NamVan56/QuantForge

Allow REFERENCE status in seed selection and few-shot example queries to guide dynamic mutations without simulating or submitting them

Replace specific private successful alpha expressions with generic public accounting seeds in the golden template pattern

Optimize mutation prompts and proven seeds library with the golden fundamental change momentum templates from upstream successes

Plug funnel leaks: evaluate metrics regardless of check status, implement unified Phase F background refinement for failed alphas, and route polling check failures to refinement

Optimize alpha diversity and orthogonalization: rotate through 12 proven seed templates, improve LLM prompts, and enhance local fallback to preserve Sharpe

Optimize miner: stagger generation, implement key rotation, two-tier evaluation, and filter refinements

Resolve merge conflicts, optimize alpha quality with Gemini 2.5 Flash, and merge submission logic

Fix: Optimize miner throughput, increase submission limit to 20/day, add queue replenishment loop to 150 target, and handle submit 403 demotion

Fix: Remove unsupported vector_neut operator in refiner prompts and local fallbacks

Fix: Restore strict criteria (Sharpe >= 1.25, Fitness >= 1.0) and enhance refiner to optimize for these thresholds

Fix: Add sign-flipping for negative Sharpe, WAL checkpointing in DB backup, and retroactive promotion of historical alphas

Fix: Lower submission criteria, add is_submittable flag, fix delay 0 fallback, improve alpha templates

fix: remove vector_neut and other vector operators from prompts and fallbacks

fix: route sk-ant API keys directly to native Anthropic endpoint and add anthropic package to requirements.txt

fix: stagger simulation submissions and add exponential retry backoff to prevent CONCURRENT_SIMULATION_LIMIT_EXCEEDED (429) errors

fix: ignore SELF_CORRELATION pending state to prevent submission deadlocks and loop stalls

fix: resolve 7 critical pipeline issues causing zero alpha submissions

Return full alpha fields in /api/queue for pending details modal

Cap decay settings parameter at 512 to prevent API rejection

Add View Details button to Pending Submission block on dashboard

Migrate to event-driven immediate submission, increase daily cap to 20, and rename SP500 to TOPSP500

Replace SP500 with TOPSP500, set daily limit to 15, and enable immediate submissions

Fix requests.Response boolean check bug and increase simulation submission max_attempts to 10

Lower cache fields threshold to 4000 to prevent startup re-scrapes

fetch and cache all 7000+ data fields across all universes, backup/restore data_fields_cache.json to Hugging Face dataset

fetch and cache fields for all allowed universes (TOP3000, TOP2000, TOP1000, TOP500, TOP200, SP500)

allow multiple universes in settings and fallback mutations (TOP3000, TOP2000, TOP1000, TOP500, TOP200, SP500)

add settings diversity to LLM prompts and fallback mutations

clean up logs by changing is_home_wifi to DEBUG level and removing debug endpoint

add debug_whitelist endpoint for diagnosing IP matching issues

fix: merge env vars and settings for ip whitelist, add ip logging

docs: add title and description to README to trigger rebuild

fix: add network exception handling to submission retries, cap concurrent simulations to 2

fix: sanitize sci notation for ALL queued alphas on boot + eliminate WARNING log spam from polling

fix: sanitize scientific notation format (1e-9) to standard decimal floats across generation and DB normalization phases, and update LLM prompts to prevent scientific notation output

fix: add retry loop on simulation submission level for 429 concurrent limit error

feat: include simulation settings in /api/archive for details modal display

Enforce a minimum rate limit retry duration of 10 seconds in wq_client.py

Fix database normalization unique constraints boot crash, timer drift using client-side seconds, rate-limit retry logging noise, and add Details modal for Tested Archive

Fix variable hallucination: restrict local fallback fields to verified ones, add prompt warnings, normalize variables on startup, and prevent infinite resets on unknown variable errors

Auto-start miner, enable SQLite WAL, implement local smooth countdown, and sort submittable archive by latest first

Implement evolutionary genetic loop, vector operator guardrails, automated orthogonalization, and smart retry backoffs

Fix timezone display to local IST and add extensive sort and filter controls to Tested & Submittable archives

Add Submitted At column to Submittable Archive table

Persist and restore miner_settings.json from HF dataset backup repository

Optimize dashboard UI/UX for responsiveness, dark mode premium aesthetics, and PnL pre-loading

Remove redundant is_home_wifi checks from settings and miner control routes, relying entirely on auth state

Fix template syntax error, allow login from any IP, and auto-authenticate home WiFi devices

Implement guest dashboard mode with IP whitelisting for write/credentials access, and add day-wise categorization for submittable archive

Implement expression/variable name validation, dynamic few-shot learning prompt injection, and fix pending checks sync bug