Equity Protection Swap (EPS) Explained

• EPS is a derivative contract adding a terminal cash flow based on the simple return of a reference portfolio, providing partial downside protection and conditional upside sharing.
• It employs static replication using European options to achieve transparent, path-independent pricing and hedging without altering the invested principal.
• Different configurations such as basic, buffer, and floor forms adjust loss and gain participation to optimize risk management in superannuation accounts.

An Equity Protection Swap (EPS) is a stand-alone derivative contract written on the simple return of a reference equity portfolio over a fixed horizon $[0,T]$. It was proposed as a new class of investment insurance product for holders of superannuation accounts in Australia, with the central idea that the holder obtains partial protection against losses on a reference portfolio and, in exchange, agrees to share portfolio gains with the protection provider if realized returns exceed a predetermined threshold. In the original formulation, the contract is transparent, path-independent, option-replicable, and can be configured so that the fair premium at inception is zero, so that protection is financed by conditional upside sharing rather than by an upfront premium (Xu et al., 2023).

1. Formal definition and payoff architecture

Let $S_t$ denote the value of the reference portfolio and define the simple return over $[0,T]$ by

$R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$

The EPS does not touch the principal invested in the portfolio. Instead, it adds a single terminal cash flow at $T$ that is a piecewise-linear function of $R_T$. From the provider’s perspective, this cash flow per unit notional is $\Phi(R_T)$, called the adjusted return; from the holder’s perspective, the net portfolio return is $R_T-\Phi(R_T)$. Economically, the overlay partially reimburses losses and claws back a share of sufficiently large gains (Xu et al., 2023).

In the generic specification, $\Phi\colon [-1,\infty)\to \mathbb{R}$ is continuous, piecewise-linear, non-decreasing, and satisfies $\Phi(0)=0$. The return space is partitioned by breakpoints

$S_t$0

with marginal participation rates $S_t$1. The derivative of the adjusted return is stepwise constant,

$S_t$2

If $S_t$3, a small increment $S_t$4 changes the provider’s payoff by $S_t$5, and the holder keeps $S_t$6 of that marginal return (Xu et al., 2023).

The paper re-parameterizes the contract by loss thresholds $S_t$7, gain thresholds $S_t$8, loss-side participation rates $S_t$9, and gain-side participation rates $[0,T]$0. This yields a natural decomposition into a protection leg and a fee leg,

$[0,T]$1

where $[0,T]$2 on the loss region and $[0,T]$3 on the gain region. The structure is therefore modular: downside insurance is specified by the protection leg, while financing is embedded in the upside fee leg (Xu et al., 2023).

2. Standard contract forms

The simplest specification is the basic proportional EPS. On the loss side, the provider takes a constant share $[0,T]$4 of all negative returns; on the gain side, the provider takes a constant share $[0,T]$5 of all positive returns. The provider payoff is

$[0,T]$6

Using

$[0,T]$7

the payoff is exactly a position in one put and one call struck at $[0,T]$8. In economic terms, the holder buys a fraction $[0,T]$9 of loss protection and sells a fraction $R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$0 of upside participation (Xu et al., 2023).

More practically relevant designs introduce thresholds. A buffer EPS uses a loss buffer $R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$1 and a gain threshold $R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$2, together with participation rates $R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$3 and $R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$4. Its provider payoff is

$R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$5

The contract is dormant when $R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$6; below $R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$7 the provider indemnifies the holder, and above $R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$8 the provider receives a share of excess gains (Xu et al., 2023).

A floor EPS modifies the loss-side shape by capping the provider’s total payout once returns fall below a specified floor $R_T := \frac{S_T-S_0}{S_0}, \qquad R_T \ge -1.$9. With gain buffer $T$0, loss participation $T$1, and fee participation $T$2, the provider payoff is

$T$3

For moderate losses $T$4, the provider covers a fraction $T$5 of the loss; for very large losses $T$6, payout is capped at $T$7; for gains above $T$8, the holder pays a fraction $T$9 of excess gains (Xu et al., 2023).

The original paper also defines proportional, buffer, cap, and buffer-cap fee legs, together with proportional, buffer, floor, and buffer-floor protection legs. This suggests that EPS should be understood less as a single contract template than as a structured family of return-based overlays whose geometry is determined by breakpoints and participation rates (Xu et al., 2023).

3. Relation to adjacent instruments and recurring misconceptions

EPS is explicitly compared with a total return swap (TRS) and a Registered Index-Linked Annuity (RILA). Like a TRS, it transfers equity risk between two parties through a swap-style settlement with no principal exchange. Unlike a TRS, however, the holder does not exchange the total return on the reference asset; instead, only a piece of downside and a piece of upside above a threshold are transferred. The holder retains a base exposure to the reference portfolio rather than giving up all upside and downside (Xu et al., 2023).

The comparison with RILAs is more structural. Both EPS and RILAs offer partial downside protection and limited upside, and both can be decomposed into portfolios of European options. The difference is institutional and contractual. A RILA is a variable annuity product with insurer-managed account mechanics, mortality and longevity features, and embedded guarantees and fees. An EPS is a pure derivative overlay with no principal transfer, no mortality or income features, and direct pricing and hedging through index options. Later work retains this interpretation and treats EPS as an insurance-based overlay that can be used with or without variable annuities (Rutkowski et al., 25 May 2026).

A frequent misunderstanding is to treat EPS as equivalent to a protective put. The papers do not make that identification. A protective put guarantees a floor on the entire portfolio and is financed by an explicit premium; an EPS can mimic a fractional protective put through its protection leg, but the protection is financed by conditional upside sharing rather than necessarily by an upfront premium. A second misunderstanding is to view EPS as a path-dependent portfolio insurance rule. The original framework states the opposite: the payoff depends only on $R_T$0, hence only on terminal $R_T$1. There are no rebalancing rules, no lookback features, and no optimal policy decisions, which sharply distinguishes EPS from CPPI and from dynamic variable-annuity riders such as GMWB arrangements (Xu et al., 2023).

4. Static replication, arbitrage-free pricing, and fair design

The central technical result is a model-free static hedging decomposition into European options on the reference index. The pricing framework assumes frictionless and arbitrage-free markets, liquidly traded European calls and puts with maturity $R_T$2, and a risk-neutral measure under which discounted prices are martingales. Because the implementation can use market option prices directly, the EPS price is model-free whenever option markets are sufficiently rich and liquid (Xu et al., 2023).

For the buffer EPS,

$R_T$3

with strikes $R_T$4 and $R_T$5, one obtains

$R_T$6

The provider’s static hedge is therefore long $R_T$7 units of the put and short $R_T$8 units of the call, so that the hedge payoff equals $R_T$9. The floor EPS is replicated by a put spread on the loss side plus a short call on the gain side (Xu et al., 2023).

For the generic EPS, the static hedge takes the finite linear form

$\Phi(R_T)$0

where $\Phi(R_T)$1 and $\Phi(R_T)$2. Hence the fair premium per unit notional is

$\Phi(R_T)$3

The article’s design criterion for a fair EPS is $\Phi(R_T)$4, meaning that the hedge costs exactly zero at inception and the holder pays no upfront premium. A typical design sequence is: first choose the desired downside protection profile $\Phi(R_T)$5; then solve for fee rates $\Phi(R_T)$6 using option prices so that the premium vanishes (Xu et al., 2023).

For a one-year buffer EPS with a single loss threshold and a single gain threshold, this zero-premium condition becomes

$\Phi(R_T)$7

so that

$\Phi(R_T)$8

The reported comparative statics are monotone: higher protection participation $\Phi(R_T)$9, more generous loss protection, or a higher gain threshold $R_T-\Phi(R_T)$0 all require a higher fee participation $R_T-\Phi(R_T)$1 to preserve zero premium (Xu et al., 2023).

5. Numerical behavior and empirical evidence

For numerical illustrations, the original paper embeds the option-based pricing formula in a Black-Scholes setting,

$R_T-\Phi(R_T)$2

with illustrative parameters $R_T-\Phi(R_T)$3, $R_T-\Phi(R_T)$4, $R_T-\Phi(R_T)$5, and $R_T-\Phi(R_T)$6 or $R_T-\Phi(R_T)$7 years. These calculations are not the core pricing claim, since the paper emphasizes that implementation can instead rely on observed option prices, but they show how fair premiums and fee rates vary with protection shares, thresholds, and maturity (Xu et al., 2023).

The empirical study uses daily closing levels of the S&P 500 and S&P/ASX 200 from January 2020 to December 2022 and computes realized trailing one-year returns. Several fair EPS profiles were calibrated on 2 February 2022 using market prices of one-year European options on the S&P 500. The six tested contracts were the following (Xu et al., 2023):

EPS profile	Parameters	Fee rate
Buffer 1	$R_T-\Phi(R_T)$8	$R_T-\Phi(R_T)$9
Buffer 2	$\Phi\colon [-1,\infty)\to \mathbb{R}$0	$\Phi\colon [-1,\infty)\to \mathbb{R}$1
Buffer 3	$\Phi\colon [-1,\infty)\to \mathbb{R}$2	$\Phi\colon [-1,\infty)\to \mathbb{R}$3
Floor 1	$\Phi\colon [-1,\infty)\to \mathbb{R}$4	$\Phi\colon [-1,\infty)\to \mathbb{R}$5
Floor 2	$\Phi\colon [-1,\infty)\to \mathbb{R}$6	$\Phi\colon [-1,\infty)\to \mathbb{R}$7
Floor 3	$\Phi\colon [-1,\infty)\to \mathbb{R}$8	$\Phi\colon [-1,\infty)\to \mathbb{R}$9

In the market downturn window from May 2021 to December 2022, using 164 S&P 500 and 165 ASX 200 one-year trailing returns that were predominantly negative, the original S&P 500 return distribution had minimum approximately $\Phi(0)=0$0 and median approximately $\Phi(0)=0$1. Under Buffer 2, the minimum net return improved from $\Phi(0)=0$2 to $\Phi(0)=0$3. Under Floor 3, the minimum improved further to approximately $\Phi(0)=0$4, with more uplift in moderate loss quantiles. Because large positive returns were rare in that interval, realized upside forfeiture was limited, so the contracts exhibited substantial downside mitigation with little realized cost during the downturn (Xu et al., 2023).

Over the full 2020-2022 sample, using 499 S&P 500 and 502 ASX 200 one-year trailing returns, the S&P 500 original returns ranged from roughly $\Phi(0)=0$5 to $\Phi(0)=0$6 with median about $\Phi(0)=0$7. Strong EPS profiles still reduced the left tail markedly: under Floor 3, the minimum improved from $\Phi(0)=0$8 to about $\Phi(0)=0$9. This left-tail improvement came with explicit upside truncation: under Buffer 1, the maximum net return fell from $S_t$00 to $S_t$01, while under Floor 1 it fell to around $S_t$02. The reported distributional effect is a narrowing of returns, with smaller variance and shorter tails. The empirical conclusion is that EPS works as designed: it trades some upside tail for protection against large downside and can substantially reduce tail risk and variance of annual returns over one-year superannuation-type horizons (Xu et al., 2023).

6. Extensions, limitations, and subsequent developments

The original framework acknowledges several limitations. Static hedging relies on liquid index options; for bespoke portfolios, hedging becomes basket-option-like and less exact. If option strikes or maturities are illiquid, interpolation, extrapolation, or model-based synthesis becomes necessary. Basis risk arises whenever the actual superannuation portfolio deviates from the reference index. Provider default risk is also material, and the paper suggests collateralization or provision by large super funds as possible responses. Finally, Black-Scholes simulations under-represent crash dynamics, motivating work under jump-diffusion and stochastic-volatility models (Xu et al., 2023).

Two subsequent arXiv papers develop these directions explicitly. One extends EPS to cross-currency reference portfolios relevant for internationally diversified retirement savings. It distinguishes separate domestic and foreign EPSs from integrated EPSs written on aggregated effective or aggregated quanto returns, and it studies exact static hedging with basket options together with superhedging via single-asset European options when bespoke basket options are unavailable. That paper also uses Monte Carlo simulation, geometric averaging, and moment matching to price the relevant basket options and reports that geometric and moment-matching approximations track Monte Carlo prices extremely well in the studied setting (Rutkowski et al., 2024).

A second extension studies valuation and hedging under Merton jump-diffusion and independent random-time default. Its core result is that the static option decomposition of EPS remains valid under jumps, so jump risk changes the initial hedge cost but not the algebraic replication at maturity. The same paper shows that counterparty default introduces an unhedgeable residual loss because long put positions in the hedge may fail to pay in crisis states. It therefore defines a default-adjusted initial premium $S_t$03 that adds expected default loss to the no-default hedging cost, and its numerical analysis highlights the sensitivity of EPS premiums and hedging costs to jump intensity, jump severity, and default intensity (Rutkowski et al., 25 May 2026).

A related but conceptually distinct line of research studies default-dependent equity claims in an equity-plus-CDS market. Within that framework, default-contingent EPS-type claims can be utility-indifference priced, and the combination of equities and CDS completes the market accounting for default when the underlying equity market is complete absent default. That result suggests a broader interpretation of EPS as part of a modular family of equity-risk-transfer contracts whose hedge set may include both vanilla options and credit derivatives, depending on the source of protection being engineered (Fei et al., 10 Apr 2025).