And here is part four of the Chainflip crypto-economic series with me, Simon. This time, rather than shilling our unique AMM design, we are diving into real cryptoeconomics today. We’re going to be throwing around words like “emission” “inflation” “incentives” “game theory” “return” and other wild and wonderful economic terminology that is sure to tickle your adrenal glands.
So here we are, part 4 of 5 parts. The final part terrifies me, but here’s what we’ve covered so far:
- Part 1: Uncapped Supply
- Part 2: Validator Auction Theory
- Part 3: Swapping Logic
- Part 4: Incentive Design (you’re reading it)
- Part 5: Economic Security (soon)
Let’s get stuck into some made up numbers.
Almost every token project under the sun has some sort of incentive design. It’s one of the key features of programmable money that has made crypto interesting at all. Every single token released today is an individual experiment in the nascent, made-up, and barely recognized multidisciplinary field of Tokenomics/Cryptoeconomics.
I make no claim that FLIP is somehow the pinnacle of token design. There is probably no such thing. The objectives of each protocol differ substantially, and external factors unrelated to the token economics can significantly skew the perceived results of any given token design. Accurately measuring success is next to impossible. All we can do is make educated guesses based on what we would like to achieve in any given token powered protocol. After working on and proposing several adjustments to the cryptoeconomics of the Oxen blockchain over the years, I’ve been fortunate to have some real world experience with this before. What I’m presenting today is an evolved plan of what I came up with for Chainflip last year. I hope by the end of this article you will at least see the thinking that's gone into the Chainflip token design and can understand it a little better. Better yet, if you can help me and tear this thinking to shreds, I’m all ears.
Looking at Chainflip, there are three main objectives I had in mind when designing the cryptoeconomics:
- FLIP holders should be natively rewarded just for holding the token. The protocol should operate in such a way where the token can capture value and sustain itself through greater utilization of the protocol.
- The FLIP token needs to strike a balance where it rewards validators for securing the network so that we can safely store as much liquidity as possible in the network without triggering excessive inflation.
- Liquidity rewards should be designed to attract “sticky” liquidity. In other words, when liquidity rewards are turned off, all of the liquidity doesn’t just vanish: it sticks around.
Overall, I perceive this to be a balanced set of objectives. A holy trinity of sorts. My hope is that Chainflip as a protocol not only survives, but thrives for over a decade as a general concept. As the market evolves, we hope to as well, so our token design has to follow this evolution by maintaining a balanced and sustainable design with long term upside potential for all holders, including the labs.
The crypto spot trading market has always been the single largest source of revenue in the industry, fragmented across thousands of markets. We do not touch fiat, and within individual blockchain ecosystems, native on-chain experiences like Uniswap exist. We hope to capture a sizable chunk of the spot market by covering the needs of the majority of users moving between and within fragmented ecosystems on-chain.
Objective 1: Achieving Value Capture
The business model for the protocol is therefore pretty simple: capture fees from spot trading. I discussed the buy and burning model at length in Part 1 of this series, which you can read here, but for now I’ll quickly summarise how it works:
- A 0.1% Network fee is deducted from each trade and collected in USDC
- The USDC is used to buy FLIP tokens in the Chainflip AMM
- The tokens are then removed from circulation by the validators
To spice up this analysis, I’ve assembled this model to help illustrate how this burning scheme will impact the fundamental floor price of FLIP. The Y axis shows a range of daily volumes achieved across all Chainflip AMM pairs with a 0.1% fee, and the X axis represents the price of FLIP. The resulting chart shows the percentage of the FLIP supply that would be bought and burned annually at each of these levels.
In my estimation, a high percentage value would imply the FLIP price is too low and the fees from the volume alone would cause the price of FLIP to sharply rise to a stable level. Low percentages suggest that the buying and burning of FLIP alone would not sustain this price. I arbitrarily estimate that it is around the 5-6% level that the price would not fall below this level before the buying and burning offsets any natural churn from emissions being sold on the market. Given we expect most validators to retain their rewards to stay in the set (demonstrated later in this article), I assert that 5-6% is a reasonable estimate for realised inflation year to year. Fight me.
This price chart leaves absolutely no room for speculative premium or premiums associated with competition for validator slots. It also does not factor in any future protocol fee changes, or any other fees already in the protocol. This is a purely fundamental analysis of a single aspect of the Chainflip protocol token-based entirely on raw protocol utilisation.
I strongly believe Chainflip will be able to achieve volumes higher than most DEXes for a number of reasons, the primary being that the biggest trading pairs in crypto are not ERC20-ERC20 token swaps. Binance’s BTC/USDT pair alone has the same volume as all of the pairs on UniV2 and UniV3 combined. The size of the spot market that Chainflip plays in is just way bigger. Solana, BNB, BTC, ETH - these 4 assets alone, all on separate blockchains, vastly outstrip existing DeFi volumes 8-fold. The cross-chain spot exchange pie is BIG.
Using this same methodology, you can extrapolate this model to figure out what the price of FLIP would be if the protocol achieved an average of $100bn daily average volume - a territory only Binance currently plays in and no decentralised swapping protocol has yet achieved. Under these conditions, the burn rate would be around 5% at about the $3000 per FLIP mark. Again, this leaves no room for other premiums or net inflows and is based purely on fee revenue, but personally, I see the maximum size of the market for Chainflip being more like 20-30% of all cross-chain spot trading, but it’s fun to think about nonetheless.
In my mind, this is a great model to capture value for token holders, and the fundamentals point to a considerable upside even at achievable volume levels considering the cross-chain spot volume far exceeds that of the ERC20 token markets. That's the first design element checked off the list.
Objective #2 Sensible Validator Reward Levels
Chainflip’s vaults, which store all assets across the external chains, are more lucrative targets for attackers and require stronger security guarantees from the validator network than most typical blockchain networks. Each of the validators holds a secret which, if used in conjunction with at least 99 other validator’s secrets, can be used to wrongly take liquid user funds from all of the active Chainflip vaults.
In order to dissuade validators from colluding to drain the vault funds maliciously, we have to design an economic security scheme that addresses this risk. The main method by which we combat this risk is with the threat of slashing the stake of malicious validators. If you do a bad thing, access to your precious FLIP tokens will be rendered non-existent, creating a material punishment for malicious behaviour.
And thus, the more FLIP that is staked across validators, the more liquidity the Chainflip protocol can secure in the vaults that they control. The exact relationship between liquidity security and stake took me a 13-page paper to explore and will make up Part 5 of this series. For now, all we need to know is the more FLIP is staked, the better.
More FLIP staked = more liquidity capacity = more volume = more fees = happy days.
As you might expect, the way to get FLIP staked in the protocol is to incentivise validator operators to do it. Typically, Proof of Stake networks achieve this by printing block rewards, and Chainflip is no different. Our approach is to set a fixed annualised emission rate, distributed evenly between the validator set, but Chainflip validators do not earn transaction or protocol fees. Instead, the fees are all burned, evenly distributing the value of these fees across all protocol participants.
Without additional fees on top of the block reward, the validator incomes are quite predictable. The number of tokens paid to each validator should only change slowly over time unless the protocol emission parameters are adjusted during an upgrade.
Given the number of tokens given to each validator is easy to predict, really what we are doing now is estimating how many FLIP validators will stake in order to win that reward. If I know that a given validator slot is going to yield approximately 30,000 FLIP over the course of the next year, the real question is how many FLIP tokens would I be willing to stake in order to win it? 50,000? 100,000? 200,000? 1m?
The definitive market-derived answer to that question won’t come until after launch. However, we can make some predictions based on how much validator operators are willing to stake in other protocols for a given reward. This number, represented as a rate of return or APY/APR, can be called the Staking Equilibrium. It is the reward APY whereas a whole, validators are staking and unstaking in equal amounts. If the APY falls below the Staking Equilibrium, we would expect to see validators unstake until the rewards go back up, and similarly, if the APY is above it, we would expect to see more people stake into validators to get the rewards until it reaches the equilibrium APY.
As protocols mature, it’s likely that the Equilibrium goes down over time, but we can take a look at some other protocols and their reward levels to assess their Staking Equilibriums.
This is a rough assessment, but the TL:DR is that I think saying that a normal equilibrium value for validator operation is about 15% after 12 months of operation. These projects were chosen to show what mature projects with mostly very complex validator software are getting in terms of lockup and how much they are paying validators.
There are two factors that separate Chainflip from the pack. Firstly, it is that almost all of the above protocols allow some form of staking delegation. Chainflip does not, because it introduces the nothing-at-stake problem which is particularly serious for liquidity security in protocols like Chainflip and THORChain, which is why we don’t do it. You stake your own capital, not someone else's, and risk losing it if you play badly. This will likely cause our staking equilibrium to sit a bit higher than other protocols, as there are fewer people who actually can stake in validators.
Secondly is that Chainflip’s validator software aims to be considerably less painful to operate through its baremetal-friendly design. Grace periods are also built in to encourage a stable and mature validator set that can maximise protocol resilience. There’s nothing like getting your tokens aggressively nuked because you didn’t check your alerts for 2 hours. We want people to stake more, and not less, and so by providing a validator software package that is both highly configurable and straightforward, and by allowing some slack on the part of individual validators through a positive reputation system, we expect this will lead to more staking, or more accurately, a lower Staking Equilibrium, over time.
So, what does all this mean? If we take these points of difference on balance, and assume 15% to be a good starting number for an estimated Staking Equilibrium, then we can show on a table how many tokens would get staked and what the minimum active bids would be for a given lockup ratio:
This equilibrium table demonstrates that in order to attract a good chunk of FLIP into the validator network, we don’t actually need to go too crazy on overall emissions.
If we set the initial validator rewards rate at 5%, we will likely see the validator stakes rise to about a 35% lockup ratio of all fully diluted tokens, or roughly 32 million FLIP tokens. For Sandstorm, this is perfectly adequate.
When we get to the later releases that start to require the validator network to secure real liquidity, in particular the Berghain release, I assess that 7% would be the best target emission to start with. At this rate, we would expect to see closer to a 50% lockup ratio, or around 45 million FLIP tokens in just validators. In addition to this, FLIP will be locked in backup validators and bond holders too, along with the remaining vesting schedule.
The reason why I don’t think any higher makes sense is because of the size of the minimum active bids beyond these levels. To achieve a 50% lockup ratio, we are already talking about a minimum active bid of 300,000 FLIP - a number already out of reach of many participants where no delegation is available. Therefore, I believe that if we further increased the emissions, we would see an increase in rewards but perhaps not as significant as an increase of actual tokens staked. In the future, we may have secondary validator sets in Chainflip to support a greater number of assets simultaneously, which would lower the overall staking requirement per node, and thus higher lockup ratios would be possible with reward increases. However, this is not a near term requirement, and for the time being, increasing emissions beyond a few % points makes little sense.
Long story short, here’s what I propose:
- Set the Validator rewards at 5% at the Sandstorm launch
- Increase the rewards to 6% at the Ibiza release
- Increase them again to 7% at the Berghain release
- Review the rewards scheme again after 6 months, with amendments possible through validator-approved chain upgrades
And that, I hope, is the answer to Objective number 2.
Objective #3 - Sticky Liquidity
In the summer of DeFi, yield farming took off. It proved once and for all that, if you print enough tokens, you can make almost anything extremely liquid. It also proved that as soon as you stop printing those tokens, that liquidity disappears as quickly as it arrived. A lot of projects last year realised that making something liquid alone doesn’t make the price go up - at least not for very long. Eventually, liquidity rewards catch up on you.
This begs the question - how do you reward liquidity providers? They are essential to any DeFi protocol. Without liquidity, Chainflip or any other DEX becomes worthless. At the same time, these programs are also very expensive in the sense that the majority of liquidity providers dispose of their token rewards very quickly, as they know that everyone else receiving the rewards will also want to sell them. It’s like a game of liquidity chicken. How can you both encourage liquidity providers but also ensure the tokens used are spent efficiently, yielding a net positive for the protocol overall?
For ideas, I think we can look at other protocols.
Uniswap’s LP tokens facilitated the phenomenon of yield farming initially, but Uniswap itself also had a moment where it felt compelled to offer liquidity incentives. This was a direct result of Sushiswap’s initial ‘vampire attack’ which aimed to pull liquidity away from Uniswap by offering a token incentive. Up until this point, Uniswap’s pools had not been incentivized by Uniswap itself. People were putting liquidity into these pools for one of two reasons:
- Because a separate protocol team was offering incentives for specific pairs (yield farming offerings on smaller coins), or;
- Because the LP genuinely believed they would be able to earn sizable fees from the pool.
With Univ2 style liquidity pools, impermanent loss is unavoidable without very complex and costly hedging strategies. Despite this, there had been enough demand from users in the major liquidity pools that the 30 bps fee offered to LPs was, for a long time, worth the risk. After the incentives ended, UniV3 came out and threw the AMM rulebook out of the window. Now, for all of the major pairs, LPing became an easy way to make money without any token incentives anywhere in sight.
And that hidden there, I believe, is the key to sticky liquidity. The real solution to this problem is not to offer incentives to LPs, but to create an economic structure where there is enough money to be made by LPing that the protocol doesn’t ever have to directly reward the providers.
This looks very different for different types of cryptocurrency markets. The ETH/USD market is structurally different from a smallcap ERC20 token/USD pool. UniV2 is still used as the primary liquidity pool type by many smaller tokens on the market - and justifiably so. The LP token system works well to incentivise a minimum pool of liquidity for the token to boost depth and volume across all of the pairs for that token. LPs are rewarded in tokens for taking on the impermanent loss risk, but that risk-taking is actually what keeps that token’s markets alive and liquid.
For the major pairs, this is not required as there is just so much liquidity out there. All sorts of risk mitigation strategies become possible. You can sell spot on one exchange and simultaneously go long on a derivatives platform, you can arbitrage large trades, you can do a million things that are just too risky or simply impossible for tokens with smaller markets.
Chainflip happens to fit into the latter category, where we will pretty much exclusively be supporting the major market pairs. Looking at this category, we can see that by using UniV3 style pools, LPs can find ways to make excellent profits with very low risk. Token incentives are simply not required to sufficiently reward LPs. Chainflip shares this approach and extends it. The dynamic range order system in Chainflip means that LPs have even more control over their liquidity than in UniV3 and as such can execute more traditional market-making and arbitrage strategies to bridge liquidity from other markets and make significant profits on every trade.
Trouble is, there’s a bootstrapping problem.
When Chainflip launches, there needs to be liquidity for the protocol to attract users and therefore volume. With no volume, there’s little incentive to put your liquidity on Chainflip.
So to get around this Catch-22, Chainflip needs to incentivise liquidity. We have to create a scheme that means LPs will have the incentive to set up their trading software in the first place and start competing with each other on the AMM, even before there are enough users to keep those LPs around from user-funded LP fees alone.
Despite already having some minimum liquidity commitments from strategic partners, the Chainflip protocol can provide a more generic rewards scheme that achieves the above-stated goal.
If we just followed the usual yield farming scheme and gave LPs tokens for putting money in the pool, we would attract a bunch of money that is not competing for user trades and does not follow the intended behaviour of liquidity providers in the system. It would just sit there, mostly unused, and as soon as the incentives stop, that money vanishes. That’s not sticky.
If instead, we paid Liquidity Providers tokens based on how much LP fees they earned from the protocol, that’s a different story. We would be teaching Liquidity Providers how to compete using the Chainflip protocol by providing strong incentives to maximise the efficiency of the liquidity they choose to deploy. LPs would quickly learn how to front-run each other to capture those fees, and how to remove risk from the trade through external management strategies. LPs have to be active in order to earn fees, otherwise, those fees will just be taken by active participants and therefore get all of the token rewards taken too. This incentive scheme might even give early LPs the incentive to accept greater risk in order to guarantee they get the tokens.
This scheme would mean that real users are benefiting by getting access to the best (maybe even better than index) pricing straight away, even before liquidity has really had time to build up in the system, meaning they’d be more likely to come back and create more volume again later on.
Over time, the volume generated through swaps should increase as more and more users and integrations come to the protocol to access this pricing and convenience. Gradually, as that volume picks up, so too do the LP fees, and at some stage, no additional token incentives will be needed to keep LPs around. 25bps LP fees in each pool is more than enough of a spread to entice major market makers and will mean that as soon as we attain decent volume, it’s virtually guaranteed that as long as people want to trade through Chainflip, deep liquidity will always be there. Now that is sticky.
Although this scheme would probably be implemented manually at first, this scaling reward scheme can be added to the emission schedule of FLIP such that a 1%-2% budget is introduced to fund this token subsidy during periods where the volume on the protocol drops below a certain threshold (for example, $100m USD Daily). This would ensure that there is a baked-in mechanism to keep the protocol liquid over time, whilst not printing tokens unnecessarily during times when the protocol is sufficiently liquid.
Chainflip will pay LPs a share of a token reward pool based on the amount of LP fees they earned over a window of time (probably 24 hours). These rewards will be scaled back as volume increases.
Objective #3, I think we may have a solution - let’s hope it works.
In this article, I attempted to address the final pieces of the puzzle for Chainflip’s incentive scheme. The final details of these schemes will be detailed in a simple breakdown which will include emission rates for all reward types and the rules that can modify them. In fact, I’m hoping to vastly simplify the Cryptoeconomics series into a single paper at the end of all this. I will of course try and build some sort of one-pager to go on the website so no one has to read these rambling justifications for too long if they don’t want to.
The purpose of this series is to document a set of arguments for the starting parameters of the Chainflip token’s economics, not just as a way to promote the protocol, but as a future reference for further debates, discussions, and analysis around FLIP. Once we hit the big green button, I fully expect that they will arise.
Until then, stay tuned for the fifth and final part of the series - the dreaded 13 page long Economic Security article. Oh god.