In this thesis, the risk-adjusted performance of AI-driven robo-advisors and traditional human-managed funds is compared across conservative, moderate, and aggressive investment strategies. Using Markowitz mean-variance optimization, revisited through Sharpe ratio maximization, robo-advisory portfolios were simulated based on historical market data (2017 – 2020) and tested empirically from 2021 to 2023. The results confirm that robo-advisors delivered strong returns with favorable risk-adjusted metrics in stable market conditions (2021) but suffered substantial declines and increased volatility in turbulent markets (2022). Human-managed funds, by contrast, were found to be more flexible and able to capture market recoveries more effectively, particularly for aggressive strategies. Statistical tests against the S&P 500 benchmark reaffirmed the robustness of the simulations. The study further reveals the disadvantages of robo-advisors’ reliance on historical data and fixed allocation strategies, and suggests that future work should investigate hybrid models that integrate systematic asset allocation with some element of human judgement to improve the portfolio’s ability to respond to unusual market conditions.
In questa tesi viene confrontata la performance aggiustata per il rischio dei robo-advisor basati sull’intelligenza artificiale e dei fondi gestiti tradizionalmente da esseri umani, attraverso strategie di investimento conservative, moderate e aggressive. Utilizzando l’ottimizzazione media-varianza di Markowitz, rivisitata attraverso la massimizzazione del rapporto di Sharpe, sono stati simulati portafogli robo-advisory basati su dati storici di mercato (2017 – 2020) e testati empiricamente dal 2021 al 2023. I risultati confermano che i robo-advisor hanno ottenuto rendimenti elevati con metriche favorevoli di rischio-rendimento in condizioni di mercato stabili (2021), ma hanno subito cali sostanziali e un aumento della volatilità in mercati turbolenti (2022). I fondi gestiti da esseri umani, al contrario, sono risultati più flessibili e capaci di cogliere più efficacemente le riprese del mercato, in particolare per le strategie aggressive. I test statistici rispetto al benchmark S&P 500 hanno riaffermato la robustezza delle simulazioni. Lo studio rivela inoltre gli svantaggi della dipendenza dei robo-advisor dai dati storici e dalle strategie di allocazione fissa, e suggerisce che lavori futuri dovrebbero indagare modelli ibridi che integrino l’allocazione sistematica degli asset con un elemento di giudizio umano, per migliorare la capacità del portafoglio di rispondere a condizioni di mercato inusuali.
Il futuro del portfolio management: analisi comparata del rendimento rischio-corretto tra strategie umane e AI
POLENGHI, LAURA
2023/2024
Abstract
In this thesis, the risk-adjusted performance of AI-driven robo-advisors and traditional human-managed funds is compared across conservative, moderate, and aggressive investment strategies. Using Markowitz mean-variance optimization, revisited through Sharpe ratio maximization, robo-advisory portfolios were simulated based on historical market data (2017 – 2020) and tested empirically from 2021 to 2023. The results confirm that robo-advisors delivered strong returns with favorable risk-adjusted metrics in stable market conditions (2021) but suffered substantial declines and increased volatility in turbulent markets (2022). Human-managed funds, by contrast, were found to be more flexible and able to capture market recoveries more effectively, particularly for aggressive strategies. Statistical tests against the S&P 500 benchmark reaffirmed the robustness of the simulations. The study further reveals the disadvantages of robo-advisors’ reliance on historical data and fixed allocation strategies, and suggests that future work should investigate hybrid models that integrate systematic asset allocation with some element of human judgement to improve the portfolio’s ability to respond to unusual market conditions.File | Dimensione | Formato | |
---|---|---|---|
Tesi_Polenghi_Laura_mat523425.pdf
accesso aperto
Descrizione: This thesis compares AI-driven robo-advisors and human-managed portfolios using simulated strategies, Sharpe ratios, and statistical tests to assess whether algorithmic investing can outperform traditional approaches in risk-adjusted returns.
Dimensione
967.94 kB
Formato
Adobe PDF
|
967.94 kB | Adobe PDF | Visualizza/Apri |
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: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/29131