Risk Model
Understanding the risk
There are many risks of offering permissionless leverage on any asset. The greatest risk however is with thinly traded assets with potentially no underlying value; it is not very expensive for a single user to 1000x the price of the asset or take the asset to 0 in a literal block. This means that in theory the market can accumulate bad debt in a literal instant; much faster than any liquidator can liquidate positions.
More common are "run of the mill" market manipulations. Users can manipulate markets against each other; however if exposed to market manipulation risks, Imperial's LPs will get fleeced in tiny markets.
Traditional risk models include measures for:
Price Volatility
Market concentration
Unfortunately in this environment, these protections are easily routed. For volatility, a user could ensure the market looked stable, then having a large percentage of the tokens, could dump them in an instant. Similarly while market concentration could in theory be a useful way to tell how risky a market is (more concentrated market in a few hands being more risky) any ways to measure this are futile in a permissionless world where a user can circumvent protections by spreading their position across multiple wallets.
Core Risk Model
Imperial's core risk model is based on delta neutrality. Imperial's markets do everything in their power to incentivize users of those markets to leave those markets holding no directional risk. When users open/close a position, the Imperial market trades with an underlying AMM pool. It buys the asset with the opening of a long and sells the asset with the opening of a short (if it has any to sell). The entry/exit price the user gets on their position is the price the Imperial market got trading the underlying asset, preventing market manipulation tactics such as sandwich attacks.
Notably however, Imperial markets are exposed to 2 core risks directly related to this delta neutral risk strategy:
"Short attack" - Imperial cannot maintain delta neutrality if a market is net-short. Users can take a market net short, then tank the underlying asset price leading to unmitigated losses to the market and LP pool.
"Long attack" - Also could be referred to as a "pump & dump" scheme, where a user uses the act of the market buying tokens to maintain delta neutrality as a way to introduce an artificial buyer for their own positions in the token.
Delta Neutral Positions
Target long/short ratio
Each market has a target long/short ratio where the market is slightly long. The mental model for this is similar to a shuffle board. The most capital efficient long/short ratio is where there are an equal number of longs to shorts; however because the market cannot hedge risk if it is net short it wants to be slightly net long to act as a buffer from "falling off the table".
The protocol can maintain delta neutrality if it is net long by buying the underlying asset. When a user opens a short, the user is opening a position against existing longs. Note there is no way hedge risk if there are more shorts than longs.
Market Scenarios
1:1 Ratio Longs/Shorts
In an even market between longs/shorts, from a capital efficiency perspective this is the best place for a market to be. This is because the market is delta neutral yet needs to own no tokens to maintain it's delta neutral position; however as noted earlier, markets that are perfectly balanced have no buffer if a long decides to close their position.
Net Long
In a net long market, the market maintains delta neutrality by buying the underlying asset to pay the longs with in the event they close their positions. The LP pool's assets are used for these purchases. The market is protected from large price swings up; however, if the price of the underlying asset drops faster than liquidators can liquidate positions, the pool's LPs might be stuck in a worthless asset. This is a serious risk of the protocol which is why before the protocol allows a market to trade, market tokens must guarantee the market in exchange for backstop funds. This mechanic is described in detail later.
Net Short
In a net short market, the market loses it's delta neutrality, as the only way to short a market is against longs. If those longs close, the market is exposed to losses if the market drops in value. A quadratic, dynamic fee structure is designed to ensure the market doesn't go net short. This mechanic is described later in detail.
Healthy Market Mechanics
Short Limits
The first line of defense the market has to maintain it's slightly net long position is to not allow users to open shorts if opening that short would take the market net short. Notably however; the protocol allows users to close longs at any time. A slightly savvy user could circumvent this restriction by opening some longs, then some shorts, then closing their longs to achieve the desired exposure.
Fees
Fees are the secondary mechanism the market has to maintain delta neutrality. Fees act as a nudge, incentivizing users to open/close positions to keep the market healthy; they also act as a final deterrent preventing users from opening/closing positions that take the market into deep short exposure. Note that fees rise to 100% of the position size essentially prohibiting a user from opening/closing a position if a market is in deep unhealthy territory and their position change will worsen the health of the market.
Unconventionally to other perpetual dexes, fees change on a user both on position opens AND on position closes
Typical protocols adjust open fees, but keep close fees constant. Imperial updates close fees in addition to open fees. This requires users to understand both the price movements of the underlying market, but also the long/short ratio of the Imperial market and how their positions will affect the health of that market. It is a new metagame for perpetuals but an old one for veterans of the crypto game at the same time.
Market Backstop
Because Imperial has no way to control the underlying market, it is exposed to bad debt losses if the underlying market drops before positions can be liquidated. To hedge these losses (and ensure they don't become an attack vector), the protocol requires backstop funds for each market. These are funded partly by the initial sale of market tokens, and partly by ongoing fee collection. This mechanic is in place primarily to prohibit long attacks.
Long Attack
When the market hedges positions to maintain delta neutrality, it puts buy pressure on the asset market. This buy pressure can be used to launch an attack against the protocol. Because these markets are so small, a user can attack the market by using the protocol as the buy pressure for their assets. For this to be a risk to the protocol however, the user must sell all their assets before liquidators can liquidate positions.
For a long attack to have a material impact on the protocol, the user attacking the protocol must be a large token holder of the underlying asset. The idea of the backstop is the backstop is an influential holder of the underlying asset that feels comfortable risking assets if the underlying market tanks faster than positions can be liquidated.
This means that in effect, those who pay fees into the market and those who originally bought market tokens funded this backstop and in turn, it's treasury of that asset, and are in a sense, net "long" that asset while also believing in it's volume over time.
Element of time
Notice that if the price of the asset drops over time, liquidators have the opportunity to liquidate positions before they represent bad debt to the LP Pool. One of the key elements of the risk is the element of time. On Solana with an AMM, the user can sell all the tokens in a single block; essentially in an instant, making this a legitimate risk to to the protocol, that can only be hedged, not eliminated entirely.
Backstop Fund Dynamics
As far as the risk model is concerned, the backstop fund has two jobs:
Hedge bad debt losses
Socially verify markets
When bad debt occurs, the backstop fund is the first one to take losses. The backstop fund also acts as a social layer letting the market take on the risk of offering leverage on this risky asset.
Last updated