When apps become chains.
A small note before we begin. Nothing written in this piece is financial advice. I am just nerding out some elementary napkin calculations and commenting on what I think. I would appreciate inputs on the model used if you are a fellow finance enthusiast.
My interest in finance keeps me looking for news from DeFi projects. On October 24, dYdX announced that their appchain was live. In the following days, dYdX proposed enhanced utility of the DYDX token.
What changes with the launch of the chain? The token (DYDX) gets three utilities instead of one:
Governance – DYDX is already the dYdX application's governance token. It now expands to be the governance token for the chain too.
Gas token – All the gas fees paid on the new chain will be in the DYDX token.
Security – As a POS-based chain, DYDX will be staked to secure the chain. As a result, stakers will get staking yield. The upside here compared to other staking is that even the fees accrued by the dYdX application (in USDC) will be distributed to stakers (and validators).
It is meaningful because, in the last 30 days, the dYdX exchange has generated over $6 million in fees. The figure is at $65 million YTD. The fees will be distributed to token holders. But how does this differ from an exchange like Uniswap?
Unlike most DEXs, dYdX is an order-book-based exchange. It matches order books off-chain, whereas most DEXs are automated-market-maker-based (AMM-based). A critical feature of providing liquidity to AMM pools is that the liquidity provider (LP) invariably ends up holding more inventory of the token that devalues compared to the other token in the pool.
Let me explain.
AMMs are typically based on the A*B = constant model, meaning the product of values (price times quantity) of the two tokens in a pool remains constant.
Say an LP adds $100 worth of tokens X and Y in a pool with a total liquidity of $1000 (after the LP adds), a 10% share of the pool. Assume the price of X is $1, and Y is $5, and the LP added 50 X and 10 Y. The pool has 500 X and 100 Y.
Product of values of X and Y = ($1 * 500) * ($5 * 100) = 250000.
Each Y is worth 5 Xs.
After a few trades, assume the pool now has 1000 X and 50Y tokens. The prices are now $0.5 and $10, respectively, i.e., each Y is now worth 10 Xs. Notice the product is still ($0.5 * 1000) * ($10*50) = 250000. But the LP now has 100 X tokens (10% of 1000) and 5 Y tokens (10% of 50) with a total value of $100 instead of 50 X and 10 Y tokens (value = $125). So, the LP has lost $25 by adding liquidity to the pool.
In traditional markets, where exchanges are based on order books and derivatives markets are mature, market makers have much more flexibility around hedging. In the AMM design, hedging is constrained. As a result, providing liquidity on AMM-based exchanges is more difficult (and, thus, risky) compared to orderbook-based exchanges.
Naturally, AMM-based projects must offer more incentives to LPs. Without liquidity, there are no traders. And there are no fees or revenue without traders. So, exchanges end up sharing a significant chunk of fees with LPs.
On the other hand, since MMs on order books have more freedom to hedge their inventory, they don’t need as many incentives from the exchange. Their incentive lies in earning the spread between the bid and ask prices. For similar reasons, we saw Hubble Exchange move from AMM-based to orderbook-based designs. Often, exchanges compensate market makers via rebates to incentivise deeper liquidity.
Don’t get me wrong: I’m not saying that the order books are better than AMMs in an absolute sense. Orderbooks require active participation and an off-chain component. AMMs are fully on-chain (so they are more transparent) and are perhaps a better choice for LPs who aren’t as active as traditional market makers.
The easiest way to see how DEX peers have evolved over the years is to see how much trading volume they supported and the fees they earned. Notice how spot exchanges (like Uniswap and Pancakeswap) dominated. Read this piece if you want to read more about DeFi derivatives.
The order book design allows dYdX to share all the fees with token holders and validators. This is why $75 million annual fees earned by dYdX differ from those earned by an AMM exchange. While dYdX earns fee revenue, it also spends DYDX tokens as incentives, as almost all applications do. And how does that impact applications?
The chart below compares how much token incentives exchanges have given out compared to the revenue they earned: that is, revenue generated for every $1 spent on incentives.
Every exchange needs to incentivise liquidity in some manner. Some do it with token incentives, and others do it via sharing fees with LPs.
In the chart above, ~180% for Perpetual Protocol in 2022 suggests that for every $1 they spent on incentives, token holders received $1.8 in revenues. After the recent overhaul of Synthetix, more volume started flowing through it, and it started earning more fees. As a result, for every $1 on incentives, revenue increased from ~$0.2 in 2022 to $0.8 in 2023.
GMX has a low ratio because most of its fees are shared with LPs instead of token holders. So, for GMX to give value to its token holders, either the fees need to increase a lot, or the proportion of fees shared with token holders needs to grow.
Before delving into the valuation details, I must emphasise that all valuation models are inherently flawed. Some are more flawed than others, and this one is probably more flawed. However, I resort to this exercise because it helps me think through different projects and compare the ones belonging to the same sector using the same metrics.
Keep reading with a 7-day free trial
Subscribe to Decentralised.co to keep reading this post and get 7 days of free access to the full post archives.