Pith. sign in

REVIEW 3 major objections 4 minor 28 references

Performance of tail hedged portfolio with third moment variation swap

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A third moment variation swap moves a skewed, fat-tailed return distribution toward symmetric, Gaussian-like tails under both stochastic volatility and jump-diffusion models.

desk verdict Solid Heston-side numerics with a real convergence check; the SVJD section hides a mean-field approximation and the empirical demonstration is in-sample. read the letter →

arxiv 1908.05105 v1 pith:2ZVPPZKH submitted 2019-08-14 q-fin.PR q-fin.CPq-fin.PMq-fin.RM

Kyungsub Lee , Byoung Ki Seo This is my paper · ORCID
classification q-fin.PRq-fin.CPq-fin.PMq-fin.RM MSC 91G2091G80
keywords thirdmomentvariationvarianceswaptailriskhedgingskewnesskurtosisHestonmodeljumpdiffusionFokker-Planckequation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that a derivative contract called a third moment variation swap can reshape an asset's return distribution so that its negative skew and fat tails largely disappear. The contract's floating leg is the negative realized quadratic covariation between the return process and its squared process, so it pays when the asset drops sharply and costs when the asset rallies. The authors derive a partial differential equation for the joint density of the hedged portfolio return, the underlying return, and the stochastic variance, and they solve it with an alternating direction implicit finite-difference scheme. Under Heston stochastic volatility, a hedge ratio of about 30–40 makes skewness nearly zero and kurtosis minimal; under a stochastic volatility jump-diffusion model, a hedge ratio of about 45–60 does the same. If the calculations are correct, the swap offers a practical instrument for transferring skew and tail risk rather than merely measuring it.

What carries the argument

The central object is the third moment variation $[R,R^2]_t$, defined as the quadratic covariation between the return process and its squared process, with the swap's floating leg equal to $-[R,R^2]_T$. The carrying tool is the forward Kolmogorov equation for the joint density $f(x,r,v,t)$ of the portfolio return, underlying return, and variance; after a Fourier transform in the portfolio-return coordinate, the density is computed on an $(r,v)$ grid by the Peaceman–Rachford alternating direction implicit method. In the jump-diffusion version the generator becomes a partial integro-differential equation, and the portfolio's jump size for an underlying jump of size $z$ is $z-2\beta R_{s-}z^2-\beta z^3$, which the paper simplifies by replacing $R_{s-}$ with its expectation. The numerical density is converted to skewness and kurtosis through the characteristic function obtained from the transformed PDE.

What would settle it

Simulate the SVJD model with the true jump map $z_X(s)=z-2\beta R_{s-}z^2-\beta z^3$ for $\beta=45$ and $\beta=60$, estimate skewness and kurtosis of $X_T$ by Monte Carlo, and compare with Table 2; a discrepancy beyond Monte Carlo error would demonstrate that the mean-field simplification changes the conclusion.

Watch

Extended reading notes

Core claim

The discovery is a numerical demonstration that the third moment variation swap serves as a tail-risk hedge: at a suitable notional amount, the distribution of the total portfolio becomes close to symmetric and close to Gaussian in the tails. In the Heston model the computed skewness falls from −0.4281 at $\beta=0$ to −0.0671 at $\beta=30$ and 0.0600 at $\beta=40$, while kurtosis falls from 3.3741 to about 3.16; in the SVJD model skewness falls from −0.5955 to −0.1289 at $\beta=45$ and 0.0107 at $\beta=60$, with kurtosis dropping from 3.9757 to about 3.11. The paper obtains these numbers from the Fokker–Planck density, validates the code against the closed-form Heston characteristic function for the unhedged case, and confirms the shape against Monte Carlo histograms. The mechanism is that the floating leg is defined as minus the third moment variation, so a market drop, which makes the variation negative, delivers a positive payout that offsets the underlying loss, while a rally imposes a payment that trims the right tail.

Load-bearing premise

The jump-model calculations replace the actual pre-jump return by its expected value when determining the jump's effect on the hedged portfolio; if that shortcut is wrong, the SVJD densities and moments are approximate rather than exact.

Editorial extensions

If this is right

  • A single parameter, the swap notional $\beta$, controls the whole shape of the return distribution; under Heston the near-Gaussian window is roughly $\beta\in[30,40]$, and under SVJD it is roughly $\beta\in[45,60]$.
  • Because the fixed leg shifts only the mean, the distributional results apply regardless of the swap's fair price; a non-zero fixed leg changes expected return but not skewness or kurtosis.
  • The third moment variation swap does not require a model of investor preferences or a risk premium: the hedge works through the pathwise payoff structure under the physical probability measure.
  • The computed density can be used to locate the optimal hedge number by a quantile-matching criterion, and the paper's $\beta$ values agree with simulation-based optima of 38.42 (Heston) and 45.21 (SVJD).
  • For longer empirical horizons, QQ plots from S&P 500 five-minute data show thinner tails across $T=5,20,60,250$ days, indicating the effect is not an artifact of one maturity.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Inference: the mean-field substitution in the jump model is testable; a direct Monte Carlo simulation that uses the actual pre-jump return $R_{s-}$ in the jump-size map would show whether the reported SVJD moments are reliable, and could move the optimal $\beta$.
  • Inference: because the same Fokker–Planck machinery is model-agnostic, the hedging analysis could be repeated under models with non-affine variance dynamics or self-exciting jumps; the paper does not do this, but the PDE framework would extend naturally.
  • Inference: the method's output is a full density, so the hedge could be tuned to risk measures such as value-at-risk or expected shortfall rather than skewness and kurtosis; the paper stops at standardized moments.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper studies a 'third moment variation swap' whose floating leg is the negative realized third moment variation [R,R^2]_T, and analyzes the return distribution of a portfolio consisting of the underlying asset plus beta units of this swap. The portfolio log-return is approximated by X_T = R_T - beta[R,R^2]_T. The authors derive backward and forward PDEs for the joint density of (X,R,V) under the Heston stochastic volatility model and under a stochastic volatility jump-diffusion (SVJD) model, solve the transformed Fourier PDEs with an alternating direction implicit finite-difference scheme, and report skewness and kurtosis for various hedge ratios. They find that for Heston parameters with beta about 30-40, and for SVJD parameters with beta about 45-60, the hedged portfolio has near-zero skewness and kurtosis closer to the Gaussian value. Section 2 also presents an empirical illustration on S&P 500 five-minute returns from 1990-2007.

Significance. If the derivation were fully rigorous, the paper would offer a model-based pricing and hedging framework for skew and tail risk using a tradeable third moment variation, complementing variance swap and realized skewness literature. The Heston part is a genuine contribution: the PDE derivation is careful, the ADI discretization is described in detail, and the beta=0 numerical solution is checked against the known Heston characteristic function with RMSE 3.19e-4. The central claim for the Heston model is credible. However, the SVJD section contains an unstated mean-field approximation that changes the distribution being solved, and the empirical section fits hedge ratios in-sample; both issues affect how strongly the paper's broader conclusions can be drawn.

major comments (3)
  1. [Section 4, Eqs. (11)-(13) and Table 2] The exact portfolio jump at a jump time s is Delta X_s = z - 2 beta R_{s-} z^2 - beta z^3. In defining z_X(s) = z - 2 beta (mu - theta/2) s z^2 - beta z^3, the paper replaces the random pre-jump return R_{s-} by its expectation. This is a mean-field approximation, not a consequence of the stated assumption that the jump size z is independent of R_{s-}. The exact Fourier-transformed adjoint jump term would be lambda * integral e^{-i phi (z - 2 beta r z^2 - beta z^3)} \hat f(r-z,v) psi(z) dz, whereas Eq. (13) uses e^{-i phi z_X(t)} \hat f(r-z,v). These differ by a state- and time-dependent phase, and the discrepancy need not be small at the beta values reported in Table 2. The skewness and kurtosis figures in Table 2 and the density comparison in Figure 10 therefore solve an approximate model unless the approximation is justified by a convergence argument or by simulation using the exact pre-jump R_{s-} mapping. This is load-bearing for the SVJD claim.
  2. [Section 2 and Figure 1] The hedge ratios 242.9, 80.8, 46.2 and 16.2 are selected by minimizing the L2 distance between the realized quantiles of the hypothetical hedged portfolio and the corresponding normal quantiles on the same S&P 500 sample used for the QQ plots. Showing that the hedged portfolio looks more normal on that same in-sample period is partly self-fulfilling and does not establish out-of-sample hedging performance. The authors should either provide an out-of-sample exercise or clearly label Figure 1 as an in-sample calibration rather than evidence of hedging efficacy.
  3. [Section 4, text and Table 2 caption] The text states that the jump size standard deviation is sigma_j = 0.01, while the caption of Table 2 reports sigma_j = 0.02. Since the reported skewness, kurtosis, and optimal beta 45.21 depend on sigma_j, this inconsistency prevents reproduction of the numerical results and should be corrected.
minor comments (4)
  1. [Section 3.3] The convergence check in Figure 9 is only for beta = 0. The claim that the hedged portfolio computation has the same accuracy is plausible but not directly established, because the term 2 beta r v in the drift changes the operator.
  2. [Section 4, notation] The jump amplitude z_X(t) in the transformed PDE should be written consistently as z_X(s) with the integration variable s, since the mean-field replacement E[R_{s-}] = (mu - theta/2)s depends on the jump time.
  3. [Eq. (10)] The characteristic function formula contains several unmatched parentheses and would benefit from a cleaner typesetting to allow direct comparison with the numerical result.
  4. [Table 2 caption] The sentence 'Table 2 list the numerically computed...' should read 'Table 2 lists...'.

Circularity Check

1 steps flagged · score 4.0 of 10

Partial circularity: Section 2's empirical 'Gaussian-like' performance claim is forced by fitting the hedge ratio to minimize distance to normality on the same sample; the PDE-based central derivation is self-contained and independently validated.

  1. fitted input called prediction [Section 2.1, empirical S&P 500 analysis around Figure 1 (pp. 4-5)]
    "The hedge numbers are determined to minimize the squares of the differences between the quantiles of the realized portfolio return, hypothetically hedged by the swap, and the normal distribution function with the same mean and standard deviation with the portfolio distribution. In other words, the L2-norm of the difference between the empirical portfolio return and the corresponding normal distribution was minimized. ..."

    The swap position beta is chosen on the same full S&P 500 sample by minimizing the L2 distance between the hedged portfolio quantiles and normal quantiles. The subsequent claim that the hedged portfolio 'tends to have a more Gaussian-like distribution' restates exactly that minimized objective: the QQ-plot improvement is enforced by the fitting procedure rather than demonstrated on a holdout sample or by an independent prediction. The empirical performance conclusion is therefore statistically forced by construction.

full rationale

The central PDE derivation in Sections 3-4 is self-contained: the portfolio return X_T is defined from R and [R,R^2], the forward/backward equations are derived from the stated Heston/SVJD dynamics, and the numerical solver is validated against the known Heston characteristic function in Eq. (10), with RMSE converging to about 3.19e-4 as the partition shrinks. The parameter scans in Tables 1-2 study a free parameter beta rather than fitting beta and then calling the resulting moments a prediction; highlighting the range where skewness is near zero is ex post selection, but it is not an equivalence-by-construction. The one clear construction-forced step is the Section 2 empirical fit: the hedge ratio is calibrated to minimize the same quantile-to-normal distance that is then reported as evidence of hedging success. This supports a score of 4: partial circularity in an ancillary empirical demonstration, while the central claim still has independent content. The self-citations (Choe and Lee 2014; Lee 2015) define the swap and estimator but are not used as a load-bearing uniqueness theorem, so they do not add circularity. The SVJD jump-term replacement of R_{s-} by E[R_{s-}] in z_X(s) is an unstated mean-field approximation that can bias Table 2 and the Figure 10 comparison, but it is a correctness/validity concern rather than a circular reduction, so it is not counted as a circular step here.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The paper introduces no new particles, forces, dimensions, or unobserved entities. The third moment variation swap is taken from prior literature (Choe and Lee, 2014), and the model parameters are standard choices for stochastic volatility and jump diffusion models.

free parameters (2)
  • beta (hedge ratio of third moment variation swap) = 242.9, 80.8, 46.2, 16.2 for T=5,20,60,250 days in Section 2; scanned 0-60 in Sections 3-4; simulation 'optimal' 38.42…
    Beta is the number of swap contracts per unit of underlying. In the empirical section it is fitted to minimize the distance between portfolio quantiles and the normal distribution on the same data; in the theoretical sections it is scanned and the value showing near-zero skewness is highlighted.
  • Heston and SVJD model parameters (mu, kappa, theta, gamma, rho, lambda, sigma_j) = mu=0.05, kappa=18, theta=0.1 or 0.05, gamma=1, rho=-0.62, lambda=20, sigma_j=0.01 or 0.02
    These parameters are chosen by hand for the numerical demonstrations and are not calibrated to market data. The headline conclusion about symmetry and thin tails is demonstrated only for these parameter values.
assumptions (6)
  • standard math The asset return is a semimartingale and the third moment variation is the quadratic covariation [R,R^2]
    Section 2.1 defines the quantity using standard stochastic calculus from Protter (2013).
  • domain assumption Under the Heston model, return and variance follow the specified square root dynamics with Brownian shocks correlated by rho
    Section 3, page 7, assumes Heston dynamics for the return and variance process.
  • domain assumption The variance process stays positive because the Feller condition 2 kappa theta > gamma^2 holds
    Section 3.2, page 13, imposes this condition to justify zero boundary conditions and positivity.
  • domain assumption Jump sizes are normally distributed with zero mean and are independent of the return level; jump times follow a Poisson process with intensity lambda
    Section 4, page 18, assumes a normal density psi for jump sizes, Poisson arrivals, and independence of jump size and arrival time.
  • ad hoc to paper In the jump generator, the pre-jump return R_{s-} is replaced by its expectation E[R_{s-}] = (mu - theta/2)s when defining the jump size in the portfolio return
    Section 4, near Eq. (11), page 19, substitutes the expectation into the jump-size map z_X(s) without stating that this is a mean-field approximation rather than an exact generator.
  • domain assumption The PDE domain is truncated to a bounded region and the density is set to zero on the boundary
    Section 3.2, page 13, imposes zero boundary conditions for the numerical scheme, which is an approximation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Performance of tail hedged portfolio with third moment variation swap." pith.science (2026). https://pith.science/paper/2ZVPPZKH

@misc{pith2026190805105,
  author       = {Pith},
  title        = {Pith review of: Performance of tail hedged portfolio with third moment variation swap},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2ZVPPZKH}},
  note         = {Machine review of arXiv:1908.05105}
}
read the original abstract

The third moment variation of a financial asset return process is defined by the quadratic covariation between the return and square return processes. The skew and fat tail risk of an underlying asset can be hedged using a third moment variation swap under which a predetermined fixed leg and the floating leg of the realized third moment variation are exchanged. The probability density function of the hedged portfolio with the third moment variation swap was examined using a partial differential equation approach. An alternating direction implicit method was used for numerical analysis of the partial differential equation. Under the stochastic volatility and jump diffusion stochastic volatility models, the distributions of the hedged portfolio return are symmetric and have more Gaussian-like thin-tails.

Figures

Figures reproduced from arXiv: 1908.05105 by the authors.

Figure 1
Figure 1. QQ plots of the underlying asset returns (left) and the hedged portfolios’ returns [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. The dynamics of S&P 500 index (left) and the hedged portfolio’ value (right) from [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. The dynamics of the hedged portfolios’ values with transaction cost 0.2% (left) and [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: When φ = 0, the solution of the Eq. (1) is the joint probability density function of the return and variance in Heston’s model A transformed PDE with respect to y is represented by ∂gˆ ∂t =  −µ + 1 2 v + ργ ∂gˆ ∂r + v 2 ∂ 2 gˆ ∂r2 + {−κ(θ − v) + γ 2 } ∂gˆ ∂v + γ 2 2 …
Figure 5
Figure 5. Figure 5: The real (left) and imaginary (right) parts of the characteristic functions with [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]
Figure 6
Figure 6. Figure 6: Probability density functions and histograms of simulated data : underlying asset [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]
Figure 7
Figure 7. Figure 7: Probability density functions with various hedge numbers [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: Probability density function of the third moment variation [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]
Figure 9
Figure 9. Figure 9: Global truncation error with respect to partition size ∆ [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]
Figure 10
Figure 10. Figure 10: Probability density functions and histograms of simulated data under stochastic [PITH_FULL_IMAGE:figures/full_fig_p021_10.png]
Figure 11
Figure 11. Figure 11: Probability density functions with various hedge numbers [PITH_FULL_IMAGE:figures/full_fig_p022_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 24 canonical work pages

  1. [1]

    and Andreasen, J

    Andersen, L. and Andreasen, J. (2000). Jump-diffusion processes: Volatility smile fitting and numerical methods for option pricing. Review of Derivatives Research , 4:231--262

    work page 2000

  2. [2]

    G., Bollerslev, T., Diebold, F

    Andersen, T. G., Bollerslev, T., Diebold, F. X., and Labys, P. (2003). Modeling and forecasting realized volatility. Econometrica , 71:579--625

    2003

  3. [3]

    Bakshi, G., Kapadia, N., and Madan, D. (2003). Stock return characteristics, skew laws, and the differential pricing of individual equity options. Review of Financial Studies , 16:101--143

    work page 2003

  4. [4]

    Barndorff-Nielsen, O. E. and Shephard, N. (2002). Econometric analysis of realized volatility and its use in estimating stochastic volatility models. Journal of the Royal Statistical Society: Series B (Statistical Methodology) , 64:253--280

    work page 2002

  5. [5]

    Barndorff-Nielsen, O. E. and Shephard, N. (2004). Power and bipower variation with stochastic volatility and jumps. Journal of Financial Econometrics , 2:1--37

    work page 2004

  6. [6]

    and Cont, R

    Bentata, A. and Cont, R. (2009). Mimicking the marginal distributions of a semimartingale. arXiv preprint arXiv:0910.3992

    arXiv 2009

  7. [7]

    and Cont, R

    Bentata, A. and Cont, R. (2015). Forward equations for option prices in semimartingale models. Finance and Stochastics , 19:617--651

    work page 2015

  8. [8]

    Choe, G. H. and Lee, K. (2014). High moment variations and their application. Journal of Futures Markets , 34:1040--1061

    work page 2014

Show all 28 references
  1. [9]

    Christoffersen, P., Heston, S., and Jacobs, K. (2006). Option valuation with conditional skewness. Journal of Econometrics , 131:253 -- 284

    2006

  2. [10]

    and In't Hout, K

    Haentjens, T. and In't Hout, K. J. (2012). Alternating direction implicit finite difference schemes for the Heston-Hull-White partial differential equation. Journal of Computational Finance , 16:83--110

    2012

  3. [11]

    and In't Hout, K

    Haentjens, T. and In't Hout, K. J. (2015). ADI schemes for pricing american options under the heston model. Applied Mathematical Finance , 22:207--237

    2015

  4. [12]

    Hansen, P. R. and Lunde, A. (2006). Realized variance and market microstructure noise. Journal of Business and Economic Statistics , 24:127--161

    2006

  5. [13]

    Hanson, F. B. (2007). Applied Stochastic Processes and Control for Jump-Diffusions: Modeling, Analysis, and Computation . Society for Industrial and Applied Mathematics

    2007

  6. [14]

    and Siddique, A

    Harvey, C. and Siddique, A. (2000). Conditional skewness in asset pricing tests. Journal of Finance , 55:1263--1295

    2000

  7. [15]

    Harvey, C. R. and Siddique, A. (1999). Autoregressive conditional skewness. The Journal of Financial and Quantitative Analysis , 34:465--487

    1999

  8. [16]

    Heston, S. (1993). A closed-form solution for options with stochastic volatility with applications to bond and currency options. Review of Financial Studies , 6:327--343

    1993

  9. [17]

    and Foulon, S

    In't Hout, K. and Foulon, S. (2010). ADI finite difference schemes for option pricing in the Heston model with correlation. International Journal of Numerical Analysis and Modeling , 7:303--320

    2010

  10. [18]

    and Kim, J

    Jeong, D. and Kim, J. (2013). A comparison study of ADI and operator splitting methods on option pricing models. Journal of Computational and Applied Mathematics , 247:162 -- 171

    2013

  11. [19]

    Kozhan, R., Neuberger, A., and Schneider, P. (2013). The skew risk premium in the equity index market. Review of Financial Studies , 26:2174--2203

    2013

  12. [20]

    and Litzenberger, R

    Kraus, A. and Litzenberger, R. H. (1976). Skewness preference and the valuation of risk assets. The Journal of Finance , 31:1085--1100

    1976

  13. [21]

    Lee, K. (2015). Probabilistic and statistical properties of moment variations and their use in inference and estimation based on high frequency return data. Studies in Nonlinear Dynamics & Econometrics

    2015

  14. [22]

    Mykland, P. A. and Zhang, L. (2009). Inference for continuous semimartingales observed at high frequency. Econometrica , 77:1403--1445

    2009

  15. [23]

    Neuberger, A. (2012). Realized skewness. Review of Financial Studies , 25:3423--3455

    2012

  16. [24]

    Pappalardo, L. (1996). Option pricing and smile effect when underlying stock prices are driven by a jump process. Working Paper

    1996

  17. [25]

    Peaceman, D. W. and Rachford, Jr, H. H. (1955). The numerical solution of parabolic and elliptic differential equations. Journal of the Society for Industrial and Applied Mathematics , 3:28--41

    1955

  18. [26]

    Protter, P. E. (2013). Stochastic integration and differential equations . Springer

    2013

  19. [27]

    Schoutens, W. (2005). Moment swaps. Quantitative Finance , 5:525--530

    2005

  20. [28]

    Wang, C. D. and Mykland, P. A. (2014). The estimation of leverage effect with high-frequency data. Journal of the American Statistical Association , 109:197--215

    2014

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.