Introduction
Arbitrage is the practice of ‘buying something in one place and selling it in another where the price is higher’.
Not only is arbitrage a fascinating tool to master market structures and trade execution, but it is also a captivating instrument to understand market behaviours.
We have been arbitraging ten crypto markets for more than a year. Seven of them are well known enough to be listed in the top twenty of the CryptoCompare Exchange Benchmark.
Utilising over one billion lines of market data Suiteki has recorded, we present what we perceive to be a relevant and fair representation of the arbitrage opportunity. Analysing the breakdown of opportunities by pairs, duration, and market types (centralised vs decentralised), we attempt to remove the mirage and present the real opportunities at stake.
To whom may enter this market, or who may enjoy a deep dive into piranha-infested waters, here are our humble findings and takeaways:
1- The daily $10,000
2- The healthy $7,000
3- The Triple 13 Club
4- The Veterans and the 70%
5- No fiat’s land and BTC King
6- The 1 second and 60% rule
7- The reward of asynchronicities
Before we go through the takeaways in detail, let us give you the scope of the analysis and our approach with the data.
Scope
The analysis has been done on a 5-month time frame, from 01–09–2019 to 31–01–2020.
Our solution covers today 2 types of spot exchange and order matching methodology/ tech (cf. below).
Our algorithmic solution is made of a collection of services. Each service aims at achieving a specific task within the lifecycle of an arbitrage. We have 7 main services:
1- Orderbook fetcher: collect real-time data from markets
2- Arbitrage finder: find theoretical arbitrage opportunities
3- Analyser: analyse profitability and check availability of funds
4- Executer: create order with the right attributes (price, quantity, etc.)
5- Risk manager: identify, replace and close any open position at risk
6- Closer: check the status of an arbitrage and release funds
7- Balance manager: store and keep track of balance modifications
As you guess, the goal here is not to assess how well Suiteki executes and manages certain risks (3 to 7), but rather to assess the market opportunity (1 and 2).
When the solution is up (almost 90% of the time), service 1 and 2 collect every price difference, between two exchanges, for the same asset. This record takes the form of a theoretical arbitrage with many attributes such as: buy platform, sell platform, buy price, sell price, timestamp, buy currency, sell currency, etc.
This analysis is based on those recordings.
Data Approach
Assessing the opportunity of a tireless economic practice, on a maturing industry, is like going out on a limb. Many different approaches and quantitative methods might exist in the context of this data analysis.
Here is our approach in 3 points.
1- Keep it simple
First, we used flat rates (1) for the period and converted every theoretical arbitrage profit to USD. Second, we discarded every arbitrage opportunity smaller than $1 in profit.
2- Common sense
As we record arbitrage opportunities every millisecond, when the estimated profit is the same for one trading pair between exchange A and B, we considered that as one and unique opportunity.
To put it simply, we did not count twice what seemed to be a unique opportunity.
3- Theory and practice
In theory, you enjoy superpowers as you have a perfect control of matter, space, and time. In terms of crypto arbitrage, that means:
- Matter: you have access to an infinite supply/ amount of capital
- Space: you can transfer instantly digital assets from one exchange to another
- Time: you execute instantly and see the future
In practice, the story is different as you have multiple constraints.
Although Flash can certainly execute an arbitrage every millisecond, we assumed that we can execute only one arbitrage every hour, between two exchanges, for the same crypto pair. But we can execute in one hour as many arbitrages as available crypto pairs between two exchanges.
In addition to the hour interval, we added the notion of cluster. A cluster is defined as a succession of arbitrages between two exchanges, for a same crypto pair. If we had a cluster of arbitrages, we assumed that the theoretical arbitrage opportunity equals to the maximum profit available on this cluster.
Neither too conservative nor too optimistic, we believe that those assumptions provide a fair representation of the arbitrage opportunity on our scope.
Takeaways
1- The daily $10,000
The total arbitrage opportunity amounts to $1.5M, from 01–09–2019 to 31–01–2020. That estimate gives an approximately $10,667 arbitrage opportunity per day, which gives roughly a $4M annualised opportunity.
Note: This table shows the distribution of arbitrage opportunities and other metrics.
Based on our assumptions and considering that, (i) you have the right asset inventory, and that, (ii) this portfolio is at the right place on both legs of the arbitrage before the opportunity, you would theoretically need $2.6M of initial investment to capture the $4M annualised arbitrage opportunity. That would yield a 147.825% annualised return. Yum-yum!
However, such a conclusion must be nuanced. Two questions must be asked behind every theoretical arbitrage.
- Is it a trap?
If you are a fanatic of the Efficient-Market Hypothesis (EMH), that stakes that asset prices reflect all available information, you will wonder: is what appears to be mispriced really so?
Example 1: at the time of writing, you can buy 1 BTC at €8,008 on Bitstamp, and sell 1 BTC at €8,030 on Binance. If you take this arbitrage, you will end with €8,030 on Binance, capturing €22 of net profit. Let us say that you have the same opportunity the following day, but no Euro on Bitstamp. You will then decide to withdraw the €8,030 from Binance to your bank account and re attempt the arbitrage. If you have a business account, you can use Advcash service on Binance. However, a bank transfer from your Advcash wallet to your bank will cost you €10 plus 0.5% fee. Therefore, you will end up with €7,979.85 in your bank account. You have wiped out your profit and lost money. It’s a trap.
Example 2: at the time of writing, you can buy 1 GRIN :) at $0.429 on Poloniex, and sell 1 GRIN at $0.496 on Bittrex, capturing almost 16% of net profit! However, the GRIN wallet is under maintenance on Bittrex. You can’t deposit 1 GRIN on Bittrex in the perspective of taking this arbitrage opportunity. It’s a trap.
Unless you have solved the problems of example 1 and 2, respectively the bank transfer and the wallet maintenance, the theoretical opportunities are impracticable.
The information that you have a trap may be contained in the price.
Recurrent and very profitable theoretical arbitrages are often ‘too good to be true’.
- Is the asset hedgeable?
Arbitrage strategies aim to rake profits with very limited market risks. If a client gives you $100 to arbitrage with, you will need to create the best arbitrageable portfolio. Having $50 on exchange 1 and $25 worth of BTC on exchange 2 is a good start. Note that you will use $25 to hedge the $25 worth of BTC you bought.
GRIN is a good example of a crypto that is not easily hedgeable. You will not be able to have it in your arbitrage portfolio, unless your client agrees to be exposed to its price movements. Good luck! That’s not arbitrage but rather speculation plus arbitrage.
What would be “The daily $10,000” considering the above?
2- The healthy $7,000
Removing the potential traps and non-hedgeable assets lead to a new daily opportunity of approximately $7,000, shrinking the initial global opportunity by 30%.
The new total arbitrage opportunity amounts to $1.1M, from 01–09–2019 to 31–01–2020. That estimate gives roughly a $2.6M annualised opportunity.
Note: What we consider as non-hedgeable assets (2), at least not easily, and hedgeable assets (3).
3- The Triple 13 Club
26% of the above opportunity rely on arbitrages in BTC and USD:
- 13% in BTC and USD
- 13% in BTC and USD-denominated stablecoins
Our scope includes data from the following stablecoins: Tether (USDT), trueUSD (TUSD), paxos standard (PAX), and USD coin (USDC).
Therefore, approximately $675,000 of the annualised opportunity is concentrated on the above possibilities and trading pairs: BTC:USD, BTC:USDT, BTC:TUSD, BTC:PAX and BTC:USDC.
Adding EUR and GBP fiats on top of that and you have the triple 13 club!
- 13% in BTC and USD
- 13% in BTC and USD-denominated stablecoins
- 13% in BTC and EUR/GBP
The triple 13 club represents approximately $1M of the annualised arbitrage opportunity, 39% of the reviewed opportunity.
Although BTC dominates the total crypto market capitalisation by more than 60%, other cryptos occupy a significant place in this market.
4- The Veterans and the 70%
We consider the veterans as the following crypto assets: BCH, BTC, ETH, LTC, and XRP.
The veterans’ opportunity represents 70%+ of the reviewed opportunity. That gives approximately a $5,000 arbitrage bounty per day, which is roughly a $1.8M annualised opportunity.
In other words, the trading pairs involving the veterans, are to the crypto spot arbitrage world (at least on our scope), what Alexa is to the US smart speaker market, a 70% market share!
Obviously, that would assume that you have in your arbitrage portfolio all the quote/fiat currencies (USD, EUR, USDT, BTC, etc.) that can be traded with the veterans.
Although fiat currencies seem to be essential, what would be the theoretical arbitrage opportunity with no fiat in your porfolio?
5- No fiat’s land and BTC King
A bit less than 30% of the healthy $2.6M annualised opportunity are crypto-to-crypto arbitrage opportunities (stablecoins excluded). A daily $2,000 arbitrage opportunity where BTC as a quote/fiat currency accounts for 94% of the opportunity! The missing 6% representing XXX:ETH type of trading pairs, where ETH is the quote/fiat currency.
Note that based on our dataset, Asian-based exchanges are involved nine times out of ten in the top ten crypto-to-crypto arbitrage profits.
That might suggest that crypto-to-crypto arbitrage opportunities are initiated by a price change in Asian-based exchanges.
More data could potentially emphasise (or not) the assumption that Asian-based exchanges are the centre for crypto-to-crypto price discovery.
6- The 1 second and 60% rule
Basic arbitrage strategies take liquidity. Like Pac-man eats ghosts, you eat existing orders, lifting the offer somewhere and hitting the bid somewhere else. That’s it. Although not entirely true in every case (depending on the order matching methodology/ tech, cf. above), how fast you take those orders will often determine how successful you are.
In this type of strategy, the winner takes it all and the looser risks it all. What makes the success of the strategy is often within the description itself: low-latency arbitrage strategy.
As an example in the traditional world, ten years ago, such a reality has led Spread Network to connect Chicago and New York by fiber optic (in a nearly straight line) for an estimated cost of $300M. Round-trip communication between both cities was reduced from 16 milliseconds to 13 milliseconds…
To put things into perspective, the blink of a human eye lasts 400 milliseconds. Ten times faster is our instinctive reaction when we face an immediate danger. This instinctive reaction operates within 40 milliseconds.
As mentioned by James Lovelock in Novacene: “As Einstein said, ‘The intuitive mind is a sacred gift and the rational mind is a faithful servant. We have created a society that honours the servant and has forgotten the gift.’ ”
But I digress.
Crypto arbitrageurs are far from the 3-millisecond battle. Crypto trading is a web-based industry that depends on the reliability of cloud-based services. As highlighted by an article on CoinDesk, ‘Web-based technology puts the emphasis on serving the greatest number of users concurrently, not on serving a subset of users deterministically and at the lowest latency possible.’
Our dataset shows (cf. below) that 60% of the arbitrage opportunities, considering the Triple 13 Club or The Veterans, last less than a second. Less than a second in web standards can be considered as relatively fast. 40% of the opportunities last less than approximately 500 milliseconds. Finally, 20% of the opportunities last more than 4 seconds.
7- The reward of asynchronicities
Arbitraging a centralised exchange with a decentralised exchange means that you have cracked a problem of asynchronous order execution between both.
Our solution is connected to a well-known public ledger. At the time of writing, the ledger interval is nearly 4 seconds. In other words, every new validated block of transactions is added every 4 seconds to the ledger. If you are executing a specific peer-to-peer transaction, being in the 4-second interval doesn’t require latency investments. You’re not competing on time but rather on other criteria such as transaction fee.
The Centralised-Decentralised arbitrage opportunity amounts to almost $117,000, from 01–09–2019 to 31–01–2020. That estimate gives roughly a $281,000 annualised opportunity, almost 11% of the healthy $2.6M annualised opportunity.
Conclusion
Comparing the crypto spot arbitrage opportunity, at least on our scope of analysis, with the E-mini S&P 500 vs the SPDR S&P 500 arbitrage opportunity, is almost like comparing the size of the UK (242,495 km²) with the size of Australia (7,692M km²). If we have had the madness of February and March in our scope, the gap would have been reduced. Both months are the reflection of exceptional market events and could be the object of a future dedicated Medium.
According to this research study, the E-mini vs SPDR race has averaged $75M in arbitrage opportunity per year, between 2005 and 2011, almost 30 times more than the $2.8M identified prize on our scope. And that’s only one trading pair between two markets!
Although latency is key, we believe that the prisoner’s dilemma on the topic is not as acute as in traditional finance. In the crypto world, to be relatively fast can be sufficient or acceptable. For instance, the DeFi ecosystem offers new anti high-frequency market designs, that are very interesting in terms of arbitrage mechanisms. Moreover, arbitraging crypto is also about having the best arbitrageable portfolio. Analysis of recent data might gave us some appreciable clues and patterns.
Based on the above assumption, we believe that crypto arbitragers are pushed to be more holistic in their arbitrage approach.
In a market where more than 30% of the daily volume take place on derivative instruments, according to a recent CryptoCompare report, we think that a basic spot arbitrage strategy must be expanded on that segment of the market.
Finally, in latin, arbitrage also means arbiter, which means to judge. Isn’t it all about to judge what’s the best option available at a certain time?
By Didier Pironi
About us
Suiteki develops and runs crypto arbitrage solutions. Since its incorporation on 28 June 2018, Suiteki has raised around €300,000 in seed funding.
Any interest in what we do, please let us know!
https://www.linkedin.com/company/suiteki
Acknowledgments
Thanks to my partners in crime, Arthur and Aki, for contributing to this post!
Disclaimer
Suiteki does not endorse or promote any of the projects or cryptocurrencies mentioned in this blogpost. Any descriptions of functionality and services provided are for information only. Suiteki is not responsible for any loss of funds or other damages caused as a result of using the projects described above.
References
(1) {‘ADA’: 0.08, ‘ADX’: 0.1, ‘AE’: 0.14, ‘ARDR’: 0.05, ‘ARK’: 0.2, ‘ATOM’: 2.5, ‘BAT’: 0.2, ‘BCD’: 0.6, ‘BCH’: 200, ‘BCHSV’: 175, ‘BCPT’: 0.02, ‘BNT’: 0.8, ‘BTC’: 8000, ‘BTG’: 8, ‘BTS’: 0.02, ‘CMT’: 0.01, ‘CND’: 0.01, ‘CVC’: 0.03, ‘DASH’: 70, ‘DCR’: 15, ‘DGB’: 0.01, ‘DGD’: 40, ‘DNT’: 0.01, ‘DOGE’: 0.001, ‘EDO’: 0.6, ‘ELF’: 0.1, ‘ENG’: 0.3, ‘ENJ’: 0.2, ‘EOS’: 2.5, ‘ETC’: 6, ‘ETH’: 250, ‘EUR’: 1.1, ‘FCT’: 1.7, ‘FUN’: 0.004, ‘GAME’: 0.9, ‘GAS’: 1.7, ‘GBP’: 1.25, ‘GNT’: 0.05, ‘GO’: 0.015, ‘GRIN’: 0.5, ‘GRS’: 0.2, ‘GTO’: 0.01, ‘HC’: 1.2, ‘ICX’: 0.3, ‘INS’: 0.16, ‘IOST’: 0.005, ‘IOTA’: 0.2, ‘IOTX’: 0.005, ‘KMD’: 0.7, ‘KNC’: 1.2, ‘LBC’: 0.03,’LINK’: 4, ‘LOOM’: 0.02, ‘LRC’: 0.1, ‘LSK’: 1.2, ‘LTC’: 40, ‘MAID’: 0.1, ‘MANA’: 0.04, ‘MCO’: 5, ‘MITH’: 0.006, ‘MTL’: 0.3, ‘NANO’: 1.15, ‘NAS’: 0.3, ‘NAV’: 0.15, ‘NEO’: 10, ‘NMR’: 20, ‘NPXS’: 0.0001, ‘NULS’: 0.4, ‘NXS’: 0.2, ‘NXT’: 0.01, ‘OMG’: 1.5, ‘ONT’: 0.5, ‘OST’: 0.01, ‘PAX’: 1, ‘PIVX’: 0.5, ‘POLY’: 0.05, ‘POWR’: 0.1,’QTUM’: 1.6, ‘RCN’: 0.07, ‘REP’: 17, ‘RLC’: 0.6, ‘RVN’: 0.02, ‘SC’: 0.004, ‘SNT’: 0.02, ‘STEEM’: 0.2, ‘STORJ’: 0.15, ‘STORM’: 0.002, ‘STRAT’: 0.5, ‘SYS’: 0.03, ‘THETA’: 0.2, ‘TRX’: 0.015, ‘TUSD’: 1, ‘USD’: 1, ‘USDC’: 1, ‘USDT’: 1, ‘VIA’: 0.2, ‘VIB’: 0.015, ‘VTC’: 0.3, ‘WAVES’: 1.2, ‘WTC’: 0.3, ‘XEM’: 0.04, ‘XLM’: 0.07, ‘XMR’: 65, ‘XRP’: 0.2, ‘XVG’: 0.007, ‘XZC’: 4, ‘YOYO’: 0.007, ‘ZEC’: 50, ‘ZEN’: 7, ‘ZIL’: 0.02, ‘ZRX’: 0.3}
(2) {‘ADX’, ‘AEB’, ‘AEE’, ‘ARDR’, ‘ARK’, ‘BCD’, ‘BCPT’, ‘BNT’, ‘BTG’, ‘BTS’, ‘CMT’, ‘CND’, ‘CVC’, ‘DCR’, ‘DGB’, ‘DGD’, ‘DNT’, ‘DOGE’, ‘EDO’, ‘ELF’, ‘ENG’, ‘ENJ’, ‘FCT’, ‘FUN’, ‘GAME’, ‘GAS’, ‘GNT’, ‘GOB’, ‘GRIN’, ‘GRS’, ‘GTO’, ‘HCB’, ‘HCE’, ‘ICX’, ‘INS’, ‘IOTX’, ‘KMD’, ‘KNC’, ‘LBC’, ‘LOOM’, ‘LRC’, ‘LSK’, ‘MAID’, ‘MANA’, ‘MCO’, ‘MITH’, ‘MTL’, ‘NANO’, ‘NAS’, ‘NAV’, ‘NMR’, ‘NPXS’, ‘NULS’, ‘NXS’, ‘NXT’, ‘OMG’, ‘OST’, ‘PIVX’, ‘POLY’, ‘POWR’, ‘RCN’, ‘REP’, ‘RLC’, ‘RVN’, ‘SCB’, ‘SCE’, ‘SCU’, ‘SNT’, ‘STEE’, ‘STOR’, ‘STRA’, ‘SYS’, ‘THET’, ‘VIA’, ‘VIB’, ‘VTC’, ‘WAVE’, ‘WTC’, ‘XEM’, ‘XVG’, ‘XZC’, ‘YOYO’, ‘ZEN’, ‘ZIL,’ ‘ZRX’}
(3) {‘ADA’, ‘ATOM’, ‘BAT’, ‘BCH’, ‘BCHSV’, ‘BTC’, ‘DASH’, ‘EOS’, ‘ETC’, ‘ETH’, ‘EUR’, ‘IOST’, ‘IOTA’, ‘LINK’, ‘LTC’, ‘NEO’, ‘ONT’, ‘PAX’, ‘QTUM’, ‘TRX’, ‘TUSD’, ‘XLM’, ‘XMR’, ‘XRP’, ‘ZEC’}
