This thesis investigates variance-optimal hedging for equity-index options under additive normal tempered stable (ATS) dynamics. In jump-driven incomplete markets, perfect replication of arbitrary contingent claims is generally not possible, and any admissible self-financing hedging strategy leaves a residual risk. Quadratic hedging provides an optimization-based alternative to sensitivity-based strategies by selecting the portfolio that minimizes the mean-squared hedging error. The theoretical contributions of this thesis can be summarized as follows. First, we verify that the martingale-corrected ATS forward process satisfies the assumptions required by the variance-optimal hedging framework, explicitly characterize its exponential-moment domain, and specialize the framework to the ATS setting: since the forward is a square-integrable martingale, the feedback term vanishes, the variance-optimal strategy coincides with the Föllmer-Schweizer integrand, and semi-explicit discrete-time hedge ratios for European calls and puts follow directly. Second, we establish intermediate-horizon quadratic projection results in both continuous and discrete time: at each intermediate horizon, the Föllmer-Schweizer integrand minimizes the expected squared difference between the conditional value of the claim and the value of the corresponding hedging portfolio, providing a rigorous theoretical foundation for the hedging procedure implemented in the numerical analysis. A further contribution is the numerical assessment of the methodology in two complementary settings: a Monte Carlo simulation study, in which the ATS model is used as the data-generating process, and an empirical analysis based on 70 at-the-money SPX calls over the period 2016-2017. In the simulation study, the variance-optimal strategy achieves the lowest terminal standard deviation and the lowest aggregate pathwise RMSE among the strategies that trade only in the forward. A complementary sensitivity analysis shows that its relative advantage over the ATS delta hedge depends materially on the ATS distributional parametrization and becomes more pronounced in the more asymmetric configurations considered. The empirical results are broadly consistent with the simulation evidence, as the variance-optimal hedge outperforms all delta-only benchmarks. The largest improvement, however, comes from introducing a second option to control gamma exposure. Among the strategies considered, the ATS delta-gamma hedge achieves the lowest terminal replication RMSE. It also outperforms the corresponding Heston delta-gamma strategy in terms of terminal replication accuracy and generates lower average transaction costs. Overall, the results show that variance-optimal hedging provides the most effective control of quadratic replication risk within the class of forward-only strategies, while enlarging the set of hedging instruments produces the largest overall reduction in hedging risk.
Questa tesi analizza strategie di hedging variance-optimal per opzioni su indici azionari nell’ambito di dinamiche additive normal tempered stable (ATS). Nei mercati incompleti caratterizzati dalla presenza di salti, la replica perfetta di un generico contingent claim non è generalmente possibile e ogni strategia di copertura autofinanziante ammissibile lascia un rischio residuo. L’hedging quadratico rappresenta un’alternativa alle strategie basate sulle sensitività, fondata su un criterio di ottimizzazione che seleziona il portafoglio in grado di minimizzare l’errore quadratico medio di copertura. I contributi teorici della tesi possono essere riassunti come segue. In primo luogo, si verifica che il processo forward ATS, corretto in modo da soddisfare la condizione di martingala, rispetti le ipotesi richieste dal framework di hedging variance-optimal. Se ne caratterizza inoltre esplicitamente il dominio dei momenti esponenziali e si specializza il framework al caso ATS. Poiché il forward è una martingala quadrato-integrabile, il termine di feedback si annulla, la strategia variance-optimal coincide con l’integrando della decomposizione di Föllmer-Schweizer e si ottengono rapporti di copertura semi-espliciti in tempo discreto per opzioni call e put europee. In secondo luogo, si stabiliscono risultati di proiezione quadratica a orizzonti intermedi sia in tempo continuo sia in tempo discreto. A ogni orizzonte intermedio, l’integrando di Föllmer-Schweizer minimizza la differenza quadratica attesa tra il valore condizionale del contratto e il valore del corrispondente portafoglio di copertura, fornendo così una base teorica rigorosa alla procedura di hedging implementata nell’analisi numerica. Un ulteriore contributo è rappresentato dalla valutazione numerica della metodologia in due contesti complementari: uno studio di simulazione Monte Carlo, nel quale il modello ATS costituisce il processo generatore dei dati, e un’analisi empirica basata su 70 opzioni call at-the-money sull’indice SPX nel periodo 2016-2017. Nello studio di simulazione, la strategia variance-optimal consegue la minore deviazione standard dell’errore terminale e il minore RMSE pathwise aggregato tra le strategie che operano esclusivamente sul forward. Un’analisi di sensitività complementare mostra che il suo vantaggio relativo rispetto alla strategia delta ATS dipende in misura rilevante dalla parametrizzazione distributiva del modello ATS e diventa più pronunciato nelle configurazioni parametriche associate a una maggiore asimmetria. I risultati empirici sono sostanzialmente coerenti con l’evidenza ottenuta nella simulazione, poiché la strategia variance-optimal supera tutti i benchmark basati esclusivamente sulla delta. Il miglioramento più rilevante deriva tuttavia dall’introduzione di una seconda opzione per controllare l’esposizione alla gamma. Tra le strategie considerate, la strategia delta-gamma ATS consegue il più basso RMSE terminale di replica. Essa supera inoltre la corrispondente strategia delta-gamma di Heston in termini di accuratezza della replica terminale e genera costi medi di transazione inferiori. Nel complesso, i risultati mostrano che l’hedging variance-optimal fornisce il controllo più efficace del rischio quadratico di replica all’interno della classe delle strategie che operano esclusivamente sul forward, mentre l’ampliamento dell’insieme degli strumenti di copertura produce la maggiore riduzione complessiva del rischio di hedging.
Variance-optimal hedging under Additive normal Tempered Stable dynamics: theory and market evidence
ROSATI, DALILA
2025/2026
Abstract
This thesis investigates variance-optimal hedging for equity-index options under additive normal tempered stable (ATS) dynamics. In jump-driven incomplete markets, perfect replication of arbitrary contingent claims is generally not possible, and any admissible self-financing hedging strategy leaves a residual risk. Quadratic hedging provides an optimization-based alternative to sensitivity-based strategies by selecting the portfolio that minimizes the mean-squared hedging error. The theoretical contributions of this thesis can be summarized as follows. First, we verify that the martingale-corrected ATS forward process satisfies the assumptions required by the variance-optimal hedging framework, explicitly characterize its exponential-moment domain, and specialize the framework to the ATS setting: since the forward is a square-integrable martingale, the feedback term vanishes, the variance-optimal strategy coincides with the Föllmer-Schweizer integrand, and semi-explicit discrete-time hedge ratios for European calls and puts follow directly. Second, we establish intermediate-horizon quadratic projection results in both continuous and discrete time: at each intermediate horizon, the Föllmer-Schweizer integrand minimizes the expected squared difference between the conditional value of the claim and the value of the corresponding hedging portfolio, providing a rigorous theoretical foundation for the hedging procedure implemented in the numerical analysis. A further contribution is the numerical assessment of the methodology in two complementary settings: a Monte Carlo simulation study, in which the ATS model is used as the data-generating process, and an empirical analysis based on 70 at-the-money SPX calls over the period 2016-2017. In the simulation study, the variance-optimal strategy achieves the lowest terminal standard deviation and the lowest aggregate pathwise RMSE among the strategies that trade only in the forward. A complementary sensitivity analysis shows that its relative advantage over the ATS delta hedge depends materially on the ATS distributional parametrization and becomes more pronounced in the more asymmetric configurations considered. The empirical results are broadly consistent with the simulation evidence, as the variance-optimal hedge outperforms all delta-only benchmarks. The largest improvement, however, comes from introducing a second option to control gamma exposure. Among the strategies considered, the ATS delta-gamma hedge achieves the lowest terminal replication RMSE. It also outperforms the corresponding Heston delta-gamma strategy in terms of terminal replication accuracy and generates lower average transaction costs. Overall, the results show that variance-optimal hedging provides the most effective control of quadratic replication risk within the class of forward-only strategies, while enlarging the set of hedging instruments produces the largest overall reduction in hedging risk.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_07_Rosati_executive summary.pdf
solo utenti autorizzati a partire dal 30/06/2027
Descrizione: Executive Summary
Dimensione
1.09 MB
Formato
Adobe PDF
|
1.09 MB | Adobe PDF | Visualizza/Apri |
|
2026_07_Rosati_Tesi.pdf
solo utenti autorizzati a partire dal 30/06/2027
Descrizione: Tesi
Dimensione
2.52 MB
Formato
Adobe PDF
|
2.52 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/260800