Harden alpha grammar and auth-aware research

Diversify alpha research lanes

Prevent research generator duplicate exhaustion

Improve alpha research selection and submission verification

Introduce authoritative platform submission verification and boot-time reconciliation

Implement adaptive seed pool with family-grouped empirical weights and retired dead-ends

Refine submitted seed mutation strategy, add fallback toggles, and introduce generation queue backoff

fix: remove contradictory standard-field-only constraint in user prompt to allow alternative data generation

Treat 403 INQUIRY_INCOMPLETE responses as pending status during biometric polling

Add rate limit (429) backoff handling and error logging to biometric authentication initiation

Configure biometric poll URL to hit the original inquiry location instead of authentication, and handle 202 status code

Inject rich diagnostic logging into biometric polling loop to track inquiry locations and cookies

Simplify biometric recovery polling by periodically re-POSTing basic auth to get 201 when Persona scan completes

Follow biometric redirects back to WorldQuant and capture session cookies, and add /force-cookie command to bot

Capture and attach challenge cookies to biometric verification get requests

Separate Basic Authentication header from client defaults in biometric auth initiation

Instantiate biometric client with pre-configured headers to preserve agent headers during polling

Correctly serialize allowed_updates array for Telegram API, use direct biometric URL, and simplify alert formatting

Implement interactive biometric verification polling and inline callback commands in Telegram bot

feat: implement QuantForge v2.5 payout optimizations: AST cloning filter, high-liquidity directives, and safe simulation timeout slot recycling

feat: implement v2.4 campaign theme compliance patch with dynamic tags (PowerPoolSelected, RegularAlpha) and pure power pool theme guards (HTVR ratio >= 0.75, pv1 dataset ban)

feat: tag PowerPoolSelected in submission PATCH and enforce date-based strict theme rules (HTVR test and dataset bans) locally in evaluator

feat: implement strict date-based Power Pool theme targeting to plug TOP3000 universe leak

fix: nest description inside regular property sub-object and handle PATCH failures properly

feat: QuantForge v2.3 - dynamic Sharpe evaluator, proactive auth, relative rate limits, and raised polls timeout

Allow advanced dataset prefixes, raise concurrent limits to 10, raise daily submission limits to 20, and automatically patch structured descriptions to alphas prior to submission

Implement lookback parameter mutation on historically submitted seed alphas

Resolve empty metrics envelopes by fetching alpha details endpoint and recover incomplete alphas automatically in main loop

Introduce ExpressionFactory grammar generator, store raw result payload in DB, and handle incomplete/timeout simulation statuses

Implement resilient retry logic, submission locks, rate-limit headers, and dynamic concurrency tracking

Implement wake_event synchronization to instantly resume recovery loop upon HTTP cookie updates

feat: implement /quick-auth and mobile browser bookmarklet for one-tap biometric session cookie syncing

fix: switch TelegramNotifier to HTML formatting to resolve markdown entity parsing crashes

feat: Add Telegram API Proxy support to bypass Hugging Face firewall blocks

fix: rewrite Telegram notifier client to use httpx.AsyncClient to prevent aiohttp uvloop connection timeouts

fix: disable SSL verification for Telegram notifier client to prevent container-level certification crashes

feat: Add keep-alive loop, Telegram auth recovery system, and fix main.py NameError crash

fix: rescue 71 stuck SIMULATING alphas on boot; lower Sharpe pre-filter 1.5->1.2 to unblock submissions

fix: send sanitized expression (1e-6->0.000001) to BRAIN payload; ban sci notation in prompt

fix: add auth lock + exponential backoff to stop BIOMETRICS_THROTTLED cascade on session expiry

feat: add per-call variation seeds to eliminate LLM expression recycling; expand strategy list 10->25

Execute Phase 16: Wrap queue database session commits with IntegrityError handlers to prevent crash on duplicate expression generation

fix: remove ts_skewness/ts_kurt, add ts_median/ts_zscore/ts_av_diff, block double group_neutralize, HTVR targeting

Execute Phase 13.5: Support High Turnover (HTVR) Campaign with evaluator thresholds, lookback constraints, and boosted USA regional weight

Execute Phase 13 updates: Purge ts_skewness and ts_kurtosis, define explicit safe list of operators, and prevent constant/no-trades expressions

Execute Phase 13: Synchronize all remaining local prompt adjustments, regional weight configurations, and operator diversity instructions

Phase 15: Expose silent NVIDIA errors and deep purge few-shot dataset examples

Phase 14: Align prompt schema for Nvidia NIM, OpenAI, Kimi, Zhipu, Deepseek

Execute Phase 13: Rescue reasoning outputs with fallback parsing, and switch primary NVIDIA model to standard text model meta/llama-3.3-70b-instruct

Restore primary NVIDIA NIM model to llama-3.3-nemotron-super-49b-v1.5 after 70B 404 validation