Hello all,

We hope you’re enjoying the middle of summer. Here’s a quick recap of the latest improvements and additions we’ve prepared for Quantpedia during the past month

– API users can now directly download the full research papers written by Quantpedia
– 10 new Quantpedia Premium strategies
– 2 new related research papers
– 7 new backtests
– and finally, 5 new posts on our Quantpedia blog

Trend following was one of the most persistent anomalies in finance for nearly two centuries, yet its performance deteriorated sharply after the 2008 financial crisis. An analysis of approximately 100 liquid futures contracts from 1995 to 2025 shows that this decline is highly selective. The decisive factor is not asset class, liquidity, market electronification, or strategy crowding, but volatility-normalized tick size. After 2008, trend-following profits collapsed almost entirely on small-tick contracts across all signal horizons, while remaining largely intact on large-tick contracts. This finding suggests that modern trend-following portfolios are fundamentally split into two distinct regimes governed by market microstructure rather than traditional asset classifications.

A profitable backtest is rarely the end of a research process. In professional quantitative research, the more important question often comes after the first positive result: is the strategy genuinely new, or is it simply another version of an already known factor, timing rule, or anomaly?

The goal is not to replace human research judgment. The goal is to create a structured robustness checker that helps researchers understand whether a new strategy is truly differentiated, redundant with known effects, or simply a variation of an existing Quantpedia strategy profile.

Recent interesting research from Cakici and Zaremba, highlights an often-overlooked aspect of machine learning for equity return prediction: the choice of prediction target. Rather than focusing on increasingly sophisticated model architectures or feature engineering, the authors show that how returns are represented during training has a much larger impact on predictive performance. In particular, models trained to predict stock ranks instead of raw return levels generate substantially stronger portfolio performance—roughly doubling both returns and Sharpe ratios in large-cap universes.

Financial markets are often viewed primarily through lens of equity index movements, as they attract most of the attention. However, commodities represent an important component of the global economy, and shocks in commodity markets can have a significant impact on broader financial assets.

These events highlight the importance of analyzing portfolio behavior not only during equity bull and bear markets, but also during commodity-driven shocks.

Large language models are already capable of summarizing financial research, but are they ready to conduct it? In their latest paper, researchers from Google, Boston College, and Columbia introduce a framework where a large language model doesn’t just fetch data—it acts as an autonomous AI research agent capable of navigating the “hypothesis discovery loop.” By placing an LLM within a human-designed laboratory—complete with a symbolic language of 66 accounting primitives and a standardized backtesting pipeline—the authors tested whether AI can move beyond black-box predictions to generate economically legible and statistically robust signals. This isn’t just about throwing a transformer at a price series; it is a systematic attempt to automate the “propose–test–reflect” cycle that defines empirical finance.