Discover more from Artemis Big Fundamentals in Crypto
Deep Dive: Maple Finance
Here to stay?
We are Artemis.xyz, a lean team building a platform that adds context, standardization, and analysis to blockchain data. We are rapidly iterating on our MVP, and while we don’t cover Maple currently, we are working to bring on data for Maple and projects like it. We’d love your honest feedback on our product and content - let us know what we got right, wrong, or missed altogether! Subscribe to our newsletter below, and follow us on Twitter.
Maple is a leading DeFi under-collateralized lending platform. Approved financial institutions (“delegates”) use Maple’s infrastructure to source lender capital (primarily USDC) and deploy it to borrowers on-chain.
Lenders are a mixture of retail, DAO treasuries and asset managers.
Delegates include Orthogonal Trading, and Maven11.
Borrowers are mostly crypto native market makers and asset managers such as Alameda, Amber Group, and Wintermute (collectively > 50% of the loan book).
Since launch, over $1.5B in loans have been originated on Maple, which takes a 66 bps fee (annualized) on this volume. QoQ origination growth was excellent in Q1 ‘22 at > 70%, but has shown signs of fading in Q2, driven by the broader turmoil in crypto and recent liquidity challenges.
The bull says —> DeFi and DeFi lending are here to stay, and will grow significantly in the long run. Under-collateralized lending will be an increasingly important segment of DeFi, as it frees up borrower funds for growth initiatives and working capital. Maple is currently a leader in this segment, and has executed very well to date (significant growth, partnerships with large and mostly reputable institutions, and no defaults to date). It has a chance to continue to build a trusted brand, aided by strong and reputable partners. It can also continue to build a long term moat by addressing new customer segments such as miners, offering additional lending products, and building value add technology tools for all stake holders (workflow, underwriting, risk management, etc.).
The bear says —> Maple connects non-KYC / AML lenders (including retail) to borrowers that invest in risky assets without pledging collateral. What could go wrong? Maple is exposed to systemic risks and downward reflexivity in crypto, which could lead to (a) washout of market makers / asset managers leading to defaults or drop in borrower demand, (b) general FUD leading to a pullback in the lending supply (we are currently beginning to see this, and it is unclear how losses will be allocated across lenders), or (c) USDC de-peg or bank run, which we view as far less likely. Current climate aside, Maple’s borrower base is also fairly concentrated (25% Alameda), and the borrower diligence performed by delegates is opaque. Additionally, there could be long term pressure on volume and fees driven by competition (ClearPool, TrueFi, etc.). Token rewards that have been used to incentivize usage are likely unsustainable. Finally, it goes without saying that there are meaningful legal and regulatory risks.
MPL is the protocol’s primary token, and can be used for staking (into xMPL), providing pool cover (akin to junior debt), and governance. Maple currently uses 50% of protocol revenue to buy back and distribute MPL to stakers, though this is likely to come down over time.
There are many ways for market participants to play Maple: holding, trading, or staking MPL, lending capital for yield, providing cover to earn even higher yield, or pursuing a combination of the above.
We recognize that we are publishing this piece during a period of extreme uncertainty in crypto markets (see June 2022 updates section below). Many of the numbers below are likely to change, and we have tried to avoid specific predictions about Maple’s future. Rather, we’ve tried to provide frameworks for thinking through its growth prospects, key risk factors, and value capture potential. We hope this is useful for anybody interested in Maple, DeFi lending, DeFi investing, or protocol / token economic design.
Maple Finance has built a platform that allows pre-approved financial sponsors or “delegates” (like Orthogonal Trading, Maven11…and to a smaller degree Celsius) to build under-collateralized lending businesses. Maple itself does not lend or borrow. It builds technology on top of which delegates can source capital from lenders, and fund borrowers. In this sense, Maple’s business model is somewhat akin to that of Shopify or Wix, where the rails are provided for third parties to build their own two sided business.
Maple is currently operating on Ethereum (~$1.5B originated), and recently launched on Solana (~$100M originated). See below for the key stakeholders and an overview of how capital flows through Maple.
Pool Delegates — Delegates are the financial sponsors (“funds and industry professionals with credit experience”) that use Maple to raise and allocate capital. There are currently 5 delegates on Ethereum, and 2 on Solana, each of which were approved by Maple. Delegates are tasked with (1) attracting capital to their pool, (2) sourcing and performing diligence on borrowers, and (3) coming to terms with both sides. Delegates are the key stakeholder in the Maple ecosystem, as they dictate both capital inflows and loan book quality.
Lenders — Depending on with whom they want to invest, holders of USDC or wETH can participate as a lender on the platform to earn expected APYs ranging from 3% to 11%. Some lending pools such as Orthogonal’s Alameda pool (which we have labeled Alameda for simplicity because it is the sole borrower) are only open to select lenders. Other pools (such as Orthogonal’s primary pool) are open to any lender with no KYC / AML process.
Cover Providers — Cover is a more junior tranche of capital. Assets pledged as cover will be liquidated first in the event of a default (after collateral, if any, is liquidated, but before lenders are impaired). In return for assuming higher risk, cover providers earn higher yields (Orthogonal’s pool, for example, currently offers 10% APY for lenders but > 25% for cover providers). The pool delegate is required to put up cover itself (a minimum of $100K USDC, though often much larger). The cover is quite small as a percentage of the overall loan pool, ranging from 0 to 3% of the total principal outstanding, meaning that the cushion in case of a default is small.
Borrowers - Borrowers, generally crypto native asset managers and market makers, engage with delegates to negotiate loan terms and source capital. Maple suggests that most borrowers are delta neutral (not taking directional price risk, but rather betting on arbitrage / spreads), which would in theory mitigate risk and make them a better fit for under-collateralized lending. That being said, 51% of the borrower base is “diversified” while 34% are market makers, so it is a bit difficult to tease out the actual amount of delta risk in the borrower base, and how borrower cash generation is tied to market volatility, price drawdowns, and liquidity.
For simplicity, when breaking down lending pool data in this exercise, we will focus on Maple’s Ethereum segment given its relative size.
Lenders - Lenders earn an interest rate on the capital that they provided to a pool. These interest rates range from roughly 4% to 10% APY. Maple currently does not offer a breakdown of loan-level interest repayment schedules (i.e. what has been paid, when is the next payment, is the loan current, etc.), but does allow borrowers to track how much interest they’ve accrued to date.
Components of interest - Lenders earn via:
Interest paid by the borrower in the pool’s denominated currency (mostly USDC per the table above).
Interest paid by the protocol in MPL tokens. These MPL rewards artificially enhance the yield earned by a lender, and thus helps increase the supply of capital on the platform while keeping borrower interest rates low (more on this later, as it is an important dimension when considering Maple’s sustainability).
Where interest flows - While it depends on the pool (Orthogonal’s Alameda pool has set a different fee schedule), the interest generally flows as follows:
80% to lenders
10% to pool delegate
10% to cover providers
Cover Providers - Cover providers generally earn a fixed 10% of interest, though it depends on the pool (5% in the case of Alameda’s pool). The APY earned by cover providers is generally larger than that earned by a lender. This risk premium is priced by the market, though, as cover APYs can flex up or down depending on the size of the cover pool (fixed interest spread over a larger or smaller pool of capital).
To provide cover, one needs to pledge USDC and / or MPL to receive a tokenized receipt in the form of balancer pool tokens (BPT). Delegates must pledge $100k + in USDC and MPL, which along with participation from other cover providers, helps fund a pool that drives liquidity between USDC and MPL (this is similar in concept to an LP token for an AMM). This means that cover providers are taking on impermanent loss risk, capital impairment risk, and asset price risk.
Pool Delegates - Delegates earn both upfront and ongoing fees.
Establishment Fee - Delegates earn 33 bps (.33%) of each loan as origination fee. This occurs when a borrower takes out a loan (the establishment fee is deducted from their proceeds), not when the capital is raised. This up front fee helps cover the costs associated with a delegate’s underwriting. It is unclear if delegates make a margin on these fees.
Ongoing Fee - As noted above, delegates generally earn ~10% of interest, though it depends on the loan pool.
Maple Treasury - These are the revenue streams captured by Maple itself, after other revenue is directed towards delegates, lenders and cover providers.
Establishment Fees - The Maple Treasury captures 66 bps annualized from the establishment fee (99 bps total = 66 bps to treasury + 33 bps to pool delegate). This is the largest and highest quality revenue stream generated by the protocol, and represented 83% of the total revenue in Q1. See below for an illustrative example of how establishment revenue flows.
Staking Fees - While it is not clear what constitutes the “staking” portion of Maple’s revenue, it is likely from the treasury earning yield on MPL or other on-chain assets (see below for revenue breakdown from Maple’s Q1 ‘22 treasury report).
From an investor’s perspective, we would not ascribe much value to this revenue stream, as (a) it is unclear how this stream will scale as Maple’s core business grows (whereas establishment fees can be easily modeled), and (b) especially given the recent tumult in crypto markets, we do not underwrite yield farming or liquidity mining as a reliable source of return.
Expenses - Maple includes cost detail in its treasury reports, noting that it turned a profit in Q1 ‘22.
It is unclear if these financials are accrual or cash based (we would guess cash). As Maple / DeFi matures, there will need to be more disclosure and consistency.
These financials do not reflect the treasury beginning to spend 50% of revenue on MPL buy-backs (see below).
It is unclear how, if at all, treasury profitability relates to value accrual for Maple token holders, lenders, or borrowers (beyond Maple having enough capital to continue operating).
TL/DR: the protocol itself earns most of its revenue via a one-time origination fee, while lenders and cover providers earn yield. Pool delegates benefit from both.
Role of Token - Maple’s core token, MPL, has both governance and economic characteristics.
Governance - Token holders are able to vote on Maple Improvement Proposals (MIPs), assuming they hold at least 25 tokens.
Fees - Owners of MPL can stake their MPL into xMPL to earn yield. As a token class, xMPL earns MPL rewards from token buy-backs / distributions (more detail in the “Value Capture” section below).
Pool Cover - Finally, holders of MPL can help fund the USDC / MPL balancer pool, through which they can receive the BPT required to participate in pool cover (and thus earn higher yields than staking or lending in isolation).
Trading Behavior - MPL currently trades at ~$14 (as of 6/23/22), roughly at its level from EOY 2021, but significantly below its peak of > $60 in early April 2022. Quarter to date, MPL has dropped slightly more than ETH (down 74% vs. 67% for ETH).
Supply / Distribution - There is a max supply of 10 million MPL tokens, allocated as follows.
The 2.5M tokens allocated to the Maple team and its advisors vests over 24 months (currently ~15 months through, meaning ~1.5M tokens have been unlocked to date). Similarly, seed investors are on an 18 month linear vesting schedule, meaning that nearly all of the 2.6M tokens have been unlocked. Roughly 6.2M tokens are circulating currently.
Maple’s token economic design should, in theory, drive demand for MPL as platform usage and revenue grows. This is because, thanks to MIP 008, Maple uses 50% of protocol revenue (33 bps of originations, or 50% of the 66 bps earned by the treasury) to buy MPL tokens on the open market, and distribute them pro rata to xMPL holders (e.g. if revenue was $2M, and $1M of MPL tokens are purchased, you would receive $100k in MPL tokens if staking 10% of total xMPL tokens). Moreover, one needs to own MPL tokens to participate in governance and earn the higher yields generated by providing pool cover. It is likely the Maple team will continue to develop uses that drive demand for and reduce sell pressure on MPL.
That being said, a few nuances make it difficult to predict how MPL’s price action will correlate to platform usage going forward.
xMPL does not currently have a lock up, meaning that holders of MPL can stake into and out of MPL instantly. If rewards dry up, or MPL loses economic and utilitarian value, xMPL holders can de-risk and sell their MPL (including accrued rewards) on the open market for a more stable asset. This related to the “velocity problem.” This may help yields self-adjust, though, as rewards would be spread out over a smaller number of stakers.
As noted, Maple currently spends 50% of protocol revenue on buying back MPL and distributing it to stakers. This aggressive buy-back program will likely be unsustainable, and protocol revenues will need to increase dramatically to preserve xMPL yields.
Widespread default in the portfolio could trigger MPL liquidity issues, and put downward pressure on MPL’s price. Maple’s balancer pool currently has ~$15M of liquidity (as of 6/23/22), which has declined nearly $1M in the last week. Cover pools liquidating in excess of that amount could be problematic.
Though MPL is a liquid token, we still view it as a venture bet (like most other tokens in crypto) given the asymmetric upside, but potentially existential market and regulatory risks. If you are considering a long term hold, we would not recommend investing solely off of run rate revenue or P/E multiples that appear reasonable relative to crypto or TradFi comps. This may give too much credit for recent performance, and incentive and buy-back levels that could be unsustainable. We will propose a few specific mental models and frameworks on how to value projects like Maple in a future post, but currently believe the market is too volatile to propose a high conviction view on valuation range.
How to play Maple
To be clear, we do not want to be in the business of providing investment advice. Rather, we want to equip people with the tools to make the best decision for themselves. APYs are likely to be dynamic in the coming weeks, with lenders and borrowers adjusting their risk apertures in light of the volatility within crypto. We have already seen some DAO treasuries pull out of Maple lending pools and many other lenders request withdrawals. We will continue to monitor this trend.
Buying, holding, or trading MPL tokens - This is a pure play venture bet on price appreciation in MPL, and will not drive any yield for the holder. Above, we have attempted to analyze the value capture potential of the token, but there are more variables than answers at this stage given token economic design that is likely to change.
Staking MPL into xMPL - Staking gives MPL holders the opportunity to generate a modest yield on their MPL tokens, via the token distribution mechanism described above. Yields are currently 3.4% APY (which is only slightly higher than the 2.86% 1 year US treasury rate. Staking MPL does not currently require a lock up, so there is little disadvantage of staking as a token holder. Yields will depend on (a) the amount and value of tokens staked, and (b) the amount and value of MPL tokens bought back and distributed to xMPL holders. This is a somewhat circular dynamic, so it is difficult to forecast long term yields especially given the rising rate environment which will likely lead to capital reallocation.
Lending - Lending provides an opportunity for USDC and / or wETH holders to earn 3-10+% yields on their capital (again, part of these overall APYs are juiced by MPL token rewards). Lending into a pool is usually associated with a 90-180 day lock-up. As noted, some pools do not have enough liquidity to support the currently very high withdrawal volume.
Providing cover - If one wants to earn a higher yield (but is comfortable with increased risk), they can participate in a cover pool by pledging USDC and MPL into the balancer pool, and redeeming BPTs. As noted, cover yields can be > 25% (currently ~27% for both Maven11 and Orthogonal USDC pools), and have risen dramatically in recent days. Our hypothesis is that this reflects a risk-off mentality within crypto markets, with people pulling capital from cover pools (and thus each participant / unit of capital in a cover pool is entitled to more interest).
Some combination of the actions above - You can also mix and match the strategies above to stack yields (for example, you can stake and provide pool cover at the same time). Maple itself calls this out (via a very simplified chart with no axis units) in their token economic overview.
Market Environment Updates (as of 6/23/22)
Maple has no direct 3AC exposure, per their website.
Celsius is a delegate on Maple’s platform, but all of the capital comes from Celsius’ own balance sheet (and thus Celsius customers), rather than lenders on Maple’s platforms. In other words, the delegate pool is not open to outside lenders. Borrowers out of the pool, at first glance, seem to be of a similar profile to the borrowers of other pools. Finally, the size of the pool is relatively small with ~$15M of loans outstanding (~2% of total). To Maple’s credit, they have been proactive and clear in addressing their Maple exposure, both through updates on their website, as well as the team’s Twitter activity. Maple does not have direct exposure to 3AC.
The Orthogonal Trading pool, the largest on the platform, has a $10M loan outstanding to Babel, a CeFi asset manager. Orthogonal has posted an update in which it suggests there have been no late or missed payments across its pool, and that it is working towards a solution with Babel.
Maven11, a pool delegate has published a helpful update trying to mitigate FUD and withdrawal activity. The TL/DR is that the loan book appears to be in tact, with borrowers continuing to pay back interest on time.
As noted, there have been liquidity issues amid significant withdrawal volume on the platform (from lenders that are past their lock up period). The Company is advising that more cash will be available to support withdrawals as active loans within each pool mature and get paid back. Still, it is unclear how much default risk exists in the outstanding loan book.
Unique “reputation based” model - As noted, Maple offers technology on top of which institutional lenders can build a loan book. While the number of delegates today is fairly concentrated and approval requires permission from Maple, this approach is arguably very scalable (each delegate helps bring capital / borrowers to the platform), and results in higher diligence standards due to the delegates’ credit expertise (this is the thesis, at least). As noted above, we view the business model and go to market as similar to that of Shopify in that Maple provides tooling to delegates, and will benefit from their growth by taking a percentage (currently 66 bps annualized) of loans originated.
Potentially key role in DeFi’s growth - Many of DeFi’s original lending success stories (Maker, Aave, Compound etc.) have offered over collateralized lending products (e.g. lock $150 to borrow $75). While this makes conceptual sense from a risk standpoint, under-collateralized lending does drive growth and credit creation in any economy, as it frees up more borrower capital to use for growth, working capital, or other initiatives. For context, unsecured debt as a percentage of total corporate debt has been climbing for decades in TradFi. If you are a believer in DeFi and on-chain lending, it is not unreasonable to believe that the mix will shift towards under collateralized lending going forward (depending on the use case). There are many risks that will need to be addressed (un-collateralized lending can be very problematic and lead to systemic issues), and we have not yet seen the full impact of a market meltdown on loan book quality in this space.
Impressive growth - Since launching in 2021, Maple’s origination volume and revenue have grown dramatically. In Q1 2022, Maple originated > $600M of loans (> 70% QoQ growth). ~$400M of additional loans have been originated quarter to date, which if pro rated, would exceed last quarter, but given the current state of the market, it is highly possible origination volume comes in down QoQ. Maple logged $1.9M in total revenue in Q1 (up 128% QoQ), though > $300k came from non core staking revenue which we would be hesitant to underwrite from an investment standpoint. Maple recently went multi-chain, launching on Solana which could be a boon to growth going forward (it also plans on launching a Solana specific token).
No defaults to date - Through mid June 2022, the platform had not yet experienced a default, an impressive feat given its under collateralized nature. Especially given the chaos in crypto markets right now (which will likely have an impact on market maker revenue and profitability), we are likely to see at least some defaults going forward (Babel is a borrower). Still, to date, the delegate driven sourcing and diligence has proven to be robust.
Potential to build ecosystem - Over time, Maple can leverage its deep relationships with delegates (and to some extent lenders and borrowers) and introduce new, value add products (underwriting tools, risk management tools, and additional software and services that allow delegates to source and deploy capital more efficiently). Other possible growth levers include breaking into new customer segments, introducing new loan types and durations (currently almost all 90 or 180 day), and continuing to build on other chains.
Credit & Underwriting - There is a significant degree of idiosyncratic risk in play for Maple stakeholders, specifically lenders and cover providers.
Lenders entrust delegates to source and diligence borrowers, but while some loan level detail is available (borrower, amount, interest rate, length, and whether it is active or matured), there are key items missing. Specifically, it is not clear whether outstanding “active” loans are at risk of a potential default, and whether interest has been paid on time to date.
Moreover, it is not clear what diligence the delegate is actually performing on borrowers, and there is no clear way to assess a borrower’s trading strategy, team quality, financial performance, past repayment history, or other proxies of creditworthiness. It is also not unreasonable to imagine a scenario in which delegates overlook the leverage granted to a borrower by other on-chain or off-chain sources (this is similar to what occurred in the Archegos Capital blowup in the equities market).
It is not clear whether the principle associated with the loans is being paid back out of cash flow, or because the borrower is re-financing. There may be risk to default in the existing loan book if sources for borrower financing dry up (whether through Maple or elsewhere) .
Alignment is also a potential issue, as delegates make an upfront fee (33 bps) and only need to put minimal capital at stake ($100k + as cover), meaning even if all loans default, it is possible the delegate still makes money if they had originated more than ~$30M (in fairness, many loans in TradFi are syndicated / sold off). There is also no way to test for potential conflicts of interest between delegates and borrowers.
Finally, there is a degree of concentration within the borrower base which should be recognized (50% on a capital basis from Alameda, Winter Mute, and Amber Group).
Broader macro / systemic risk in crypto - DeFi grew dramatically in a very short period of time (exceeding $250B+ of TVL at its peak), and still sits at ~$70B in TVL. There is a web of interconnectedness between funds, DeFi protocols, and other stakeholders, that has already shown clear signs of downward reflexivity given all of the cross-collateralization and leverage in the system. While one could argue that the ecosystem has been resilient all things considered, it has of course been an incredibly tumultuous stretch that has resulted in multiple blowups and the destruction in $1T+ of market cap across crypto. Maple could be exposed via:
A washout of market makers or a dramatic decline in liquidity in the space which could result in default or reduce borrower demand.
General fear among both retail investors and institutions looking for yield, leading to an increase in withdrawals and a pullback in lending supply.
A de-peg or bank run on USDC, which we view is unlikely.
Moreover, Maple has never experienced withdrawal volume like it currently is. In the event of widespread default, it is unclear how losses would be absorbed across lenders in the pool. If some are able to withdraw earlier than others, do they recoup 100% of their principle, while those locked up until a later time absorb all of the impairment?
Competition and fees - Maple’s delegate base today is highly concentrated (5 delegates in total on Ethereum, which have originated > $1B). If any of these delegates depart for other under collateralized platforms like Clearpool or TrueFi for any host of reasons, lenders and borrowers could follow. Moreover, it is possible that competition puts downward pressure on fees (origination fees could drop from 66 bps, and token rewards could be pressured). Maple will likely attempt to develop additional products and token incentives to offset the competitive pressure.
Legal and regulatory risk - We have more questions than answers on this front, but do believe that high volume institutions borrowing money provided from non KYC / AML customers (including retail) could be problematic in the eyes of regulators, especially if there is significant impairment or there are tax avoidance and collection issues. Moreover, because there have not yet been defaults, it is unclear what recourse Maple would have (based on an interview with management, there do appear to be legal agreements backing each on-chain loan, enforced in the Cayman’s but pursuable in NYC).
Sustainability of yields & role of MPL - As noted above, MPL rewards have multiple implications:
They artificially raise lending APYs (a significant portion of yield comes from MPL rewards rather than actual interest paid by borrowers) —> This keeps interest low for borrowers while raising interest paid to lenders, thus artificially increasing demand on both sides of the market.
Staking yields are propped up by spending 50% of protocol revenues on MPL buy-backs / distributions —> This reduces MPL sell pressure and preserves price.
These token incentives (of course very common across crypto) may taint the ability to thoughtfully assess true product market fit and business model sustainability.
These rewards will likely run dry at some point, and the APY uplift from rewards will diminish meaningfully if MPL’s price drops, and / or the size of the loan book increases.
Tech risk - While there are no specific red flags to call out, smart contract risk, technology, and cybersecurity risks are of course an overarching concern when such large sums of capital are transacting immutably.
For exploratory purposes, the authors and other members of the Artemis team purchased and staked nominal amounts of MPL, and participated as lenders in the Orthogonal Trading pool.
Maple Finance Discord