Pricing S&P 500 Variance Futures Based on the Instantaneous Variance of S&P 500 Index with Double
Jump Processes
Shu Su1, Jiling Cao2, Xinfeng Ruan3, Wenjun Zhang2
1 School of Applied Business
Unitec Institute of Technology - Part of Te P¯ukenga 2 School of Engineering, Computer and Mathematical Sciences
Auckland University of Technology 3 Otago Business School
University of Otago
December 7, 2022
1 Introduction Background Literature review
2 Prcing S&P 500 Varaince Futures Data
Variance Futures Valuation Variance Risk Premium
Model Calibration and Estimation Process, and Results
3 Conclusion and Future Work
4 Reference
Overview
Motivation:
To better understand the changes in the U.S. stock market.
An alternative way to observe and forecast the volatility of the U.S.
stock market to reduce investment risk, leverage volatility, and diversify a portfolio.
Pricing the variance futures contract with new specifications has not been studied in the literature.
What did we do?
Explore the feature of the Standard & Poor’s 500 index (SPX) price evolution.
Discover the relationship among the S&P 500 index, CBOE volatility index (VIX), and the price of S&P 500 variance future (VA).
Deduce an accurate VA pricing formula based on the returns of SPX.
Background - SPX and VIX
Standard & Poor’s 500 Index (SPX)
Market-capitalization-weighted index of 500 leading publicly traded companies in the U.S.
Launched in 1957 by the credit rating agency Standard and Poor’s One of the best gauges of large U.S. stocks, even the entire equities market
CBOE volatility index (VIX)
The first benchmark index to measure the market’s expectation of future (30-day) volatility (Market’s ”fear gauge”)
Based on real-time prices of options on the S&P 500 Index Not tradable but can be speculated through the VIX derivatives.
SPX VS VIX
Source from: www.cboe.com/us/indices/dashboard/vix/
Background - VA
Summary product specifications for S&P 500 variance future (VA) Exchange-traded futures contracts based on the realized variance of SPX.
Launched in 2012 and listed Chicago Board Options Exchange (CBOE).
The final settlement value: the realized variance of the S&P 500 measured from the time of initial listing until the expiration of the contract.
Contract size: the contract multiplier for the S&P 500 Variance futures contract is$1 per variance unit
Literature Review
[Zhang and Huang(2010)]
Derived a formula for pricing S&P 500 3-month variance futures Found a linear relationship between the variance futures price and the value of VIX squared
The volatility risk premium is negative
[Chang et al.(2013)Chang, Jimenez-Martin, McAleer, and Amaral]
Three conditional volatility models (GARCH, GJR, and EGARCH) with three different probability densities (Gaussian, Student-t, and
Generalized Normal distribution) to value the 3- and 12-month variance futures
The value of VIX is highly correlated to the prices of 3- and 12-month variance futures
S&P 500 Variance Futures Data
42 S&P 500 variance future contracts
Contract expiration date: between 17/01/2014 and 18/12/2020 Trading period: from 24/07/2013 to 14/02/2020
Excluding the number of days between issue data and trade date is less than 90 days.
Average daily settle price: between $113.77 and$441.92
The contract with a longer trading period has a higher daily average settle price, and vice versa
Table: Descriptive statistics of S&P 500 variance futures contracts
Code No. of Obs. No. of Days Settle Price
Mean Std.
F14 54 144 188.9080 44.3120
H14 97 187 235.8399 50.3069
M14 250 369 289.3102 82.4285
U14 160 250 231.2168 67.7068
X14 87 339 228.5410 85.4947
Z14 377 496 319.3339 84.3104
F15 14 105 297.5783 146.1982
H15 159 250 313.5196 78.8818
M15 120 211 293.6429 102.3710
U15 159 250 348.1140 181.1643
Source from: www.cboe.com
Variance Futures Valuation
Issue date t0
Trade date t
Settlement date(Maturity date) T
Realized variance Implied variance strike
Let FtT be the price of variance future at timet, then we have, FtT
10,000 = 1 T −t0
h
(t−t0) RV +EQt
QVt,Ti
. (1)
where
RV = annualized realized variance = 252×PN−1 i=1
Ri2 N−1
R = log return of the S&P 500 index = ln(S /S)
Let {lnSt} be the log return of SPX log return, under physical measure P, dlnSt =
a−ΛPtk¯P+ηSvt−1 2vt
dt+√
vtdWS,tP +dJS,tP , dvt =κv(mt−vt)dt+σv
√vtdWv,tP +dJv,tP , dmt= [κm(θm−mt) +ηmmt]dt+σm√
mtdWm,tP ,
(2)
Under Qmeasure, we have dlnSt =
r−q−ΛQtk¯Q− 1 2vt
dt+√
vtdWS,tQ +dJS,tQ , dvt = [κv(mt−vt) +ηvvt]dt+σv√
vtdWv,tQ +dJv,tQ , dmt =κm(θm−mt)dt +σm√
mtdWm,tQ ,
(3)
Risk premiums for the index return (ηSJ) and the variance (ηvJ) are ηSJ = ΛPt
µPJ +ρJµPv
−ΛQt
µQJ +ρJµQv
, (4)
ηvJ = ΛPtµPv −ΛQtµQv. (5)
Variance Risk Premium
The variance risk premium (VRP) measures the average difference between the realized variance and the variance swap strike.
VRPt,T = 1 T −t
h
αQVP −αQQV
mt+
βQVP −βQVQ
vt+
γQVP −γQVQ i
. where
αQQV=
σ2J+ρ2J(µQv)2+
µQJ +ρJµQv
2
λQ1αQv +λQ2αQm
+αQv, βQQV=βvQ
1 +λQ1
σJ2+ρ2J(µQv)2+
µQJ +ρJµQv
2 , γQVQ =
σ2J+ρ2J(µQv)2+
µQJ +ρJµQv2 h
λQ0(T −t) +λQ1γQv +λQ2γmQi +γvQ, αPQV=
σ2J+ρ2J(µPv)2+
µPJ+ρJµPv2
λP1αPv +λP2αPm +αPv, βPQV=βvP
1 +λP1
σJ2+ρ2J(µPv)2+
µPJ+ρJµPv
2 ,
Model Calibration and Estimation Process, and Results
Model Calibration and Estimation Process
Using the joint data set, including the S&P 500 index, the VIX, and the variance futures data.
Applying Euler approximation on the latent processes, {m}t≥0 and {v}t≥0 in Eq. (2)
Applying joint estimation method which combines the Maximum Log Likelihood (MLE) and Unscented Kalman filter (UKF)
Estimation Results
Table:Parameters in Eq. (2)
Parameter ρ κm κv θm µQv σm σv σJ
Value -0.64 0.08 7.01 0.00 4.05 0.01 0.35 0.43 ηv ηm λQ0 λQ1 λQ2 µQJ ρJ σe
-5.61 0.28 0.00 1.17 20.95 4.05 -0.94 0.01
Market Data VS. Estimation Data
Jul 2013 Jan 2014 Jul 2014 Jan 2015 Jul 2015 Jan 2016
Trade Date 0
200 400 600 800 1000
Settle price
Z15 Market data
SVSCJ SVSMJ SVSMCJ-R SVSMCJ
Jan 2016 Jul 2016 Jan 2017 Jul 2017 Jan 2018 Jul 2018 Jan 2019
Trade Date 200
300 400 500 600 700
Settle price
Z18
Market data SVSCJ SVSMJ SVSMCJ-R SVSMCJ
400 600 800
Settle price
Z19 Market data
SVSCJ SVSMJ SVSMCJ-R SVSMCJ
Conclusion and Future Work
Sudden and quick changes occur in the evolution processes of both the SPX returns and its variance.
The long-term mean of the SPX variance changes randomly during the selected period.
The rates of occurring sudden and quick changes depend on all latent processes.
Our pricing model yields more accurate estimation results for the variance futures contracts with a time-to-maturity that is more than or equal to 180 days.
The model cannot capture the size of the sudden decrease in the prices of VF contracts well, but can capture the decreasing tendency Apply the particle filter to improve the model estimation performance.
Reference
Chang, C., J. Jimenez-Martin, M. McAleer, and T. Amaral, 2013, The rise and fall of S&P 500 variance futures, North American Journal of Economics and Finance 25, 151–167.
Zhang, J., and Y. Huang, 2010, The CBOE S&P 500 three-month variance futures, Journal of Futures Markets 30, 48–70.