AIQF Daily digest — 1 strategy, 11 new papers

New public strategies

• Monthly δ-Stable Target versus Unrestricted Target in US-Stock Minimum-Variance Portfolios

  Each month, the strategy trades a capped, long-only minimum-variance portfolio targeting the cross-sectional mean estimated return. It also solves a higher-target benchmark on the same data for diagnostics, but does not trade that benchmark.

  Sharpe 0.66 · total return 49.0% · max drawdown -31.1%

Quant Paper Radar · 6 of 11

1. 1

   Cut the chit-chat: a new framework for the application of generative language models for portfolio construction

   Francesco A. Fabozzi, Ionuţ Florescu · OPENALEX · News Sentiment / Cross-Sectional Long-Short

   quant relevance 1.00

   The study constructs daily cross-sectional equity portfolios using continuous news-sentiment scores extracted from language-model label probabilities. It reports improved risk-adjusted returns relative to discrete sentiment labels and FinBERT benchmarks. The mechanism is implementable with available US equity news and prices, although testing our news feed would not reproduce the original dataset exactly.

   original paper

2. 2

   Enhancing a multi-factor equity strategy via artificial intelligence and technical indicators

   Mita, Daiya, Takahashi, Akihiko · ECONBIZ · Machine-Learning Multi-Factor Investing with Tactical Hedging

   quant relevance 1.00

   The strategy combines gradient-boosted forecasts of five-factor returns with state-space estimates of stock exposures to select Japanese equities, then uses technical-indicator rules to short TOPIX futures for downside protection. It reports historical trading performance, but the Japanese equity universe, TOPIX futures, and several required macro inputs are unavailable on the platform.

   original paper

3. 3

   Optimal Portfolio Allocation with Gromov-Wasserstein Barycenter

   Alessandro Spelta, Nicoló Pecora, Mario Maggi · OPENALEX · Sentiment-Driven Portfolio Optimization

   quant relevance 1.00

   The method dynamically allocates among bull, bear, and neutral portfolios using an entropic Gromov–Wasserstein barycenter, with regime weights driven by the CNN Fear & Greed Index. The authors report superior full-window risk-adjusted performance relative to alternative portfolios, while noting variable year-by-year results. The optimization is computationally feasible, but its specified sentiment input is not available in the platform's data catalog.

   original paper

The rest of the batch

• Optimal Size in Market Making under State-Dependent Hawkes Process

  Jonathan Chávez Casillas, Li Gan · ARXIV · relevance 1.00

  The study develops a market-making size and quote-placement policy using state-dependent Hawkes order flow, exponential demand, and a Prony-series executed-volume model.

• Financial Context-Rich Strategy for Actor-Critic based Agents: A Deep Reinforcement Learning Approach for Stock Trading

  Hrishikesh Singh, Aarti Sharma, Millie Pant · ARXIV · relevance 0.99

  The study compares actor-critic reinforcement-learning agents for daily trading of five Indian stocks, incorporating FinBERT news sentiment into states and rewards.

• Sentiment-Driven Forecasting of Fractional Trading Activity Using LSTM Networks

  Christina Erlwein‐Sayer, Sakir Kepezkaya, Natalie Packham et al. · OPENALEX · relevance 0.95

  The study forecasts changes in retail stock ownership using clustered LSTM models augmented with news sentiment, then applies the predictions to fractional-share brokers’ inventory hedging.

and 5 more · see the full radar