Behavioral Probability Weighting and Portfolio Optimization under Semi-Heavy Tails
Ayush Jha Abootaleb Shirvani Ali Jaffri, Svetlozar T. Rachev Frank J. Fabozzi · 2025
Read Behavioral Probability Weighting and Portfolio Optimization under Semi-Heavy Tails by Ayush Jha Abootaleb Shirvani Ali Jaffri, Svetlozar T. Rachev Frank J. Fabozzi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-04208.txt.
Introduction
Ayush Jha and co-authors blend behavioral finance into portfolio design under fat-tailed market returns. Investors do not weigh odds in a straight way; fear and greed bend their sense of risk. The team pulls those bends straight out of top portfolios rather than asking people about them.
Working with Dow Jones stocks, they build frontiers that balance mean return against severe loss risk. From the best portfolios under two return models, one bell shaped and one with semi-heavy tails, they recover the implied probability weighting curves. Fatter tails stretch those curves further from straight odds. Shifts in safe interest rates also bend the curves. The results urge risk managers to model skewed returns and bent beliefs together when setting capital under extreme risk.
Catalog record
| CREATOR | Ayush Jha Abootaleb Shirvani Ali Jaffri, Svetlozar T. Rachev Frank J. Fabozzi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 20 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-04208
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Behavioral Probability Weighting and Portfolio Optimization under Semi-Heavy Tails Ayush Jha Abootaleb Shirvani Ali Jaffri, Svetlozar T. Rachev Frank J. Fabozzi Abstract This paper develops a unified framework that integrates behavioral distortions into rational portfolio optimization by extracting implied probability weighting functions (PWFs) from optimal portfolios modeled under Gaussian and Normal-Inverse-Gaussian (NIG) return distributions. Using DJIA constituents, we construct mean-CVaR 99 frontiers, alongwith Sharpe- and CVaR-maximizing portfolios, and estimate PWFs that capture nonlinear beliefs consistent with fear and greed. We show that increasing tail fatness amplifies these distortions and that shifts in the term structure of risk-free rates alter their curvature. The results highlight the importance of jointly modeling return asymmetry and belief distortions in portfolio risk management and capital allocation under extreme-risk environments. Keywords: Portfolio Optimization, Portfolio Forecasts, Probability Weighting Function, Tangent Portfolios, Distributional Laws 1 Introduction Optimal asset allocation hinges on the interplay between rational risk–return trade‐offs and the behavioral impulses of greed and fear that shape investors’ subjective perceptions of uncertainty. In classical mean–variance theory, pioneered by Markowitz 1952 and extended by Sharpe 1964 , investors maximize the Sharpe ratio under objective probability weighting, yielding a unique tangency portfolio along the Capital Market Line. Yet real‐world returns exhibit skewness and leptokurtosis ( Kraus and Litzenberger 1976 ; Harvey and Siddique 2000 ) , and empirical allocations deviate from CAPM predictions, giving rise to persistent anomalies such as the equity premium puzzle ( Mehra and Prescott 1985 ) and the low‐volatility effect ( Ang and Chen 2002 ) . Prospect theory ( Kahneman and Tversky 1979 ) and its cumulative extension ( Tversky Kahneman 1992 ) offer a behavioral alternative, positing that investors apply inverse‐S probability weighting functions (PWFs) that overweight rare tail events and underweight moderate ones. These distortions, confirmed experimentally by Prelec 1998 , underpin behavioral phenomena such as the disposition effect ( Shefrin and Statman 1985 ; Odean 1998 ) and lottery‐like demand for high‐skew assets ( Bali, Cakici, and Whitelaw 2011 ) , inducing systematic biases in portfolio allocations. Recent theoretical contributions incorporate such PWFs into optimization models. Barberis and Huang 2008 and Bordalo, Gennaioli, and Shleifer 2013 show that rank‐dependent utility generates “pyramidal” portfolios that combine safety and speculation, while Polkovnichenko 2005 attributes severe under‐diversification to probability weighting alone. However, these models often impose restrictive parametric forms, typically Gaussian or exponential, that fail to capture the heavy tails and jump risk documented in empirical return distributions ( Kelly and Jiang 2014 ) . Separately, tail‐risk measures like Conditional Value‐at‐Risk (CVaR) have been introduced into the optimization framework ( Rockafellar and Uryasev 2000 ) , and studies such as Ang and Chen 2002 show that CVaR-based portfolios tilt toward safer assets compared to their Sharpe‐optimal counterparts. Yet, even these approaches typically assume objective probability weighting, leaving a gap in frameworks that integrate behavioral distortions, through PWFs, into both variance‐ and tail‐risk–based portfolio construction. In this paper, we bridge rational and behavioral paradigms by extracting implied probability-weighting functions (PWFs) from optimal portfolios estimated under heavy‐tailed return distributions. Using the thirty constituents of the Dow Jones Industrial Average, we construct mean–variance and CVaR 99 efficient frontiers and extending the analysis to the Sharpe-maximizing and CVaR-maximizing tangent portfolios. Under two comparative distributions, Gaussian and Normal Inverse Gaussian (NIG), we derive the PWFs implied by each portfolio’s tangency point used as the subjective view relative to the objective or prior view set as the DJIA index benchmark. This approach quantifies the behavioral distortion embedded in optimal allocations under varying assumptions about return asymmetry and kurtosis. Our results reveal that heavier-tailed laws produce PWFs with sharper inverse‐S curvature, capturing heightened fear (overweighting of extreme losses) and greed (overweighting of extreme gains). These effects are magnified under CVaR 99 optimization relative to mean–variance, indicating that tail-risk minimization intensifies behavioral distortions. Additionally, we show that shifts in the term structure of risk-free rates alter the shape of the implied PWF, suggesting that funding conditions interact with behavioral risk perceptions. Together, our framework offers a tractable method for identifying and managing behavioral risk in portfolio choice, grounded in observable return distributions. 2 Methodology 2.1 Data Our empirical analysis proceeds in two parts: (i) classical portfolio optimization under the CVaR criteria, and (ii) forward-looking evaluation of Sharpe-maximizing and CVaR 99 -maximizing tangency portfolios relative to a DJIA benchmark. This section outlines the data construction and transformation steps supporting each component, along with the distributional diagnostics used as robustness checks. 2.1.1 Portfolio Analysis We construct an equity portfolio comprising the thirty constituents of the Dow Jones Industrial Average (DJIA), a price-weighted index of U.S. large-cap equities. The dataset includes unadjusted closing prices obtained from Bloomberg as of June 14, 2025, covering the period from November 19, 2020 to May 3, 2025, yielding five years of daily data. These unadjusted series exclude dividend effects and reflect capital appreciation alone, ensuring return calculations isolate price dynamics. Daily arithmetic returns are computed from raw prices, consistent with discrete-time portfolio analysis and convex optimization requirements. As noted in Shiryaev 1999 , arithmetic returns preserve aggregation properties and allow tractable solutions for CVaR frameworks. This transformation also mitigates nonstationarity in raw price levels and stabilizes risk estimates across assets. We employ two proxies for the risk-free rate: the three-month and ten-year U.S. Treasury yields, reported as annualized, continuously compounded rates. These are converted to effective daily rates via r_{f,3m}=0.042/252 and r_{f,10y}=0.046/252 , respectively, and used in the construction of capital market lines and tangency portfolios across optimization regimes. 2.2 Tangent Portfolio Analysis Tangent portfolios lie at the core of modern asset allocation, identifying optimal trade-offs between risk and return. Under Gaussian assumptions and the availability of a risk-free asset, the tangency portfolio maximizes the Sharpe ratio along the Capital Market Line (CML). In practice, asset returns display skewness, kurtosis, and discontinuities, limiting the effectiveness of variance-based metrics. Moreover, tail-sensitive investors often favor risk measures such as Conditional Value-at-Risk (CVaR). As such, we estimate tangency portfolios under both variance and CVaR objectives and assess their empirical performance in- and out-of-sample. Input: Return series r_{t} (window W ) Step 1: Estimate \widehat{\mu} & \widehat{\Sigma} (Mean and Covariance Matrix) Step 2a: Sharpe‐optimal \displaystyle\max_{w\geq 0,\;\mathbf{1}^{\top}w=1}\frac{\widehat{\mu}^{\top}w}{\sqrt{w^{\top}\widehat{\Sigma}w}} Step 2b: CVaR‐optimal \displaystyle\max_{w\geq 0,\;\mathbf{1}^{\top}w=1}\frac{\widehat{\mu}^{\top}w}{\mathrm{CVaR}_{0.99}(w)} Rolling 4-year window Generate 100-day out-of-sample returns vs. DJIA benchmark 2.3 Distributional Fit and Out-of-Sample Performance To evaluate the statistical adequacy of competing return distributions, we fit Normal and Normal-Inverse-Gaussian (NIG) laws to the 100-day out-of-sample returns of Sharpe-maximizing, CVaR 99 -maximizing, and DJIA benchmark portfolios. Table 1 reports the AIC, BIC, and Kolmogorov–Smirnov (KS) statistics across specifications. In all cases, the NIG distribution dominates on both likelihood-based and nonparametric metrics, reflecting its superior ability to accommodate skewness and excess kurtosis observed in realized returns. The difference is particularly notable for the DJIA and CVaR portfolios, where Gaussian fits significantly understate tail risk. In the Sharpe-optimized portfolio, the estimated NIG parameters imply a pronounced tail index ( \alpha\approx 0.39 ), capturing discrete jump behavior that standard Normal distributions cannot replicate. This divergence has immediate economic significance: optimization under mis-specified distributions leads to incorrect asset weights and underestimation of risk exposures, particularly in turbulent regimes. Portfolio Distribution AIC BIC KS statistic Sharpe Normal -553.411 -548.201 0.10450 NIG -590.036 -579.615 0.04144 CVaR Normal -498.331 -493.121