Sci-Finance: Crypto trading in the next decade
December 27, 2019, 2:30PM EST · 12 min read
As we enter the dawn of a new decade, the genesis of which will be marked by too many end of year posts containing pithy ‘hindsight is 2020’ jokes, I want to instead look to the future evolution of crypto trading. But we’re not going to be able to do this without first peering into finance’s past!
Historically, every epochal change in trading has been accompanied by a combination of both business model innovation and technical improvements. These innovations included quantitative improvements to valuations, such as the use of the Black-Scholes equation for options pricing, and infrastructure for electronic markets that let new competitors (such as BATS) enter the marketplace.
The past 40 years of development have warped market structure and valuation models and the next 40 years look to be no different. In the 1980s and 1990s, electronic markets and novel valuation models helped propel options and futures trading to new heights. Cumulatively, this led to new forms of algorithmic trading and market making that made it easier for passive investing — and overall market participation — to increase in the 2000s and 2010s.
In crypto, we have slowly begun to see a similar combination of business model and technical improvements that are portents of the future — namely, exchange tokens and on-chain trading and lending. These portholes to the financial systems of the future are the bellwether for a new quantitative era of crypto trading.
Exchange Tokens: Business model innovation or regulatory arbitrage?
The crypto trading boom of 2017 led to exchanges and financially-savvy entrepreneurs realizing that futures and other derivatives products were the most profitable way to package crypto's inherent volatility for customers. This led to a flood of new crypto futures exchanges, all of which are competing for BitMEX's first-mover and liquidity advantages. What came out of this race to dethrone BitMEX was an amazing business model innovation that seems innocuous enough: the exchange token. Given that most other tokens lack utility and value accrual, it is natural to wonder why these tokens are special and accrue value. For this, it helps to get some context from how revenue is shared at traditional equities venues.
In these markets, exchanges such as the New York Stock Exchange (NYSE) or the Chicago Mercantile Exchange (CME) share revenue with market makers as a function of their pro-rata share of exchange volume. Why do they do this? Exchanges don't want to hold risk themselves — they want to collect a toll yet don’t want to do any maintenance — but they need to ensure that they have tight spreads (e.g. good prices) and meet regulatory requirements (e.g. Regulation Neutral Market Service, or RegNMS). These exchanges, do this via fee rebates — if you trade more than 10,000 shares per day and you normally pay 0.01% per share traded, then the venue gives you a discount to 0.005% — which are paid quarterly and/or infrequently.
On the other hand, exchange tokens, such as Binance’s BNB, FTX's FTT, CoinFLEX's FLEX, or Huobi's HT, are designed to provide more instantaneous rebates. How does this work?
- Market makers are paid rebates in these tokens, continuously, which changes the relative supply of these tokens (e.g. new tokens are minted).
- Exchanges infrequently burn a lot of these tokens, which implicitly makes these tokens more valuable and since market makers are already quoting these tokens (e.g. providing bid/ask spreads on FTT/USDT, for instance), they can effectively get paid "continuous" rebates.
- Continuously paid rebates admit better capital optimization for market makers allowing them to
- Can reinvest their rebates into long positions (think of this as ‘reinvesting dividends’ but for market makers!)
- Take on more risk and leverage
- Quote tighter spreads
While the CME or NYSE could certainly provide this, they are reticent to do this as these tokens could easily be deemed securities by regulators. Subsequently, this is also why you're seeing these tokens boom in Asia and not the west — and they're successfully encouraging new market making participants to add liquidity to these exchanges in a way that existing exchanges cannot. Moreover, there's a feedback loop for these tokens, which is why you've seen the most successful competitors to BitMEX — Binance, FTX, Huobi, CoinFLEX — utilizing these tokens as tools to gain market share.
What can we learn from finance’s past?
This unbridled competition from exchanges resembles the heady early times of the equities markets in the late 1990s to early 2000s, when people were starting exchanges that were competitive with NYSE and NASDAQ out of their basements. At this time, the US equities market had extremely fragmented liquidity and poor data, with large data providers such as Yahoo! Finance and S&P sourcing from 100s of exchanges of varying quality, akin to CoinMarketCap. Regulators decided to improve liquidity and transparency by forcing the decentralized nature of equities exchanges to become more centralized via RegNMS.
The implementation of RegNMS in 2005 forced consolidation by requiring any stock exchange to provide the national best bid and offer (NBBO) over all other exchanges. This meant that exchanges that had very little volume would have to listen to exchanges with lots of volume and provide prices that match those high volume exchanges. Inevitably, this led to the shuttering of a number of smaller exchanges as they lost money while being forced to provide prices that deviated from their liquidity .
This boom and bust era of startup US equities exchanges ended with some of the biggest startup equities exchanges being acquired by existing exchanges and banks for billions of dollars . A natural question to ask about exchange tokens is: what will the RegNMS moment of this world look like?
Automated Market Making: The RegNMS of Crypto?
Speaking of RegNMS, the decentralized yet almost synchronous nature of US equities markets is starting to rear its head in crypto via decentralized exchanges and on-chain trading.
In this world, traders are able to non-custodially trade with smart contracts instead of centralized entities, like Coinbase, Binance, or BitMEX. Until 2019, these exchanges were historically plagued by terrible UX, expensive fees (relative to centralized competitors), costly security bugs, and riddled with valueless fee tokens. The introduction of Uniswap and constant product market making led to a dramatic shift in usability, UX, and inevitably, fees, for users of these protocols. Uniswap, which was created with a $100,000 grant from the EF, has fared much better than competitors such as Bancor (which raised $153m in 2017) because it avoided historical mistakes with automated market makers and optimized for simplicity (which leads to smaller and less complicated fees).
Let’s step back for a second: What are automated market makers? These are methods for valuing assets without needing market makers and have existed in the academic literature since the early 2000s. They were often used in early prediction markets and tended to have poor usage due to complex UX or challenges (think: Lightning/L2 game theoretic challenges).
How do AMMs work? They incentivize participants to:
- Lend their assets to a pool, from which they derive a pro-rata share of trading transaction fees. Market makers are replaced by these lenders which can earn passive income for lending their assets to the pool. One can even view the yield earned by lenders as equivalent to ‘revenue sharing with the exchange,’ which happens to be a smart contract.
- Act as an ‘oracle’, synchronizing prices of off-chain exchanges with those on-chain (e.g. incentivizing arbitrageurs to keep the ETH/DAI price on Coinbase ‘close’ to the on-chain exchange) 
The simplicity of AMMs, which starts with the absence of an order book and incentives for those who contribute tokens to the pool to keep the prices tethered to Coinbase/Binance/etc., has led to a significant amount of liquidity in on-chain vehicles. Moreover, it is easy for those working on UIs for decentralized trading protocols to integrate with Uniswap. This is how, for instance, Synthetix, a decentralized margin trading contract (which is one of many attempts at making a decentralized BitMEX) was able to bootstrap liquidity for traders of synthetic tokens. Synthetix, whose native token increased in value to be one of the largest Uniswap assets, is exemplar of how the interconnected nature of on-chain trading venues allows for new use cases to bootstrap off of existing liquidity. What is driving this increase in trading activity on-chain? It is likely a combination of the following:
- Increased KYC scrutiny (even, purportedly, at BitMEX, thanks to 5AMLD!)
- The introduction of practical AMM designs that provide reasonable (but not perfect) passive income for token lenders while also encouraging arbitrageurs prices synchronized
- Growth in on-chain lending (e.g. Compound) providing non-trivial leverage for traders
- Increased number of exchange aggregators (e.g. 1inch, DEX.ag, 0x’s new design) has proven that market forces can synchronize prices across these pools (and lower fees!) without needing order books.
- This is quite an innovation, as synchronization in normal markets is relative to NBBO (national best bid and offer), which requires an order book.
- This is sort of like a decentralized, version of RegNMS, albeit with no KYC necessary.
How will these changes affect the future of crypto trading?
Given the 2019 emergence of these two financial tools in crypto, it is natural to ask, what’s next and how will these tools change trading in the next decade? I’ve decided to break this up into four sections: Bitcoin, Ethereum, Other Layer 1, and Valuation Models of the Future.
- Lightning has always been the main form of on-chain lending activity on Bitcoin, with the lending corresponding to providing raw network value in terms of routing. I think that Lightning will make large strides in 2020, especially as exchanges such as Sparkswap and games such as Lightning Poker continue to have UX and usability improvements.
- Centralized exchange connectivity to Lightning, which began with Bitfinex’s announcement in December, will provide other arbitrage opportunities as Layer 2 cross-chain solutions will begin to have more intrinsic value (e.g. making Uniswap arbitrage opportunities more valuable).
- There are a number of projects looking to bring DeFi primitives to Bitcoin via sidechains, such as Echo and Money on Chain (built on RSK). This should provide ample opportunity for Bitcoiners who don’t want to go cross-chain to earn some yield.
- Cross-chain solutions, such as Cosmos and tBTC, will also help boost the amount of liquidity as traders can lend their tokens from one chain on another, without needed an external vendor like Genesis or Celsius.
- take: Lightning and Cross-Chain solutions will provide almost all of the on-chain lending and trading profits for Bitcoin holders, whereas sidechains will struggle to gain adoption.
- As margin trading venues begin to roll out on Ethereum, including Monte Carlo DEX (which has Bitcoin synthetics with up to 5x leverage), UMA, and Synthetix, we’re likely to see a growth in these projects that aim to use novel revenue sharing agreements (exchange tokens, pooled fees, etc.) to minimize gas usage and optimize UX.
- Ethereum lending has been growing, mainly in support of trading, and the trends for increasing these lending pools (e.g. exchange/wallet support for Compound, better UX via Dharma, etc.) and the increased need for lent tokens for derivatives should have a flywheel effect (but let’s hope that it isn’t akin to the margin that was present during the 1929 stock market crash).
- Private Ethereum transactions, for instance via Aztec Protocol’s shielded ERC-20s, will open up different styles of trading as front-running becomes less profitable and more expensive.
- Regulation of front-ends will increase. For example, Uniswap currently restricts access to their front-end webapp to users from certain countries, even though the Uniswap smart contract is running on Ethereum blockchain and is available to anyone.
- take: Layer 2 will drive a lot of the improvements in trading and lending in the foreseeable future, as opposed to Layer 1 improvements. Matter Labs is leading the charge on this and between zkRollups and StarkDEX, there will be a huge increase in the opportunity space.
- Other Layer 1s
- A lot of protocols are going to launch with on-chain exchanges, such as Celo. These protocols need these exchanges for stability (e.g. stablecoins) and/or to provide validators with on-chain liquidity during the early stages of the network.
- These networks will need to build liquidity in lending pools like those on Ethereum to help bootstrap both sides of a trade.
- Cross-chain solutions will also help bootstrap liquidity on new chains from existing ones (e.g. BTC, ETH).
- take: I’m not holding my 2020 breath for an Algorand / DFINITY / NEAR / Solana breakout trading application other than staking derivatives (which will trade lightly, as professional validation services slowly learn the lessons of Bitmain’s past).
- Valuation Models of the Future
- As the on-chain trading world matures, so will the actuarial and quantitative tools and techniques used to analyze and secure these networks.
- Front-running is already common on most Ethereum trading venues and one should expect to see more sophisticated trading strategies that take advantage of low-level transaction data dominate the marketplace. (In a recent episode of Laura Shin’s Podcast, the creator of Synthetix notes that charging higher fees for switching from long/short in the same asset has helped reduce front running.)
- This is an exciting time! Big improvements to valuation models, such as the Black-Scholes equation, often come out of improvements to market infrastructure and I suspect that 2020 will bring an ever increasing tool chest to the valuation of crypto assets.
- But hey, I’m biased… take: The value of agent-based simulation in creating the ‘Black-Scholes of crypto trading’ will be realized.
If there is nothing else that you take away from this article, just remember this: the next decade of crypto trading will look far different from the previous decade as business models and technology collectively advance. Happy New Year!
Thanks to Tony Salvatore, Matteo Leibowitz, Michael Jordan, Haseeb Qureshi, Ivan Bogatyy, Leland Lee, Joyce Yang, Rei Chiang, John Morrow, and Hasu for helpful feedback, comments, and insights.
 The story is a bit more complicated, however: smaller exchanges do have an advantage with RegNMS in that they can compete with larger exchanges by incentivizing market-makers with rebates to provide tight spreads, forcing broker-dealers to route to them because of RegNMS. Ironically, RegNMS ended up fragmenting the market structure, leading to the formation of dark pools, each with slight twists on matching rules, order types, allowed participants, etc. Thanks to Rei Chiang and Leland Lee for further discussion about this point.
 Citi bought Automated Trading Desk in 2007 for $680 million, NYSE merged with Arcahttps://www.marketwatch.com/story/nyse-to-buy-archipelago-go-public(née Archipelago) in 2006 for $10 billion, NASDAQ bought Instinethttps://www.nytimes.com/2005/04/22/business/nasdaq-to-acquire-instinet-in-19-billion-deal.html(née Island) for $1.5 billion in 2005.
 We wrote a paper that mathematically proves that Uniswap acts as an oraclehttps://arxiv.org/abs/1911.03380(up to an error dependent on the fees) if the external market is more liquid than Uniswap and there is at least one rational arbitrageur present.
© 2021 The Block Crypto, Inc. All Rights Reserved. This article is provided for informational purposes only. It is not offered or intended to be used as legal, tax, investment, financial, or other advice.