v0.91.0 — QRT Deep-mode full due diligence

QRT (Qube Research & Technologies) rewritten as a nine-section Deep due diligence per DD_STANDARD (5 parallel research agents, ~130 searches, title-level evidence).

• AUM chain completed: $1bn (2016 spin-out) → $16bn (2023-02) → $20bn (2024) → $30bn (2025-07) → $42bn (2026-02) → $50bn (2026-07)
• 2025 performance: flagship +30% / overall +22% (dual-source)
• China path corrected: QFII (\u7acb\u65b9\u79d1\u7814) + WFOE PFM (\u5bbd\u7acb\u8d44\u672c) — not QDLP; Dao fund 98% re-baselined as cumulative since Nov-2022
• 'Olivier Jaillon = CEO' premise removed (zero co-occurrence); Companies House 09867306 pending
• 2026-07 human-stockpickers unit adds a qualifier to the 'pure quant' label
• QRT Labs (Oxford/Cambridge/Imperial) = talent open, tech closed
• Regulatory record clean: Japan impersonation = victim, DJT SEC complaint = no action, HSBC short-report glitch = technical error
• 14-item to-verify list added; ann snapshot: ann/2026-09-18_qrt_deep_dd_launch.md

v0.90.0 — High-Flyer dual-engine deep-dive (50 reports)

HIGHFLYER_SPECIAL — the dual-engine king

How Liang Wenfeng quant firm funded AGI: quant engine (700亿+ closed-source, 2025 +56.6%), AGI engine (DeepSeek $45B pre-money, Big Fund lead, >500亿 from Ma/Zeng/Ding/Liu), Firefly-supercomputer lineage (3FS open-sourced), mother-silent-descendant-open pattern (High-Flyer → DeepSeek → dsh → dsh-quant), the world-AI-richest founder (360亿美元), the 2026-08 hedge-fund drawdown, and the twin-open-source contrast with UBIQUANT_SPECIAL.

94 archives + 50 research reports.

Announcement: Discussion #104

v0.89.0 — Ubiquant China AI-transform deep-dive (49 reports)

UBIQUANT_SPECIAL — the North-Jiukun AI pivot

WorldQuant-lineage founders, ~800亿 AUM with the capsizing paradigm (2018/2021 freezes, 3y lockups, self-purchase), the AI pivot (Beiming supercomputer, "AI in 80% of strategies"), the 2026 open-source IQuest-Coder-V1 (40B code model, beats GPT-5.1, streaming training), the Zhizhi Innovation Institute + SJTU lab, the 2022 -39% leverage lesson, and the "second DeepSeek moment" thesis.

94 archives + 49 research reports.

Announcement: Discussion #103

v0.88.0 — cost gate + rebalance optimizer (59 tools, 215 unit)

quant_trading_cost + quant_rebalance_schedule

• quant_trading_cost: commission + slippage + Almgren-Chriss market impact (∝ σ × √participation)
• quant_rebalance_schedule: optimal rebalance frequency (drift vs cost trade-off)

59 tools · 215 unit tests · all green.

94 archives + 48 research reports.

Announcement: Discussion #102

v0.87.0 — validation tools (57 tools, 210 unit)

Validation tools: correlation + deflated Sharpe + stress + sensitivity

• quant_factor_correlation: pairwise matrix + redundancy detection (effective count)
• quant_deflated_sharpe: Bailey & López de Prado overfitting-adjusted Sharpe (trial-corrected min-significant)
• quant_stress_test: portfolio loss under crash/liquidity/vol scenarios
• quant_parameter_sensitivity: plateau vs needle-sharp robustness

57 tools · 210 unit tests · all green.

94 archives + 48 research reports.

Announcement: Discussion #101

v0.86.0 — research→strategy→execution bridges (53 tools, 200 unit)

quant_layered_backtest + quant_trade_quality + quant_attribution

• quant_layered_backtest: quantile-layer backtest (top/bottom/long-short equity, per-layer returns) — factor → strategy sketch
• quant_trade_quality: execution quality (fill rate, slippage bps, holding bars) — sim → live expectations
• quant_attribution: portfolio attribution (per-asset + factor-exposure regression + residual alpha)

53 tools · 200 unit tests · all green.

94 archives + 48 research reports.

Announcement: Discussion #100

v0.85.0 — IC decay + portfolio optimizer (50 tools, 193 unit)

quant_ic_decay + quant_portfolio_optimize

• quant_ic_decay: IC per horizon, half-life, best horizon, signal type (short/medium/long) → rebalance frequency
• quant_portfolio_optimize: maxSharpe / minVar / riskParity weight optimizer + portfolio stats + concentration

50 tools · 193 unit tests · all green.

94 archives + 48 research reports.

Announcement: Discussion #99

v0.84.0 — AI-infra data modules (48 tools, 186 unit)

AI-infra data modules (own-module plan 1-4)

1. Data quality layer: quant_data_pit (point-in-time / survivorship / channel reliability + health score)
2. AI-infra channel view: quant_channel_guide (agent-ready access guide + readiness check)
3. CLI: quality + channel commands
4. Industry-chain loop: quant-upstream closed-loop section

48 tools · 186 unit tests · all green.

94 archives + 48 research reports.

Announcement: Discussion #98

v0.83.0 — quant_factor_neutralize repaired + 5 baselines (179 unit)

quant_factor_neutralize fixed

The tool was registered in the 46-tool set but its factorNeutralize import was broken (silent failure). Now imported, re-exported, tested end-to-end with 5 hand-computed baselines (zscore / group / ols / residual direction / preconditions). Unit suite 174 → 179, all green.

94 archives + 48 research reports.

Announcement: Discussion #97

v0.82.0 — Tokyo yen-rates-center census (48 reports)

TYO_QUANT — the yen-rates center

~9 firms (Capula stronghold / Two Sigma / Balyasny / Millennium / Squarepoint / ExodusPoint / Dimensional / Jump / AQR), the 2025 $30M yen-rates talent war (Ron Choy → Balyasny, Harimoto → Modular, Capula/Dymon bank hires), the structural analysis (why only ~9 firms but deep positioning), and the Asia-three-city division (HK China / SG crypto / TYO yen-rates).

94 archives + 48 research reports.

Announcement: Discussion #96