A deep-dive on fees in DeFi derivatives

Co-authored by Kevin Britz & Jacob Phillips

Derivatives in DeFi are heating up — but so are the fees. Fees in many derivatives protocols are multiples of what you’d see on CEXs (sometimes even 10x!). Traders are getting rekt & protocols/developers building structured products are unable to produce profitable strategies in this environment. For on-chain derivatives protocols to be sustainable & competitive over the long-term, fees will have to come down. The dream of building a whole new financial ecosystem upon derivatives is squandered if high fees are a necessary part of operation.

In this piece, we discuss each of the different types of fees users may encounter in on-chain derivatives protocols, the logic for why these exist, and how Perennial reduces/eliminates fees for users.

Taker Fees & Slippage

In mechanism design of on-chain derivatives protocols, taker fees & and slippage are related.

Taker Fees: Here, we define taker fees as costs imposed on takers when opening, closing, or adjusting positions. This will include both open/close fees AND price impact (costs baked into pricing).

The reason we group the two is that in DeFi derivatives mechanism design, the open/close fees & price impact act as effectively same mechanism used to accomplish the same outcome. Taker fees, some combination of taker fees & price impact, or price impact can be used to defend against:

1. Toxic flow — When the current market price deviates from the price of an on-chain market, a risk-free arbitrage opportunity exists. Taker fees or price impact can be used to minimize or eliminate the profitability of this arb opportunity.
2. Market manipulation — Creating a significant short-term deviation between the on-chain price and the price on external venues to profit. Taker fees & price impact can be used to raise the cost of performing this manipulation to a point that is prohibitively expensive/unprofitable.

Another way of putting it — Consider the mechanisms of two protocols: A) price impact but no open/close fees (think traditional AMM-style), and B) open/close fees but no price impact (think: Perennial / GMX). The two protocol designs work effectively the same way. If ETH is $2000, and a trader comes to open a 1 ETH position, in protocol A, the trader incurs price impact (to defend against the risks discussed), and the trade will execute at $2002, incurring 10 bps of cost. In protocol B, the trade executes directly at the current price of $2000 (with no price impact), but the trader also incurs a 0.10% trade fee, for a total trade cost of $2002. The outcome is the same. The only difference is that protocol A bakes the additional cost into the price, while protocol B separates it into two components.

So taker fees vs. price impact — it’s just semantics. We could easily reframe this “open/close fee” as a spread on the market price. And while this type of fee is traditionally thought of as static, there’s no reason this can’t be a variable, increasing with taker demand, creating synthetic price impact.

So for the sake of discussion here, taker fees & price impact can be thought of as the same mechanism.

Slippage: The difference between the attempted execution price and the settlement price. This can either be positive (price moved in a user’s favor) or negative (price moved against the user).

In on-chain derivatives protocols, this often takes the form of ‘delayed settlement.’ In this model, traders request to open a position at time t, but their request is not filled (and priced) until t+1, meaning that the trader can’t know their exact execution price beforehand, preventing toxic flow. Additionally, this trader is locked into that position for the delayed settlement period (and again during the settlement process when closing the position), indirectly making oracle manipulation unprofitable (longer time to manipulate the market = higher cost). The length of this delayed settlement is a parameter that can be tuned (high or low, fixed or variable). So here, we see that slippage can be used to prevent toxic flow and market manipulation, just like taker fees.

So to recap: taker fees (price impact) & slippage (delayed settlement) are two separate mechanisms that can be used to accomplish the same goals.

This gives mechanism designers two separate levers to work with, forming a spectrum of possible protocol designs:

Many protocols opt for the high fees (or price impact), leading to effective costs for opening a position on the order of 10–30 bps or even more (compare this to 2–4 bps on CEXs). This means active traders bleed P&L, and vaults with frequent rebalancing become entirely unprofitable.

Perennial instead leans on delayed settlement (initially). This allows Perennial to keep trading fees low (or even not have them at all). And since slippage can be positive or negative, on average, this nets out to little/no effective cost to users. For active traders & vaults, this is a huge unlock — lower costs means more profit.

At launch, Perennial has no fees for takers. When markets begin to reach equilibrium (high pool utilization), Perennial will likely introduce small taker fees (but likely closer to that of CEXs). And keep in mind, Perennial has permissionless pool creation, so anyone can create a new LP pool with a different fee structure. In the future, Perennial may expose the full spectrum of the tradeoff between taker fees & slippage, allowing users to customize their experience.

Maker Fees

Fundamentally, makers take the opposite side of takers’ trades, meaning that the same opportunities for market manipulation & toxic flow discussed in the “taker fees” section apply to makers.

Other protocols protect against this toxic flow through a combination of maker fees (or price impact), slippage, and swap fees (discussed later). Maker fees are not ideal, since protocols want to incentivize liquidity.

Again, with Perennial, the delayed settlement mechanism works to combat this issue, allowing maker fees to remain low or nonexistent.

Funding Rate

The funding rate, or funding fees, are fees paid from one side of the market to the other. These are used as a market balancing tool; if market demand is skewed in one direction, the side of the trade with higher demand will pay a tax to the side with weaker demand to put both sides at equilibrium (equal expected values). Unlike transaction fees, which are a one-time fee at the time of the trade, funding fees are ongoing throughout the entire course of the position. Funding fees have been extremely high on DeFi derivative protocols — sometimes over 50% annualized when funding on CEXs is less than 10% annualized. Fees this high put traders in a losing battle and are a non-starter for most structured products & hedging use cases.

Having funding rates this high can be attributed to inefficiencies in mechanism design. There are a few elements to this:

First, funding-rate structure is inefficient. Even at low levels of utilization, on-chain derivatives protocols often raise funding rates significantly (using a linear funding rate curve). Also, these funding rate curves are often uniform across multiple assets, despite varying levels of risk. The result: low capital utilization and/or funding rates that are well above market.

Second, complicated LP structures create friction for makers (those on the other side of the trade), opening up market inefficiencies. Complexity in the form of convoluted risks for LPs makes it harder for capital to directly arbitrage the rate to the market rate. In many no-price-impact derivatives protocols, the funding rate structure is actually more of a borrow fee from a passive LP pool, rather than a true funding rate exchanged between two sides of the market. Passive LPs will require a higher rate (more compensation for risks) than other market participants.

And last, capital inefficiency increases the cost of capital that protocols must compensate for on an ongoing basis.

In Perennial, each market is tailored to a specific payoff, so rates match the risk. Jumprate utilization curves maximize utilization without ratcheting up funding rates prematurely. Markets are also isolated, allowing market participants to be efficiently matched, creating a structure with a true funding rate between makers and takers. This, along with high in-protocol maker leverage makes it easy to arbitrage rates on Perennial.

Swap Fees

Swap fees narrowly refers to a fee users are required to pay as part of an in-protocol asset conversion, often used to abstract some functionality of the protocol. But we could broaden this bucket to include all non-standard fees that are tucked into DeFi derivatives protocols. These fees are not related to the core protocol and only serve peripheral functionality (things that arguably shouldn’t be baked into the base protocol).

For non-synthetic or non-cash-settled protocols (as is the case with many in DeFi derivatives), to create the same UX as a fully-synthetic & cash-settled derivatives protocol (how traditional markets operate), there may be functionality of the system that is abstracted from the user, leading to additional fees. For example, some protocols utilize spot assets in the maker pool & as collateral to partially hedge the exposure. Doing so may require makers & takers to convert their asset into one/more other assets to open a trade or LP position. This sometimes leads to takers paying more in swap fees than trading fees, and makers paying 20–40 bps to LP into the protocol (and another 20–40bps when withdrawing)!

Perennial is fully synthetic and cash settled — no swapping or additional fees required.

Capital Efficiency (capital cost)

On-chain Derivatives protocols require capital to operate, and capital is not free; it has opportunity cost & requires some rate of return. Thus, if we increase capital efficiency, we reduce the cost of capital, meaning users will be willing to accept a lower rate of return.

The entire fee structure of derivatives protocols can be reduced by simply creating a more capital efficient system. And the inverse is also true — compromising on capital efficiency necessarily leads to higher costs within the protocol. There’s no way around this.

Perennial’s core protocol is designed for maximum capital efficiency — LPs can use 20x+ leverage, just like takers.

Conclusion

Users may put up with high fees in the short-run, but over time, fees will have to come down or DeFi derivatives will hit scaling limits. Careful mechanism design & strong capital efficiency are the key to keeping costs low on DeFi derivatives protocols. Perennial has placed a big focus on this, and the results are clear.

At Perennial, we’re actively improving the core protocol mechanism and building extensions that enable new vaults & experiences on top. We’d love to hear your feedback.

Check out our job board — we’re hiring!
Follow us on twitter.
Come join our discord.