Thanks to Rin Haung, Yuki Yuminaga, Alex Zhu, Pablo, Terence An@Itos, Jesper, Jermaine W, Simiao Li for great discussions and feedback!

Key Takeaway:

• We might still be one market cycle away from traditional on-chain option to gain significant traction considering overall monthly crypto option volumes is only at $40B (monthly volumes for perpetual is around $1.6T). Currently, DEX to CEX Futures trading volumes is around 2%. If we assume on-chain option will reach a similar market share to on-chain perp, the expected on-chain option trading volumes should be around $800M, which might not be enough to incentivize market makers to market make on-chain.

• Derivative primitives that leverage concentrated liquidity LP positions could be a way to solve the liquidity problem since there is already a lot of liquidity sitting across concentrated liquidity pools. These products (Infinity Pool, Panoptic, Itos, Smilee, etc.) also offer significant product differentiation to CEXs like the ability to speculate on long-tail assets. The main challenge for them is to convince concentrated liquidity LPs to re-deposit into their liquidity pool.

• In the long-term, I’m bullish on Clob-based on-chain option protocol that first utilizes dYdX alike off-chain orderbook matching and on-chain settlement and then progressively putting orderbook on-chain as underlying blockchain infrastructure improves.

Option Market Overview

• In traditional finance, option is the most heavily utilized financial instrument. According to FIA, annual ETD (exchange-traded derivatives) global derivative trading volumes in 2022 (measured in terms of the number of futures and options contracts traded and/or cleared) increased 34% compared to last year, reaching a staggering amount of 83.8B. Option (54.5B) and future (29.3B) volumes accounts for 65% and 35% of the total volume, an increase of 63.7% and 0.1% from last year.

• Contrary to TradFi where options account for more volumes than futures, current crypto derivative volumes is mostly taken up by future: the trading volumes for ETH&BTC options (~$30B) are only around 2% to that of futures (~$1T) in June 2023. This is perhaps due to the introduction of perpetual futures by Bitmex that aggregates liquidity across expiry dates, resulting in much greater capital efficiency than TradFi futures. The greater capital efficiency then translates into better future pricing, lower slippages, and higher leverage opportunities that caters to the risk-loving profiles of crypto traders. For a nascent asset class like crypto, capital efficiency is crucial as it does not have as much liquidity to start with compared to traditional equity market. In the case of crypto option, it suffers from the problem of segregated liquidity across expiry dates and strike prices, leading to much lower volumes than crypto futures.

• On-Chain options take a tiny fraction of DeFi derivative volumes and TVL.

  • Currently, the total TVL for on-chain derivative protocols amounts to ~$1.5B while the TVL for on-chain option protocol is only ~$110M, implying a huge untapped market. Volume wise, within ~$30B monthly derivative trading volumes, DEX option exchanges aggregates only $114M notional volume ($3.7M premium value). This means that on-chain option market is still in its nascency with a lot of market potential.

Future vs Option

• As option gives the buyer the right, not the obligations, to execute their position at the strike price, buyer needs to pay a premium upfront to pay for such flexibility. Such mechanism translates into a convex payoff function for option buyers (fixed downside risk and unlimited upside gains) and a concave payoff function for option seller (fixed upside gains and unlimited downside risk). In contrast with the asymmetric payoff functions for buyers and sellers of option, the buyer and seller of option have symmetric payoff functions. The different payoff functions lead to different user profiles and use-cases:

  • User profiles:

    • Option: higher barrier of entry; buyer: long-gamma, short-theta; seller: short-gamma, long-theta.

    • Future: lower barrier of entry (especially in the case of perpetual future that abstracts away physical delivery and expiry dates), typical retail crypto users who demands high leverages.

  • Use Cases:

    • Option: Income generation (seller), hedging, speculation, volatility exposure

    • Future: Hedging, speculation, higher leverage (for perpetual)

Competitor landscape

There are roughly five types of option protocols based on their mechanisms: structured products, option infrastructures, automated market makers (“AMMs”), central limit order books (“CLOBs”), and protocols utilizing concentrated liquidity pools. Structured products aim to generate returns to LP capital through various yield-generating strategies, usually relying on option infrastructure to mint/settle on-chain options. Currently, structured products consist of option vaults (e.g., Ribbon, Psyoption, etc.) that provide covered call/put positions and more complicated yield products (Cega) that includes exotic options as well as staking, lending, and liquidity provision opportunities. Option infrastructures are essentially smart financial escrow protocols that allows users to create, mint, and settle various financial derivative products. Option AMMs works by using the liquidity pool to take the opposite side of traders. They use algorithms to automatically price options based on a modified Black-Scholes formula as well as the supply and demand of the option contracts. Option Clobs are marketplaces that actively matches the buy/sell orders from option buyers and sellers. Protocols utilizing concentrated liquidity pool refers to protocols that create derivative primitives by taking the opposite side of concentrated liquidity pool providers.

Currently, AMM-powered option exchange takes up the majority of the option trading market shares, similar to the on-chain perpetual landscape in the early days. This is because AMM-powered option exchanges are generally easier to build and cold-start their liquidity compared to Clob-based model. A fully-onchain Clob-based option exchange requires a matching engine that can quickly scan the book and match orders, which is hard to run on a blockchain, and extensive BD resources to find centralized market makers to bootstrap liquidity at its inception. Consequently, most order books only support option minting and settlement but not trading, which is typically done OTC through market makers auctions. Zeta circumvents this problem by composing with Serum’s orderbook infrastructure that checks its orderbook up to two times per second. However, serum’s liquidity was drained due to the FTX incident, and all the users and trading volumes migrated to openbook, a community-forked serum, making Zeta’s option trading temporarily halted. An alternative solution could be a hybrid model of off-chain orderbook and on-chain settlement, which is what Aevo is building right now. That being said, the market share today is pretty non-indicative given its nascency. As more Clob-based model comes into play and more innovations with AMM peer-to-pool model in terms of delta-neutrality and capital efficiency, the market landscape will experience significant changes. Next, I will provide some overviews in the history of option protocols, identifying some key challenges faced by them as well as the improvements that were made.

Hurdles faced by first-wave AMMs:

• Option Mispricing: The problem with AMM peer-to-pool model is that they need to accurately price the option. In traditional market, the price of an option is determined by the demand and supply of options. However, in peer-to-pool model, there is not a demand and supply of option since supply is fixed (automatically act as the counterparty of the option buyer). Therefore, how to derive an efficient pricing model to price the options becomes the main problem faced by AMM liquidity pool models. The most common approach in pricing an option is through the Black-Scholes formula, which takes in asset price, strike price, risk-free rate, time to expiry, and implied volatility. Within the five factors, the only factor that is not observable is implied volatility (IV), which is a metric that reflects the market's expectation of future volatility of a particular security. Traditionally, IV can be calculated through the demand and supply of the option contracts: a higher demand and a lower supply lead to a high IV and vice versa. However, IV is hard for on-chain option protocols due to the inconsistent flow of demand and supply. In Hegic, for example, its IV is calculated off-chain and manually update on-chain weekly. This means that whether you open a $100 or $10000 ATM option it will be priced the same. This is important because unlike makers in a CLOB-based system can dynamically reprice their quotes as they implement new information about the true price, AMM LPs solely rely on the pricing functions embedded in the smart contracts. Therefore, in times of market volatility, the LP can expose to high impermanent loss where IV can be significantly lower than realized volatility. As showed by the graph below, most of the first-generation option AMMs have a static volatility input that does not adjust dynamically based on real-time volumes.

• Unhedged LP positions: In the first-wave AMMs, protocols like Hegic, Dopex, and Premia provided unfriendly experiences for LPs because their collateralizations are unhedged. Such mechanism is due to the concerns toward capital inefficiency as delta-hedging generally needs to take part of the liquidity pool to take long/short positions based on the net delta of their shorting positions. However, this leads to LPs accumulating a high degree of exposure to the underlying assets. As a result, this increases cost of liquidity for LPs, hindering the growth of liquidity pool.

• From a trader’s perspective, they have limited selections of asset choices, strike prices and expiry dates as well as inefficient pricing compared to CEXs. Limited selections of asset choice, strike prices and expiry dates boils down to the limited liquidity because a wider selection of these criteria will fragment the already-thinned liquidity in the pool.

Improvements made by the second-wave AMMs

• Emphasizing delta neutrality to attract liquidity: Lyra’s Valon update first introduced delta-hedged AMM. Lyra mitigates risk by taking a hedged position through either GMX or Synthetix. For instance, when traders hold long positions in ETH call options, Lyra's Market Making Vault (MMV) takes a long position in ETH equal to the net delta when the position was initiated. By doing this, the MMV safeguards itself against the potentially limitless profits that traders with long call positions could earn if ETH's price were to rise. Similarly, Siren Flow also introduced delta hedging system to product liquidity from delta exposure by composing with Prennial. At the same time, other AMM liquidity pool protocols have also tried to fix problems for LP in different ways: Hegic, Premia and Dopex split liquidity pools into call and put vaults, giving LPs more control over the option they underwrite. However, this approach is inferior to Lyra and Siren’s approach because it still offloads the burden of hedging to the LPs and fragment the liquidity.

• Improving inefficient pricing models and introducing partial collateralization for better UX: Compared to the static IV problem in Hegic, protocols like Lyra, Dopex, and Siren Flow all implemented new strategies to enhance the pricing efficiency. Lyra AMM's core mechanism focuses on adjusting implied volatility (IV) and, subsequently, the cost of options in response to market conditions. When the demand for options is high, the AMM increases the implied volatility and lowers it when there is an excess supply. This approach enables the AMM to converge on a market-clearing value for IV for each strike and expiry. Upon listing an expiry Tj on Lyra, a baseline volatility value IVj is initialized along with the ratios of the listed strike volatilities to IVj. These initial values are derived from current market data, using IVj from the 50 delta (at-the-money) strike. After initialization, both IVj and the strike volatility ratios (skew) are determined by the supply and demand for options for a specific strike and its associated expiry. Siren Flow, on the other hand, implements a hybrid on-chain/off-chain Request-for-Quote (RFQ) system to deliver competitive pricing for options trades. This innovation enables Siren Flow to offer pricing comparable to centralized exchange derivatives while retaining the benefits of self-custody and decentralized trading. However, Siren’s approach determines that it cannot serves as price discovery platform as it relies on the data from centralized exchange to derive its IV.

• Additionally, there is innovative approach from the Lyra side to deliver partial collateralization for option sellers that enhance capital efficiency 4-5x. Avalon allows traders to partially collateralize short positions, allowing them to sell 4-5x as many options with a given amount of capital. Partially collateralized shorting is a big deal for two reasons: it provides options traders with a more complete experience that rivals CeFi platforms, and it allows the AMM to provide more efficient pricing. The main difficulty in implementing partial collateralization lies in the difficulty to calculate the initial collateral amounts based on various factors and a robust risk management system to prevent bad debt problems.

CLOB: Limited tractions so far but promising products are launching

• Due to the FTX incident leading to the sunset of Zeta’s option exchange, Aevo is a promising Clob-based option exchange incubated by Ribbon that utilizes a hybrid model of off-chain matching and on-chain settlement. Aevo is built on a custom EVM-rollup and is said to grant access to hundreads of instruments that can be traded across a wide range of strikes and expiries with deep liquidity. To cold-start its liquidity, other than partnering with professional market makers, it aims to integrate with Ribbon’s DOVs as the venue where the sold option contracts settle. Currently, Ribbon has a monthly notional volume of around $30M, paving the liquidity groundwork for Aevo. Additionally, this also solves the misaligned incentives faced by current DOVs (will introduce later), which can potentially drive more volumes to Ribbon. Depositors of DOVs can also take profit or cut losses before expiration, massively increasing the flexibility of Ribbon DOVs. Lastly, Aevo would create liquidity for market makers that currently buy DOV options off-chain, and they could hedge their positions directly on the exchange. Aevo could also be synergistic for other DOV protocols by functioning as an infrastructure layer.

Comparing Clob-based model to AMM liquidity pool model:

• Overall, three metrics is of utmost importance in evaluating on-chain option exchange: liquidity, capital efficiency, asset choices.

• In terms of liquidity, AMM liquidity pool model has a clear edge in attracting initial liquidity because it can easily attract retail liquidity that automate the market making for them. However, in the long term, Clob-based model has a higher ceiling because of it enables professional market makers to market make in the platform. Additionally, protocols like Elixir also unlocks market making on Clob-based system for retail users. Therefore, Clob-based model has an advantage in liquidity attraction than AMM liquidity pool model.

• In terms of capital efficiency, Clobs and AMMs both have different edges. Clob-based model can potentially act as a price discovery platform for options when the volumes is large enough, which can lead to more efficient option pricing. It also does not lead to any impermanent loss problem as faced by AMM models. On the other hand, AMM-based option protocols can improve capital efficiency through composability of its LP positions.

• In terms of asset choices, I think Clob-based model has a relatively edge in that it functions similar to a CEX that can potentially allow more assets being listed on the platform. AMM liquidity model, on the other hand, will have a hard time implementing a wider range of assets because the pricing of these option OTM will be extremely hard. Additionally, traders can make a fortune by buying straddles, draining the liquidity pool.

For structured products, they face two problems of misaligned incentives and mismatch of risk profiles:

• Misaligned incentives between market makers and structured product vaults: currently, since on-chain option exchange hasn’t taken up, most option vaults work by building a covered call strategy using LP’s deposits and then selling it to the market makers via auctions. Market makers will generally buy calls from Ribbon depositors at a negotiated premium price and then hedge by selling an equivalent amount of calls with the same specs (expiry, strike). By doing this, they capture a spread between the price they buy from Ribbon and the price they sell on Deribit. The problem here is the contradictory interest faced by market makers and option vaults since market makers want to purchase the calls as cheaply as possible while depositors want strategies that outperform the market on a risk-adjusted basis. However, since DOVs are constrained to selling the options to a limited number of market makers, most options sold are underpriced, leading to minute yields for LPs.

• Mismatch with current crypto user risk profile: Essentially, LPs in DOVs are long theta and short gamma since most vaults are covered call/put options. However, crypto is inherently very volatile, which makes LPs generally underperform in a bull market and bear a similar loss in a bear market. Such a payoff function does not match most crypto users as they are in crypto for asymmetric upsides instead of collecting minutiae yields.

• Future: The maturity of Clob-based option exchange can solve the misaligned incentives between market makers and structured products. As more Clob-based option exchange like Aevo emerges, this can create a price discovery venue for DOVs to find their sellers, solving the unbalanced power dynamic in OTC. Additionally, as mentioned above, DOVs that compose with on-chain Clob option exchanges can allow depositors to take profits or cut losses before the DOV expiration, allowing greater flexibility for depositors.

To summarize, current on-chain option protocols, whether it is Clob-based or AMM-based models, have achieved little traction in terms of volume and liquidity. This introduces a typical chicken-and-egg question if there is no liquidity, there is no volume, and vice versa. From the perspective of liquidity, on-chain liquidity providers face problems relating to the mispricing of options that leads to realized volatility often higher than implied volatility calculated, discouraging them from participating in the liquidity pool. There is also little motivation for traditional market makers to market make in those protocols given the minutiae volumes. In the far long term, I’m bullish at Clob-based model takes a similar market share of option volumes like that of DyDx. But the on-chain option protocols today do not offer enough product differentiation from that of CEX, which introduces our next wave of protocols that tap into the ample liquidity sitting at the concentrated liquidity pools.

The next wave: Protocols utilizing concentrated liquidity pools

The core idea behind these new waves of protocols is that Uniswap v3 liquidity provider (LP) positions can be seen as tokenized short puts. This is because the economic payoff function of LP is mathematically identical to selling a put option. For Uniswap V3 LPs, they are essentially short gamma and long theta as LPs incur losses when the price of underlying assets moves more rapidly and gain swapping fees as time goes by. Therefore, various protocols, including Panoptic, Infinity Pool, Smilee, and Itos, are trying to capture the ample liquidity of short option positions sitting in concentrated liquidity pool to build their derivative primitives. Although similar in their underlying philosophies, there are some notable differences among these products in terms of their designs and offerings.

Mechanism Overview

Panoptic

• Overall, Panoptic is consist of liquidity provider, trader (option buyer/seller), and liquidator. Liquidity provider needs to deposit fungible tokens in the Panoptic pool at any ratio of token1/token0. This liquidity pool can be borrowed by option sellers to create short options by depositing the liquidity to a corresponding Uniswap v3 pool. Similarly, a trader can create a long option by removing liquidity out of the Uniswap v3 pool. For example, suppose a trader wishes to buy a put option with a strike price of 1000 USDC and a width of 10%. When the trader buys this option, a portion of the USDC liquidity in the range of 909 - 1100 USDC per ETH is taken out of the Uniswap V3 pool and moved back to the Panoptic Pool. The cost of this option is the amount of fees that would have been generated if the liquidity stayed in the Uniswap pool. Now, let's consider different scenarios:

  • If the ETH price is above 1100 USDC when the option is bought, the option is out of the money (OTM) – it's not profitable yet, so no premium is accumulated.

  • If the ETH price falls below 1100 USDC, the option starts to gain value - a premium starts to accumulate. If the ETH price falls further below 909 USDC, the option is now in the money (ITM), meaning it's profitable. At this point, the option stops earning fees, and the user can decide to exercise the option.

  • If the ETH price stays above 909 and below 1100 USDC during the option's life, the option is considered "near the money" and "in range", and its cost is equal to the foregone Uniswap fees plus an additional liquidity spread. The user can decide to close their option position, but will owe the accumulated premium.

  • When exercising the option, the user has to return the borrowed liquidity, which is now in ETH. So, they send ETH to the Panoptic Pool and keep the USDC they received when they initially bought the PUT option. This means they have effectively sold their ETH for 1000 USDC, even though the market price might be less than 909 USDC.

• Offering: everlasting option with 5x leverage for option seller & 10x (could be up to 20x if the pool utilization rate is high) leverage for option buyer on any token available on Panoptic liquidity pool.

Itos

• Different from Panoptics which relies on UniswapV3 AMM to create long/short options, Itos provides similar products by building its own concentrated liquidity AMM (CLMM) with negative liquidity provisioning positions called takers on top of positive liquidity provisioning. The AMM structure has three market participants: maker, taker, and trader. Maker and traders are the same concept in UniswapV3 where makers refer to liquidity providers who deposits fungible token pairs to liquidity pools and traders refer to participants who swap one token for another in liquidity pools. Instead of providing liquidity like makers, takers reserve liquidity for the purpose of taking trades. These takers always trade with the swap, paying fees to ensure trading into the more valuable token. This allows the creation of TakerPuts and TakerCalls, which mimic put and call options respectively. For example, when a trade occurs on the decentralized exchange (DEX), takers benefit by riding on the back of that trade while paying a funding rate to secure sufficient liquidity for executing the trade. By consistently leveraging every trade, the value of their position increases as prices move, unlike the value of a market making position, which tends to decrease. Therefore, this allows the taker to experience a long-option-alike payoff function (as shown in the green line in the following graph), effectively hedging against any maker position.

• Offering: automated market taker position that can hedge external AMM LP positions & offers everlasting option for any token listed on Itos AMMs.

Infinity Pool:

• Like Itos, Infinity pool is a leverage trading DEX built on top of its own CLMM called float pool. LPs can directly deposit fungible token pairs or deposit their Uniswap V3 LP tokens (the protocol will convert the LP tokens on LPs’ behalf) to the float pool. Float pool serves two functions: 1) spot trading and 2) lent out to leverage traders. In the second case, the LP token borrowed will be extracted from the Float pool to private pools (swappers) where traders can perform unlimited number of swaps at a predetermined strike price for free. In return, traders pay a funding fee to the LPs through a fixed-term loan (1~40x leverage) or a revolving loan (40x+ leverage). On a high level, a trader can achieve leverage trading experiences on any asset available in the float pool by borrowing the corresponding LP token at desired tick ranges via an off-chain matching engine, converting it to fungible tokens in the private pool as well as any subsequent external swapping if needed, and eventually repaying the loan in any mixture of tokens along the LP curve. For example, borrowing from the example outlined in their WP, a trader can long ETH with 10x leverage by borrowing a 1000 USDC worth of ETH/USDC LP token deployed at a tight liquidity range centered at 900 USDC assuming the market price of ETH is at 1000 USDC. Since the price of ETH is currently above the liquidity range, the trader can redeem that LP token for 1000 USDC and swap the 1000 USDC for 1 ETH at any spot DEX (assuming no swap fee or slippage). If ETH drops below 900USDC, the LP position would consist of 1.11 (1000/900) ETH, which would require LP to purchase additional 0.11 ETH if they decide to close the position. Therefore, the worst scenario for trader is when ETH is at 900USDC, where the trader needs to purchase 0.11ETH with 99USDC. This means that the trader only needs to post 100USDC as collateral to achieve a 10x leverage in this example.

• Offering: Everlasting option with a perpetual alike UX that offers theoretically unlimited leverages for any token listed in Infinity Pool AMMs.

Key Takeaways:

Fundamentally, the key challenge for the above protocols is to convince concentrated liquidity pool LPs to redeploy to their protocols. Among the three, Panoptic does not need to bootstrap as much liquidity compared to Infinity Pool and Itos because their option products are created through interactions with UniswapV3 pools. However, Panoptics still need to have enough liquidity for option buys and sellers to borrowed against for leveraged trading experiences. As such, the three protocols implement different fee models to compensate LPs. In Panoptic, LPs receive a 20-60bps commission rate depending on the pool utilization rates. In Infinity Pool, LPs earns a swapping fee when they are not lent out and earns a loan interest payment based on the leverages traders take. In Itos, the LPs (makers) earns a money-market borrow rate (cost of carry) multiplied by the liquidity utilization rate in their active range that is reserved by the takers. Theoretically, all the above protocols should offer a fairly competitive rate that takes into account the expected volatility of their tick ranges compared to the fixed swapping rate offered in UniswapV3 AMMs.

The most exciting part of these derivative primitives is that they offer speculation on small market cap tokens, an experience that is not possible in any other DEX or CEX. On the other hand, it might be difficult for these protocols to compete with CEX for major asset pairs because of the relatively high fee rates. This is especially true for Panoptic because of the swapping fee in the UniswapV3 pools. For any option created in panoptic, the user will pay at least a 30bps (20bps commission rate and 10bps swapping fee in UniswapV3), which is even higher than many other derivative DEXs, let alone CEXs. In the case of Infinity pool, they still require interactions with external AMMs to successfully close their trades, leading to a relatively higher fee. Itos, on the other hand, might have a more competitive pricing compared to the other two because they do not require interaction with any external AMM and they adjust the swapping fees based on volatility. That being said, the different offerings of these protocols mean that they can tap into users with different risk profiles: Infinity Pool, with its theoretically unlimited leverage, might cater to more perpetual future traders, while Panoptics and Itos might cater to more sophisticated retail traders and DAOs who demands hedging directly on-chain.

Closing Thoughts

Overall, this new wave of derivative primitives unlocks the liquidity sitting at concentrated liquidity pools by discovering the similar payoff function between UniV3 LP positions and option sellers. In a nutshell, all these protocols aim to take the other side of the impermanent loss, which solves the liquidity problem that hinders options market growth. Additionally, I’m also excited to see various structured products built on top of them if they manage to gain some sizable liquidity.

While these new derivative primitives sound promising, it remains to be seen if they will gain sufficient traction. The key challenge will be convincing AMM LPs to deposit capital in their protocol instead of Uniswap (or deposit their Uniswap LP tokens into the protocols). Therefore, one of the main differentiators will be how comfortable they can convince LPs to deploy their liquidity positions into their liquidity pool. With both Panoptics and Infinity aiming to launch their testnets by the end of July, I’m looking forward to seeing the market reactions to this new wave of AMM-powered primitives.