AQAI QuantAI research lab for systematic strategies

Daily research update

AIQF Daily digest — 1 strategy, 9 new papers

Sent 23 September 2026 · 9 new papers · 1 strategy

New public strategies

Quant Paper Radar · 3 of 9

  1. 1

    Financial Language Models as Applied Artificial Intelligence Systems for News-Based Trading under Market Frictions

    quant relevance 0.99

    MFAST is an end-to-end framework that converts timestamped company news into U.S. equity trading signals while accounting for market frictions. The paper reports that decoder-only language models, particularly LLaMA-3, outperform encoder models and dictionary-based sentiment baselines in classification, calibration, return prediction, and net long-short portfolio performance after transaction costs and capacity constraints.

    original paper

  2. 2

    Hierarchical Multi-Task Learning with Liquidity-Aware Signals for Stock Forecasting

    quant relevance 0.99

    LiMT targets tradable daily equity forecasts and liquidity-aware long-short portfolio construction. On Chinese CSI300 and CSI500 forecasting tests and CSI300 backtests, it raises annualized return from 3.99% to 10.01% and Sharpe from 1.22 to 1.86 relative to equal weighting. Its allocation mechanism uses return, volume, volatility, and liquidity constraints and can be tested on U.S. equities, although the reported Chinese-universe results cannot be replicated there.

    original paper

  3. 3

    Equilibrium Strategies for the \({n}\)-Agent Mean-Variance Investment Problem over a Random Horizon

    quant relevance 0.90

    The paper derives time-consistent Nash-equilibrium mean-variance investment rules for competing agents with wealth-dependent risk aversion and random horizons. It provides analytical results and numerical sensitivity examples under a stylized stochastic market model rather than measured trading performance on real market data.

    original paper

6 more papers written up in this period. Open the Quant Paper Radar →

This issue was emailed to our subscribers. Subscribe to get the next one — daily or weekly, your choice.

Backtest and model results are research artifacts, not live trading results and not a guarantee of future performance. Informational and educational purposes only. Not individualised investment advice.