Having gone through each element used to determine the funding rate of perpetual futures, it’s time to see how current exchanges have implemented them.
Order Book Exchanges
There is not much difference between the different exchange implementations. The most important factor, the interest rate component is identical amongst exchanges that have even bothered to include it.
OKX and Deribit will likely have a higher bias in average price divergence (perp price less index price) due to the lack of an interest rate term. All the other exchanges use the same 0.01% per 8 hours in all markets. This is an area that could do with some optimisation to tailor the rates to the specific market.
OKX and Deribit also do not use impact prices when calculating the perpetual market price, rather relying on the last traded price. This has the potential for manipulation and increases risks.
The other major differences are the impact notionals and the maximum funding rates. There has been a gradual trend towards larger impact notionals over time to mitigate manipulation . This could further be improved by calibrating it to average orderbook liquidity rather than the initial margin fraction.
See below for a summary of the major parameters in some major exchange funding rate implementations.
|Interest Rate Component||0.000125% per hour, 0.01% (every 8 hours) equivalent to 10.95% annually||Individual to markets. Currently set to 0.01% per 8 hours for all||0.01% (every 8 hours) equivalent to 10.95% annually||0||0||Individual to markets. Currently set to 0.01% per 8 hours for all||0.01% (every 8 hours) equivalent to 10.95% annually|
|Funding Period (hours)||1||8||8||8||8||8||8|
|Sampling||1 minute||1 minute||5 seconds||1 minute||1 millisecond||1 minute||1 minute|
|Impact Notional||$500 / initialMarginFraction||$10,000||$200 / initialMarginFraction||n/a||Individual to markets||$10,000|
|Maximum Funding Rate||4 380% Ann (4% per 8 hours)||75% of initialMarginFraction per 8 hours||75% of initialMarginFraction per 8 hours||Dependant on market, 0.375 on BTC – 3% on DOGE||0.5% for BTC (547.5% Ann), 1% for ETH (1095% Ann)||75% of initialMarginFraction per 8 hours||75% of initialMarginFraction per 8 hours|
|Differences from the norm||No CLAMP||No impact prices, Linear function of premium – no CLAMP||No impact prices|
Liquidity pool exchanges
The other category of perpetual futures exchanges are the decentralised exchanges using a liquidity pool as counterparty to traders positions.
Exchanges of this type do not need to incentivise price convergence as prices are sourced from an external oracle. They rather need to incentivise a balanced book in terms of open interest. For this reason they do not need to stick to a funding rate that mathematically guarantees the payoff of holding the underlying can be replicated over some interval. They are free to incentivise trades than reduce net open interest in any way they like.
Technically most of these liquidity pool exchanges actually offer margin trading products where assets are borrowed from a liquidity pool to achieve leverage. These liquidity pool exchanges do not have a funding rate in the sense that payments happen from the dominant side of the market to the smaller side. Funding rates on these exchanges are called borrow rates as they are the rate the assets are borrowed by users for their positions. These borrow rates fluctuate according to pool usage of the asset in question. From a user perspective they affect their P&L in a similar way though. This borrow rate is always positive though and there is never a borrow rate benefit.
Synthetix perps v2 is the exception, offering completely synthetic derivatives exposed to a central liquidity pool. It adjusts the funding rate each hour depending on the market skew to keep the liquidity pool exposure balanced. An analogy would be market makers adjusting their prices to balance their inventory.
The funding rate on Synthetix increases if the market is long, and decreases it if the market is short. The rate the funding change is equal to the skew times a constant set by governance.
This makes Synthetix funding rate effectively set by market forces and it could converge to any level in equilibrium. It also creates an interesting venue for arbitrageurs to trade against limit order book exchanges due to the differentiated funding rate formation process.
Orderbook exchanges have mostly adopted the original BitMex funding rate design without much iteration. Some even leave out certain aspects of the implementation. The most glaring aspect to be optimised at this stage is the interest rate component which should be calibrated to individual market conditions. There is also scope to improve the impact price notional to ensure the most accurate prices are used at all times.
Liquidity pool exchanges mostly charge a borrow rate rather than a funding rate. This is always positive in that all positions pay an hourly rate to keep their positions open.
Synthetix is different with a funding rate of their own, set in a completely different way to order book exchanges. This could present interesting arbitrage properties.