Where does the yield come from? If you can’t answer this question, you are the yield.

Author: Nabil Abdellaoui, Data Science Lead (@randombishop)

Have you ever wondered how real these DeFi liquidity pools are that advertise juicy and delicious APYs? 50%, 150%, even 400% for providing liquidity into a trading pool? Is it an opportunity or a poisoned gift?

In this article, we will share how we answer this question and how we make decisions for Concrete capital allocation, with a case study on Trader Joe V2.1 liquidity pools.

Overview

In a previous article, we presented how we use our simulation engine (The Matrix) to forecast the outcomes of the Concrete Earn and Borrow flows.

In this article, we will zoom in on a particular aspect and simulate different strategies targeting a high yield pool on Trader Joe.

Trader Joe proposes some extremely attractive APYs here, with multiple pairs advertising +200%; but to focus on the most liquid pair and understand the mechanics of these pools while minimizing token specific aspects, we decided to zoom in on the ETH-USDT pool.

According to a DefiLlama dashboard, this pool APY has been consistently hovering between 100% to 200%, with daily peaks over 500%, for almost a year. Which begs the question: why isn’t its TVL higher? ETH and USDT are blue chip tokens, Trader Joe has been around for quite some time, so no major smart contract risk should be scaring the market, isn’t it an easy opportunity?

Well, if it looks too good to be true, it’s too good to be true.

Liquidity Provision on Trader Joe

Trader Joe provides an interesting approach to deploy liquidity, well documented here in terms of capabilities and how-to guides; but it doesn’t explain the financial aspects of the different proposed provision strategies, not-financial-advice style, leaving it to the user to DYOR and figure out what works for them.

So, we did some research.

First, what is the APY advertised on their website? It turns out it’s calculated from the total fees generated by the pool divided by the total value locked into the pool. Fair enough, that’s real revenue.

But it’s not evenly distributed to all the liquidity providers, it’s rather distributed to the users who deployed liquidity into the active price range of ETH. In other words, the more concentrated an LP position is, the more fees it will collect. We actually simulated, and also tested with real positions, how the real APY relates to the advertised APY on a given day; and it turns out this is what you can get:

• If your liquidity is very concentrated around the current price, [-1% +1%], and remains active all day, you can collect way more than the advertised APY, sometimes over 2x.

• If your liquidity is concentrated enough around the current price, [-4%  +4%] seems to be the magic range, your yield will be similar to the advertised APY, more or less.

• The wider your price range, the smaller your yield. For example, [-20%  +20%] will get you 5x less than the advertised APY.

Now, the yield is real, ETH has a very solid probability of staying in a +/- 4% range over one day, why not be greedy and eat the +150% APY every day (that’s around 0.25% daily return)?

Well, there’s a catch, actually two of them, and they’re called impermanent loss and rebalancing cost. If your plan is to exit the position if the price of ETH goes out of the [-4%  4%] range, and you’re OK with keeping only USDT if price goes up, and ETH only if price goes down, that’s an excellent deal, you basically defined 2 exits that are acceptable for you, you positioned yourself as a market maker, and you collect 0.25% per day until your exit.

But, if you were hoping to make +150% in a year, this is not going to work. Given current volatility, ETH would move up or down by over 4% several times per month. Therefore, you would need to rebalance your position and re-deploy it into a new range. And every time you do that: (a) you pay some trading fees; (b) you materialize the so-called impermanent loss. You basically make the impermanent loss permanent. These 2 components will instantly cost you anywhere between 0.5% to 1.5%.

The Experiment

We simulated 1000 scenarios with different price trajectories to visualize how this can play out; and the chart below shows the expected delta between simply holding 50% USDT and 50% ETH; vs deploying to the Trader Joe pool into a [-4% +4%] range and rebalancing every time the price exits the range:

Each decile represents an outcome with 10% probability. The 4 red bars on the left will happen when there’s more than 7 rebalancings over the course of the month, while the green bars on the right happen when the price is more stable. In the worst scenarios you could lose -2.5% in a month; and in best case scenarios make +3%; for a total average expectation across all scenarios of +0.4%.

It’s still a net positive expectation game, but very far from the advertised +150% APY.

Expectedly, if you deploy into the wider price range [-20% +20%], then you will get a lower but more consistent return:

In this case, you have 60% chances of price staying in range for one month, and won’t need to pay the impermanent loss and rebalancing cost. The overall average monthly return expectation is 0.5%.

In Summary

• Trader Joe advertised APYs can’t be obtained by simply depositing and maintaining the position over one month; they are only attainable under opinionated strategies where the liquidity provider bets that the price will remain in a narrow range for a few days.

• In a basic deposit-and-rebalance strategy, over one month, the earned fees will be offset by the impermanent loss and rebalancing costs. The expected yearly return for a pool advertising 150% APY would actually fall under 5%, with some 20% to 40% risk of being negative.

Learnings

The main takeaway from the example presented is that the APYs advertised by DeFi pools have to be deeply analyzed and simulated to understand the end-to-end risks and rewards of providing liquidity into them. While DefiLlama does a good job at calculating the actual revenue and dividing it by TVL to output an APY number, this number doesn’t reflect the actual $ denominated profit if you start from a stable denominated vault and exit back to it. It also ignores the fact that you probably want to enter and exit after a certain amount of time: they convert a daily yield into a yearly rate with a simple compounding formula, ignoring the fact that maintaining a position for any period longer than one week has risks, costs and huge variation in the income stream.

You can easily see this by comparing TVL and advertised yields for very similar pools: why does the ETH-USDT Uniswap pool have 10 times more liquidity than Trader Joe’s counterpart, while offering a smaller APY?

Also, while browsing high yield pools and running simulations, we also realize that the DeFi market is evolving towards more maturity, as true arbitrage or yield is becoming harder to find, even outside Ethereum Mainnet.

To illustrate with another example: USDT-AERO is one of the top Base pools on DefiLlama, with a TVL of $100m and +100% APY, and less sensitive to impermanent loss and rebalancing cost than the Trader Joe V2.1 pools. This one, again, smells like a very juicy steak. Of course, it would expose us to the price of AERO, but what if we built a strategy to short it at the same time? A cash-and-carry strategy where you deploy half a vault into that pool, and use the other half to short AERO using a lending protocol, would basically yield half of that APY, so +50%, still very attractive, while canceling the exposure risk to AERO. But here’s the catch, to short AERO on-chain, the funding rate is currently sitting between 50% and 100%! Long story short (pun intended), there is no free lunch in this one either.

While we concluded that these pools are not suitable for long term positions, we also discovered the single-side provision mode, which will be used by Concrete as a trading strategy to prepare collateral for liquidation protection. This essentially positions our protocol as a market maker instead of a taker when we know we will need to buy collateral at a certain price, thus earning fees instead of paying a trading cost.

As DeFi matures, and real opportunities become harder to find and arbitrage, researching and thoroughly studying the yield mechanics, risks and outcomes, is key to our work at Concrete and driving competitive returns

If you are interested in chatting about protocol design, working together, or integrating within the Concrete ecosystem, please reach out to hello@concrete.xyz

Follow along on Twitter and join the community

Concrete Protocol: https://twitter.com/ConcreteXYZ

Blueprint Finance: https://twitter.com/Blueprint_DeFi