This thesis examines one of the most persistent puzzles in empirical finance: why does the naive equal-weight portfolio strategy (1/N) so often outperform theoretically optimal portfolio optimization methods in out-of-sample tests? The study replicates and extends the influential analysis of DeMiguel, Garlappi and Uppal (2009) using updated data from 1990 to 2025 across different asset classes and datasets. The empirical results show that the relative performance of optimization-based strategies versus naive diversification is driven by the ratio M/N, where M is the estimation window lenght and N the number of assets.When the number of assets is limited, constrained strategies tend to outperform the 1/N benchmark. As the number of assets increases, unconstrained mean-variance optimization performs poorly, while naive strategies remain more stable and robust. The naive diversification puzzle persists in contemporary data: simplicity and robustness remain key determinants of performance in environments dominated by estimation risk.
Questa tesi analizza una delle più persistenti questioni della finanza empirica: perché la naive equal-weight portfolio strategy (1/N) riesce così spesso a superare i theoretically optimal portfolio optimization methods nei test out-of-sample? Lo studio replica ed estende l’influente analisi di DeMiguel, Garlappi e Uppal (2009), utilizzando dati aggiornati dal 1990 al 2025 su diverse asset classes e datasets. I risultati empirici mostrano che la performance relativa delle optimization-based strategies rispetto a quella della naive diversification è determinata dal rapporto M/N, dove M rappresenta la lunghezza della finestra di stima e N il numero di asset. Quando il numero di asset è limitato, le constrained strategies tendono a superare il benchmark 1/N. All’aumentare del numero di asset, invece, la unconstrained mean-variance optimization mostra risultati deboli, mentre le naive strategies rimangono più stabili e robuste. Il naive diversification puzzle persiste nei dati contemporanei: semplicità e robustezza restano determinanti chiave della performance in contesti dominati dal rischio di stima.
Revisiting the 1/N Portfolio: Estimation Risk and Robust Portfolio Strategies
DOVERI, FEDERICA
2024/2025
Abstract
This thesis examines one of the most persistent puzzles in empirical finance: why does the naive equal-weight portfolio strategy (1/N) so often outperform theoretically optimal portfolio optimization methods in out-of-sample tests? The study replicates and extends the influential analysis of DeMiguel, Garlappi and Uppal (2009) using updated data from 1990 to 2025 across different asset classes and datasets. The empirical results show that the relative performance of optimization-based strategies versus naive diversification is driven by the ratio M/N, where M is the estimation window lenght and N the number of assets.When the number of assets is limited, constrained strategies tend to outperform the 1/N benchmark. As the number of assets increases, unconstrained mean-variance optimization performs poorly, while naive strategies remain more stable and robust. The naive diversification puzzle persists in contemporary data: simplicity and robustness remain key determinants of performance in environments dominated by estimation risk.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi _Doveri_Finance_26.pdf
embargo fino al 30/10/2026
Dimensione
2.2 MB
Formato
Adobe PDF
|
2.2 MB | Adobe PDF | Richiedi una copia |
È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: [email protected].
https://hdl.handle.net/20.500.14239/34894