The Mechanics of DeFi yield farms and their eventual blow out.

Before I go about showing the DeFi emperor naked, I first want to say that I am materially invested in many DeFi protocols. There is enormous potential in great execution of on chain finance and what I say here today is not to diminish the outstanding projects, but rather comment on how the market has cleverly created a new bubble mechanism and its mechanics.

First, some key assumptions that I have observed while trading and participating in the yield farming craze.

1. Most trading volume on chain is centered around yield farms and the allocation of LP tokens to capture token emissions from other projects. Generally AMM (X*Y=K) with two uncorrelated assets will lose to impermanent loss. The offsetting function here is payment via farm token distributions.
2. Because almost all pools don’t actually make money via only trade fees and market participants are generating their offsetting returns via token emissions, it logically follows that most TVL (Total value locked) is generally only done so to participate in these games.
3. X*Y=K AMM facilitates farming by serving up a buyer at all prices. Once farms secure an incentivized pool with decent liquidity, they have effectively created a buyer at all times. Unlike in order book markets where Market Makers place orders predicatively with assumptions of the future, AMM’s cannot do so. So when there is a farm that essentially guarantees large amounts of selling via staked token schedule, the AMM will serve up a buyer who is agnostic of that information. Where an order book would discount the price, the AMM does not.

The Structure of Modern Farming

Example of the Primary liquidity pool used in standard farming structures. BAO Token

Here is the structure of the modern yield farms. You will see this copied all throughout DeFi and masked as a supposed “fare launch” methodology. In fact there are projects such as BAO, that are ONLY farms, and only contain a promise of a product deep into the future, and yet this structure allows them to maintain a market cap unthinkable in any other market. Our hypothetical project below is structured from a token perspective as follows — we will call it TURD.

1. The Airdrop or first launch — Designers of the game need to first distribute the initial set of tokens, the intent here is for them to find their way into an AMM pool to provide liquidity for the next wave of “investors”. Some projects airdrop tokens to users of other protocols, such as EPS did with vCurve holders. Some projects elect to simply allocate a portion of the token supply directly to an AMM pool. Other projects sell tokens to insiders or early supporters. There is no better way to obtain network effects that make people paper rich fast.
2. Now the creators need to decide on a emission schedule and tokenomics setup. Generally the TURD/ETH pool will receive the largest emissions. Because the TURD/ETH pool now gives the project a price, these emissions on paper create the illusion of a very attractive return. Its not uncommon for these projects to start out with APY figures like 20000%. Investors see this on a number of farming stats aggregators or twitter and flock to buy the token, and stake TURD/ETH LP tokens in the farm. I call this stage one of the farm. Because of the initial buying, price of the token increases (remember the pool was only recently created/seeded, so its easy to move the price). Stage one sees a large price increase, this price increase in the TURD token in turn then maintains or increases the stated APY figures of the TURD/ETH pool farm.
3. Accessory farms are created to stake other things, some times single sided liquidity like a USDC pool. There is many forms of these sub farms which are rewarded in token emissions, they will generally always pay less than the LP pool for TURD/ETH. These sub farms allow people who rightfully don’t want price exposure to TURD, but also still want to participate in the token distribution to claim some yield. Additionally, they have guaranteed exit liquidity (so long as the TURD/ETH pool has liquidity) for selling their rewards.
4. As the stage one flywheel moves forward, increasing TURD price draws in more TVL until eventually the amount of liquidity in the primary TURD/ETH pool is large enough to saturate the emissions. At this point the pool has a lot of liquidity, think a lot of buyers at all prices. This subtle trick of the market allows projects to spike prices very fast, and then as they increase in price the growth in liquidity makes the price more difficult to move. Thereby locking in an abnormally high valuation and also providing a lot of exit liquidity for the emissions sellers to continually dump into. Stage Two now begins where the price has stabilized because of the growth in liquidity.
5. Stage two is handled differently by different projects. Some choose to lockup a portion of the tokens for x amount of time. This allows them to state ultra high APY’s on their sites, like NRV/EPS/BAO/etc, but in reality they only emit a portion of that now. Other projects with no lockup incentivize early farmers to lock their earned tokens further into the primary TURD/ETH LP pool for the project. Early players typically double down until stage one is finished and in stage two they are the key players of the game.
6. The dumping begins here. Many early buyers whom have now made several x their initial investment on the stage one pump will begin selling into the TURD/ETH pool. This triggers APY’s to fall, and collectively simpleton game theory would tell you to dollar cost average out your rewards as you get them and buy back lower (if you wanted them at all). This is typically also the point at which a project such as CREAM, or FTX/Binance lists Perpetual futures. Now more sophisticated farmers can come in and collect the TURD/ETH pool emissions without TURD exposure. Many projects this marks the top on a price per token basis, additionally sophisticated investors can now sell future rewards at current prices with positive funding.
7. A high quality farm (And this is optional) will typically have some basic product, like a copy of CURVE 3pool, and with the fees generated, produce some dividend income for the protocol. Most just claim to be developing one. This is important, because even if the product offers no differentiating function, the emissions make it profitable to use. These fees then create the Revenue/TVL metric which now gives the protocol a PE and relative valuation leading to a “~ what if ~” as AlexWice would say. The farm now has everything, sans a market differentiated self sustaining product, to climb the coin gecko market cap ladder (even while high emissions force token price down).

The long and slow grind through LP incentivized liquidity

Yield Farming Supernova — the end of the farming craze.

As you can see from the structure I laid out above, the project doesn't really need to provide any value to obtain all the critical metrics required to become a high market cap project. Thats because the farm itself IS the value. And that brings us to why, of potentially how, 99% of these projects will die and strand investors within them.

1. Because AMMs (X*Y=K) don’t allow market makers to price, generally without token emissions its a guaranteed loss to be inside of these these pools. However these are the exit liquidity and are the very pools that set the monetary value of these projects. When either the token price drops low enough due to farmers selling to capitalize their yields, or emissions schedule slowing down, the total liquidity in the pool decreases. This causes price to move more freely along the AMM price curve. Now it runs in reverse. lower token price = lower APY = lower liquidity = no new capital inflows. Without capital inflows to offset the emissions schedule, our TURD token is now doomed to a death spiral. Now in practice this pool will become liquid pretty quickly, thereby stranding all of the other participants in the system. People can get very paper rich but the door is shrinking. Neutral farmers don't have to worry about this, they have most probably presold their rewards in the perpetuals market. This process of new farms popping up, rising to prominence, rewarding early holders and farmers, then when APY’s stabilize its on to the next one.
2. Alternate idea: Uniswap V3 allows market participating to play the competitive pricing game in market making. To keep this short and sweet, this means that the market participants will be rewarded for correctly predicting the future. now these farms futures are not hard to predict, lots of token dumping. This would most likely negate that initial stage one pump, thereby cutting off the farming flywheels. Additionally I have a theory that MEV on Uniswap v3 will allow miners to front run orders easily and essentially high frequency trading will dominate the order books going forward should it be adopted, but that's a write up for another day. In other words, the next gen on chain order books (if we are to believe Uniswap), are not compatible with this farming craze.

Miners Front running Transactions. Now imagine they can front run within the “orderbook”

Bull Cycle End

On the macro side of things, this is playing out at scale. The wealth effect of newly rich blue chip DeFi holders (or eth/BTC) whom borrow stables from COMP/XVS/AAVE/CREAM etc with their blue chips as collateral are why cash yields are so high, and also why they are unsustainable. Akin to 2017, as long as the capital inflows are great enough to maintain the increasing number of farms and their high paper returns, all is well. But also similar to 2017 when the number of new ICO’s expands beyond the inflow of capital, people start to loose money, and when people start loosing money it begins the cascade risk to be priced in. Very dangerous for essentially a ponzi reliant on momentum. This will be the death of these farms, and also the death of 80+% of the TVL in AMM’s. Because the most all of the functioning of DeFi currently is centered around farming in some capacity, TVL’s will drop, protocol revenues will drop, and the dissolution of all these sustainable cash flows will lead to major contractions in the blue chip tokens prices. As they come down, so will borrowed margin in farms. And so the process becomes reflexive. In my estimation, once total DeFi TVL decreases materially for several weeks straight, that has a high likelihood to mark the cycle top.