This paper investigates whether longer historical training periods improve the out-of-sample performance of a Hidden Markov Model (HMM) used for identifying financial market regimes. Four four-state Gaussian HMMs are compared, differing only in the starting point of their training data (1970, 1980, 1990, and 2000), while being evaluated on the same test period. The model uses five market-based features capturing volatility, drawdowns, credit conditions, and sector rotation to identify latent risk regimes.

The study highlights that validation methodology and evaluation criteria are as important as model design and data length when assessing regime-based investment models.

Hello all,

– Screener and API extended with two new fields: Transaction Costs and Robustness
– 11 new Quantpedia Premium strategies
– 7 new related research papers
– 8 new backtests
– and finally, 6 new posts on our Quantpedia blog

We analyze a multi-asset calendar anomaly around U.S. federal elections that is consistent across assets and simple to trade. While our previous study documents a robust pre-election drift, the cross-asset pricing of its D+1 resolution remains unmapped beyond SPY. Using a diversified ETF basket spanning technology, emerging markets, real estate, high-yield credit, gold, oil, and foreign exchange, we show that election day itself is neutral to slightly positive — an equal-weight D0 long earns a modest 1.03% p.a. at a Sharpe of 0.19, which we document as the control arm of the experiment but do not trade — while the following session delivers a systematic sell-off across all seven legs as the political uncertainty premium collapses. An equal-weight short on D+1 yields 1.86% per annum with a Sharpe of 0.31 and a maximum drawdown of -4.16%, achieved over 13 trading days in 26 years. Intermediate Treasuries exhibit the inverse pattern, selling off in the four days before the vote and rallying from D0 to D+2, with a Sharpe of 0.48.

Meme stocks are commonly studied through social-media activity, but this approach limits both the historical scope of research and the signals available to practitioners. In A Century of Meme Stocks and the Price of Coordination, Chad Schmerling develops an alternative: a machine-learning model trained to identify the holdings of the Roundhill MEME ETF using market and firm-level data rather than textual inputs. By adapting the feature set to historical data availability, the paper constructs a measure of “memeness” extending back to 1926. For traders and portfolio managers, its principal contribution is a framework for identifying speculative activity and examining how its return implications vary with the investment horizon.

The Quantpedia Awards are back—and this time, we’re taking the winners’ announcement to the stage!

If you have been developing a systematic trading strategy, testing an investment idea, or turning your research into a paper, this is again your invitation to join the race…

In a previous article Building and Testing Trend-Following Strategies on One-Minute SPY Data, we investigated whether retail activity indicators derived from one-minute SPY data could be used to construct profitable trend-following strategies. The results suggested that we are able to construct strategies that are often able to achieve superior risk-adjusted performance. In this article, we examine an alternative hypothesis. Instead of assuming that changes in retail activity signal the continuation of an existing trend, we investigate whether persistent declines in order flow may create conditions for a subsequent market reversal. More specifically, we analyze situations in which selected trading activity indicators decrease for several consecutive trading days and evaluate whether such sequences are followed by above-average SPY returns.