The "Predicted Path" (future curve) is not yet visualized.

Holiday Handling: The date calculator currently skips weekends but does not yet account for NYSE market holidays (e.g., Christmas).

Technical Indicators: The RSI and SMA values in the API response are currently placeholders awaiting the pandas-ta integration.

RateLimit yfinance is blocking requests