Cross-Market Alpha: Testing Short-Term Trading Factors in the U.S. Market via Double-Selection LASSO

Alexander Walter; Jin Du; Maxim Ulrich

arxiv: 2601.06499 · v2 · pith:SUFU44MCnew · submitted 2026-01-10 · 💱 q-fin.ST

Cross-Market Alpha: Testing Short-Term Trading Factors in the U.S. Market via Double-Selection LASSO

Jin Du , Alexander Walter , Maxim Ulrich This is my paper

Pith reviewed 2026-05-22 12:03 UTC · model grok-4.3

classification 💱 q-fin.ST

keywords short-term trading signalsdouble-selection LASSOcross-market factorsU.S. equity marketbehavioral risk premiumshigh-dimensional regressionS&P 500price volume signals

0 comments

The pith

Short-term trading signals from China generate alpha in the U.S. S&P 500 after controlling for fundamentals.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper examines whether 191 short-term trading signals created for the Chinese A-share market can enhance returns in the U.S. S&P 500. By using double-selection LASSO to account for 151 fundamental factors in data spanning 2002 to 2022, the authors identify 17 price-volume and microstructural signals with significant explanatory power. These signals appear to reflect behavioral patterns common across markets that hold up even with monthly portfolio rebalancing. Readers might care because the approach suggests a practical way to combine fast and slow information sources to build more resilient equity portfolios and reduce exposure to crowded fundamental strategies.

Core claim

While traditional equity factor investing relies heavily on slow-moving fundamental accounting metrics, this study explores how institutional investors can leverage a high-dimensional library of 191 short-term, trading-based signals to enhance alpha generation within the U.S. S&P 500 universe. Utilizing a robust double-selection LASSO framework to control for 151 established fundamental factors, we isolate 17 distinct price-volume and microstructural signals that capture significant, non-redundant risk premiums. Our empirical evidence demonstrates that these fast trading signals capture universal behavioral dynamics that do not dilute over a monthly rebalancing horizon.

What carries the argument

Double-selection LASSO framework that isolates relevant short-term signals from a large set of fundamental factors in high dimensions.

If this is right

Integrating short-term signals with fundamental data creates a dual-horizon framework for better alpha generation.
Short-term signals help mitigate model misspecification risk in factor models.
Large-cap portfolios achieve improved diversification by including these fast behavioral factors.
The signals provide non-redundant risk premiums that persist at monthly rebalancing frequencies.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Similar cross-market testing could be applied to equity markets in Europe or emerging regions to check for universality.
Future work might examine whether these signals retain power at weekly or daily rebalancing horizons.
The method could be extended to other asset classes like bonds or commodities using analogous short-term signals.

Load-bearing premise

The double-selection LASSO framework can separate the short-term signals from the fundamental factors without introducing selection bias or missing key interactions.

What would settle it

Replicating the study on post-2022 S&P 500 data and finding that the selected signals lose their predictive ability after controlling for fundamentals would challenge the results.

read the original abstract

While traditional equity factor investing relies heavily on slow-moving fundamental accounting metrics, these models frequently suffer from factor crowding and miss real-time, sentiment-driven market dislocations. This study explores how institutional investors can leverage a high-dimensional library of 191 short-term, trading-based signals, originally developed for the retail-heavy Chinese A-share market, to enhance alpha generation within the highly institutionalized U.S. S&P 500 universe from 2002 to 2022. Utilizing a robust double-selection LASSO framework to control for 151 established fundamental factors, we isolate 17 distinct price-volume and microstructural signals that capture significant, non-redundant risk premiums. Our empirical evidence demonstrates that these fast trading signals capture universal behavioral dynamics that do not dilute over a monthly rebalancing horizon. Integrating these short-term behavioral footprints with slow fundamental data offers a powerful dual-horizon framework to mitigate model misspecification risk and enhance large-cap portfolio diversification.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper applies Chinese short-term signals to the US via double-selection LASSO and finds 17 non-redundant ones, but lacks performance numbers and robustness details.

read the letter

Here's the quick read on the Du et al. paper. The core finding is that a set of 17 short-term trading signals, originally from China, survive double-selection LASSO against 151 US fundamentals and still show up as non-redundant at monthly rebalancing in the S&P 500 from 2002 to 2022. The authors argue these capture universal behavioral dynamics that don't fade when rebalanced monthly. The paper does a decent job framing the cross-market transfer and the practical motivation around factor crowding and real-time dislocations. Using double-selection LASSO to handle the high-dimensional control is a reasonable choice for this setting, and the claim of universal behavioral dynamics is at least testable. The idea of a dual-horizon framework that mixes fast and slow signals has clear appeal for institutional portfolios. Where it falls short is the missing evidence. There are no reported alphas, Sharpe ratios, or even basic summary stats on the selected signals. The abstract talks about significant premiums but leaves the reader without numbers to evaluate the size or robustness. On the LASSO side, equity factors are often correlated, so without diagnostics like selection frequency across subsamples or checks on the irrepresentable condition, it's unclear if the 17 signals are stably identified or artifacts of multicollinearity. If the full paper has these checks, that would strengthen it a lot. This is the kind of paper that might interest quant finance groups working on multi-horizon models or high-dimensional factor selection. A practitioner could get ideas from the signal library transfer, but without the actual performance data it's more of a methods sketch than a finished result. I'd send it out for review. The question is interesting enough and the approach is grounded, even if it needs more empirical meat to stand on its own.

Referee Report

2 major / 2 minor

Summary. The manuscript applies double-selection LASSO to a library of 191 short-term price-volume and microstructural signals originally developed for the Chinese A-share market, controlling for 151 established fundamental factors, to identify 17 non-redundant signals in the U.S. S&P 500 universe (2002–2022). It claims these signals capture universal behavioral dynamics whose risk premiums persist under monthly rebalancing and can be combined with slow fundamental data for improved alpha and diversification.

Significance. If the empirical separation and persistence results hold after proper diagnostics, the work would provide concrete evidence for a dual-horizon factor framework that mitigates crowding in traditional accounting-based models by incorporating fast, sentiment-driven signals across markets.

major comments (2)

[§4] §4 (Double-Selection LASSO Framework): The central claim that the 17 isolated signals are non-redundant after controlling for the 151 fundamentals rests on consistent variable selection. However, equity factors routinely exhibit pairwise correlations >0.4, which can violate the irrepresentable condition required for LASSO selection consistency (Belloni et al.). The manuscript reports the 17 signals but supplies no post-selection diagnostics such as variance inflation factors, selection stability across folds, or post-LASSO inference to confirm the separation is not an artifact of multicollinearity.
[§5] §5 (Empirical Results): The abstract asserts 'significant, non-redundant risk premiums' and 'universal behavioral dynamics' that 'do not dilute over a monthly rebalancing horizon,' yet the provided text contains no performance numbers, t-statistics, error bars, or robustness checks (e.g., sub-period analysis or data-exclusion rules). These omissions are load-bearing for the claim that the signals enhance alpha generation in the S&P 500.

minor comments (2)

[Abstract] Abstract: Consider adding one or two quantitative highlights (e.g., average monthly premium or Sharpe improvement) to make the strength of the 17-signal result immediately visible.
[§3] Notation: Define the exact penalty parameters and cross-validation procedure for the double-selection LASSO more explicitly to allow replication.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments. We address each major point below and will revise the manuscript to incorporate additional diagnostics and explicit empirical results.

read point-by-point responses

Referee: [§4] §4 (Double-Selection LASSO Framework): The central claim that the 17 isolated signals are non-redundant after controlling for the 151 fundamentals rests on consistent variable selection. However, equity factors routinely exhibit pairwise correlations >0.4, which can violate the irrepresentable condition required for LASSO selection consistency (Belloni et al.). The manuscript reports the 17 signals but supplies no post-selection diagnostics such as variance inflation factors, selection stability across folds, or post-LASSO inference to confirm the separation is not an artifact of multicollinearity.

Authors: We agree that multicollinearity among equity factors can pose challenges for LASSO consistency. Although the double-selection procedure is intended to mitigate this by orthogonalizing controls, we will strengthen the paper by adding the requested diagnostics. The revised manuscript will report variance inflation factors for the selected signals, selection stability across cross-validation folds, and post-LASSO inference results following Belloni et al. to confirm that the 17 signals remain non-redundant after controlling for the 151 fundamentals. revision: yes
Referee: [§5] §5 (Empirical Results): The abstract asserts 'significant, non-redundant risk premiums' and 'universal behavioral dynamics' that 'do not dilute over a monthly rebalancing horizon,' yet the provided text contains no performance numbers, t-statistics, error bars, or robustness checks (e.g., sub-period analysis or data-exclusion rules). These omissions are load-bearing for the claim that the signals enhance alpha generation in the S&P 500.

Authors: We acknowledge that the current draft does not sufficiently highlight the numerical results in the main text. The revised version will explicitly report performance numbers, t-statistics, and error bars within Section 5. We will also add sub-period analysis (e.g., pre- and post-2010) and robustness checks under alternative data-exclusion rules to directly support the claims of persistent risk premiums and non-dilution under monthly rebalancing. revision: yes

Circularity Check

0 steps flagged

No significant circularity: empirical cross-market test with standard LASSO selection

full rationale

The paper applies double-selection LASSO (citing Belloni et al.) to a fixed library of 191 signals and 151 fundamentals on 2002-2022 S&P 500 data, then reports the resulting 17 signals' risk premiums at monthly horizon. This is a data-driven empirical exercise whose central claim (non-dilution of behavioral dynamics across markets) is tested against observed returns rather than defined into existence. No equation equates a fitted selection outcome to a subsequent 'prediction' by construction, no self-citation supplies an unverified uniqueness theorem, and the cross-market transfer from Chinese A-shares to US equities supplies an external benchmark. The method is standard and the results remain falsifiable outside the fitted values.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

Abstract provides limited visibility into modeling choices; main unstated premises concern cross-market transferability of signals and the ability of LASSO to isolate truly non-redundant effects.

free parameters (1)

LASSO penalty parameters
Double-selection LASSO requires tuning parameters whose values are not stated in the abstract.

axioms (1)

domain assumption The 191 short-term signals contain information orthogonal to the 151 fundamental factors that LASSO can reliably extract.
Invoked by the claim that 17 signals remain after controlling for fundamentals.

pith-pipeline@v0.9.0 · 5694 in / 1220 out tokens · 53004 ms · 2026-05-22T12:03:36.725196+00:00 · methodology

Cross-Market Alpha: Testing Short-Term Trading Factors in the U.S. Market via Double-Selection LASSO

Core claim

What carries the argument

If this is right

Where Pith is reading between the lines

Load-bearing premise

What would settle it

discussion (0)