The role of AI in cryptocurrency

• Cryptocurrencies enable users to perform commercial

operations across the globe

• This has reduced dependencies across the value chain

• Cryptocurrencies such as Bitcoin, Ethereum, and Ripple have

enabled a large number of people to disintermediate

• AI has been utilized in the Banking, Financial Services, and

Insurance industry.

• The convergence of both technologies can be mutually

beneficial to AI and to the cryptocurrencies.
 The role of AI in cryptocurrency
Fiat currency Cryptocurrency
Minted by the government under the Minted by the miners or validator nodes
Fractional reserve banking under a voted scheme
Not all fiat currencies are in digital form, Cryptocurrencies are digital, transparent
because difficult to trace exact taxes transactions, ideal tax compliance.
Regulated by a national-level federal bank Cryptocurrencies are decentralized,
Hence, it is challenging to regulate them
Money transfers and settlements may Monetary transfers are near-immediate
take days
Digital money may not be highly secure, Hackers may not be able to hack all the
due to centralization and cyber threats nodes
May not be highly volatile Highly volatile compared to traditional
currencies and money markets, due to
speculation
 Cryptocurrency trading
• Trading cryptocurrencies has become a global alternative to
wealth creation

• Ranging from small crypto exchanges to billion-dollar funds,

offer wealth management services both directly and indirectly

• 2,000 cryptocurrencies floating on multiple blockchains

• Cryptocurrencies crossing 500 billion US Dollars (USD) in 2018.

• Cryptocurrency is an alternative wealth creation medium in

the long term.
Cryptocurrency trading
 Issues and special considerations
Even impressive growth in wealth creation using cryptocurrency
trading, the market is still inefficient due to the following reasons:

• Difficulty determining sharp volatilities in the market

• Lack of an intelligent order-matching engine that allows users to

estimate the right values for their holdings

• Inability to calculate predictable risks due to mixed responses in

the market

• Weak cybersecurity strategy in protecting wallets from attacks.

 Benefits of AI in crypto trading
• AI techniques such as Machine learning (ML) and Deep Learning
(DL) based on quantitative and non-quantitative

• Sharper ability to detect anomalies in trading trends

• React quickly without human intervention to critical situations

using trading bots

• Establish an aggregated signal that represents the emotions of

buyers and sellers.

• Bring remarkable security enforcements in password

generation, detecting physical characteristics such as iris, retina,
and other biometric identities
 Making price predictions with AI
• Cryptocurrency market is efficient if the growth of market
capitalization increases gradually over time

• More attention needed in making margins has always been

price prediction

• Predictive analytical concepts using for money making

through algorithmic trading

• cryptocurrency price has increased or decreased based on

dynamics such as news, announcements, and also any
regulatory actions made to a specific cryptocurrency
 Issues with price prediction
• Top two issues with regard to price prediction of cryptocurrencies
are:
– Unable to adjust the algorithms to crypto market sentiments
– Unable to maintain and measure liquidity

• Growing application of AI techniques such as neural networks in

predicting the price of a publicly traded cryptocurrency

• Crypto markets are highly volatile, with various attributes such as

– technical advancements,
– announcements, and
– emerging competition
affecting the price with relatively higher sensitivity

• So ML may not be effective in these scenarios

 Issues with price prediction
• Application of an artificial Recurrent Neural Network (RNN) is called Long
Short-Term Memory (LSTM)

• A common LSTM unit is composed of a cell, an input gate, an output gate and
a forget gate.

• The cell remembers values over arbitrary time intervals and the three gates
regulate the flow of information into and out of the cell.

• Forget gates decide what information to discard from a previous state by

assigning a previous state, compared to a current input,

• Input gates decide which pieces of new information to store in the current
state, using the same system as forget gates.

• Output gates control which pieces of information in the current state to

output
Issues with price prediction
 Issues with price prediction
• With the LSTM approach, the cryptocurrency trading data is grouped
under basic common attributes

• Open Price refers to the price of the cryptocurrency at the beginning of

the day.

• Close Price refers to the price of the cryptocurrency at the end of the
day.

• High Price refers to the highest value recorded versus the USD on that
particular day.

• Low Price refers to the lowest value recorded versus the USD on that
particular day.

• Volume refers to the total number of units traded on that particular day
Actual versus predicted prices by applying the LSTM model
 Introduction to time series

Time series information of Bitcoin price versus USD in logarithmic scale

 Time-series forecasting with ARIMA
• cryptocurrency prices are affected by various factors, it is not
easy to analyze and predict prices using simple ML models.

• Hence, the ARIMA model is used to predict the price.

• ARIMA is widely used so that we can

– apply the model,
– make a note of the error in prediction,
– retest the model after making the necessary changes,
– and calculate the MSE again.

• ARIMA models provide reasonable price prediction with an

error rate ranging from 3% to 5%
Time-series forecasting with ARIMA

Bitcoin price prediction using ARIMA

 Applications of algorithmic or quant trading in
cryptocurrency

• Significant number of trade calls happening around the world

are not made by humans

• They are programmed and automated to a certain extent by

operations

• This type of trading involves a keen eye on profits over a large

number of calls, resulting in good profits in a short term

• The trade calls made by these machines are optimized to look

at profits that humans cannot sense by reading charts and
reports.
 Arbitrage
• Arbitrage refers to the practice of leveraging different prices
of the same asset in two different locations

• In order to profit from the difference in the economic values

of the asset in trade transactions

• Exploits the different economic values of the same asset in

two or more markets

• In 650 BC buy silver coins at a cheaper price in Persia and sell

them at a premium in Greece

• Bitcoin (BTC)—is listed across more than 100 exchanges

 Arbitrage

USD price of one Bitcoin across 10 different exchanges

 Arbitrage
• Algorithms have been developed to enable instantaneous
arbitrage transactions between two kinds of cryptocurrency.

• Assume that 1 BTC costs 50 Ethers (ETH) in exchange 1 and 60

ETH in exchange 2.

• Algorithmic trading is the best-fitting cryptocurrency trade

pairs that can return maximum profits and execute the trade

• Its profiting 10 ETH in less than a few seconds.

• If such a trade can be achieved successfully, a crypto/crypto

arbitrage has been successfully established
 San Francisco Open Exchange (SFOX)
• YCombinator-backed trading platform

• It has been serving more than 175,000 traders since 2015

• The platform has a gross transaction value of nearly 11 billion

USD over the course of years

• This exploring arbitrage opportunities for its customers across

more than 20 markets

• SFOX offers industry-leading algorithmic trading

 San Francisco Open Exchange (SFOX)
• Tortoise: An optimized order routing suitable for trading lower
numbers of bitcoins, but comparatively slower than the rest of the
algorithms.

• Hare: Better price goals than Tortoise without any compromise in

speed.

• Gorilla: Optimized to execute large trade orders with suitable controls

on market fluctuations to make sure that the market does not move.

• Polar Bear: Optimized to execute hidden orders optimized for price,

triggered by the best limit set by the user. Once the algorithm
identifies the best price in the order book, it will immediately execute
the order at that price without going too deep into the order book.
 Market making
• Market making is having the important goal of offering liquidity
to cryptocurrencies in the market

• Buying parties in the markets place a bid (the bid) for

purchasing a particular crypto asset.

• A seller with an intent to sell the same type of crypto asset may
place their asking price (the ask) for the asset.

• Buyer and seller do not match because the buyers usually quote
for less value and the sellers quote for more value.

• This can create a gap between the expectations between both

parties, thereby creating a spread
 Market making
• When the disagreement on the mutual price grows, the
spread value widens and creates an illiquid token or
cryptocurrency.

• Illiquid tokens are basically non-tradeable since expectations

are not matched.

• Hence, lower trade volumes create a slowdown for the token

and reduce the market capitalization of the respective token.

• Hence, there is a need for liquidity providers who can help

both buyers and sellers meet in the middle.
Market making
 Benefits of AI in trading data
• The application of AI on semi-structured data from blockchain
transactions can pave the way for better analysis

• It provides deeper insights into making efficient trade

decisions.

• Real-time high throughput from multiple blockchains can be

analyzed by applying AI models

• This helps to make informed decisions across different

blockchain realms
 The future of cryptocurrencies in India
• India has long been one of the few developing countries with
an uncertain position on cryptocurrencies.

• A circular from the Reserve Bank of India (RBI) dated February

12, 2018 advised all the banks and RBI-regulated institutions
against supporting any individual or entities involved in virtual
currencies

• This was considered as a blanket ban on all activities

pertaining to cryptocurrencies, although exchanges in India
were unable to support deposits or withdrawals associated
with bank accounts
 The future of cryptocurrencies in India
• On March 6, 2020, the Honorable Supreme Court of India's
verdict quashed the RBI circular, thereby striking down RBI's
ban on financial institutions rendering services to these
exchanges and traders alike

• At the time of writing, RBI has reportedly filed a review

petition to challenge the verdict given by the country's Apex
court.

• It is also notable that Anti-Money Laundering (AML) and Know

Your Customer (KYC) regulations are to be updated in order to
support these businesses without harming the country's
economy