---
title: When Does Size Matter?
url: https://www.ml-quant.com/papers/ssrn/7713618/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-11-20
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: SSRN 7713618
source_url: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7713618
featured: 2026-11-20
citations: unknown
topic: Asset Pricing & Factors
---


# When Does Size Matter?

Neural networks predict stock-level returns and aggregate them into a size-premium forecast that times the Fama-French size factor, lifting Sharpe to 0.61 in the U.S.

- Source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7713618
- Identifier: SSRN 7713618
- Released: 2026-11-16
- First featured: Quant Letter No. 140 (2026-11-20): https://www.ml-quant.com/issues/2026-11-20/
- Citations (Semantic Scholar): not tracked
- Published in: not yet
- Topic: Asset Pricing & Factors
- Authors: Vitor Azevedo, Jonathan Brogaard, Sebastian Weibels

## Related

- [Deep IV Factor Models](https://www.ml-quant.com/papers/ssrn/5283770/): The Deep Implied Volatility Factor Model, combining neural networks and linear regression, is proposed for estimating the daily Implied Volatility surface of individual stock options, improving performance around earnings announcements.
- [Multi-Factor Model](https://www.ml-quant.com/papers/ssrn/4840857/): The piece introduces a new neural network-based asset pricing model that includes time-varying volatility dynamics and offers improved predictive accuracy and risk-adjusted returns.
- [Fama-French Five-Factor Model vs. Machine Learning](https://www.ml-quant.com/papers/repec/gam-jmathe-v-11-y-2023-i-13-p-2988-d-1186815/): The paper develops a seven-factor model for the A-share market, compares five machine learning algorithms, and discovers that SVM and random forests enhance fitting power, while the performance of lasso, ridge, and neural networks varies.
- [Idiosyncratic Market Effects in Global Time Series Momentum](https://www.ml-quant.com/papers/arxiv/2611.15646/): Fitting nonlinear time series momentum models to individual markets rather than globally produces modestly better out-of-sample returns despite smaller sample sizes.
- [Interpretable Machine Learning for Asset Pricing](https://www.ml-quant.com/papers/ssrn/4473746/): The paper utilizes deep neural networks to more accurately estimate equity risk premia over time, enhancing the interpretability of machine learning in economics.
- [Optimal Investment and Consumption in a Stochastic Factor Model](https://www.ml-quant.com/papers/arxiv/2509.09452/): The article discusses optimal investment and consumption in an incomplete stochastic factor model, offering a comprehensive characterization of the problem's well-posedness and an efficient numerical algorithm for computing the value function.
