Realized Volatility Forecasting in the Derivatives Market with
Multivariate LSTM: A peek into BTC-USD
1.1, Motivation
Assets with high volatility, such as commodities like oil and gold, alongside emerging
financial instruments, with a decentralized nature and minimal regulation, like
Bitcoin, have created distinct challenges and opportunities in financial markets.
Therefore, accurately forecasting their volatility becomes crucial for informed
decision-making, particularly in the derivative markets of these assets where
volatility plays a key role in the pricing of options, futures, and other contracts.
Volatility forecasting is a cornerstone of the derivatives market, playing a vital role in
enhancing risk management, refining trading strategies, and optimizing portfolios.
Traditional methods, such as GARCH models or implied volatility, often fall short in
capturing the intricate, non-linear relationships that define financial data. This
limitation underscores the need for innovative approaches.
With the advent of deep learning, Long Short-Term Memory (LSTM) networks have
emerged as a good candidate for time series analysis. By incorporating such models,
we can explore the interactions between various market factors, with a view to create
more precise and reliable forecasts.
The purpose of our project is to leverage the utilization of LSTM to forecast the next
7 days’ average realized volatility of BTC-USD. By diving into the intricate patterns
of market behavior, we aim to take a forward-looking approach, offering insights that
could enhance decision-making in the derivatives market. Through this endeavor, we
seek, not only to improve financial forecasting ,but also to demonstrate the practical
impact of machine learning in quantitative finance.
1.2, Derivatives and volatility forecasting
Derivatives are sophisticated financial instruments whose value is derived from an
underlying asset, such as stocks, commodities, currencies, or indices. These
instruments include options, futures, and swaps, and they are pivotal tools in modern
financial markets. They serve multiple purposes: hedging against potential losses,
leveraging speculative opportunities to profit from price movements, and managing
complex portfolios. However, the profitability of trading in derivatives is not solely
dependent on market trends but also heavily influenced by the accurate assessment
and forecasting of market volatility, a critical factor in pricing and strategy
development.
In our project, we will take an attempt to forecast the volatility of BTC-USD and
AAPL using LSTM. Therefore, there are the basics we need to know beforehand, in
order to get a grasp of how an accurate projection of Volatility helps us in
outperforming the market:
1.2.1, Introduction to short-long positions
Long Position: We buy it in hope that the assets’ prices will increase in
the future
• Definition: Buying an asset to profit from a price increase.
• Example: Buy at $50 now, sell at $70 in the future; profit = $20.
• Risk: Limited to the amount invested (if the asset's price drops to $0).
Short Position: We buy it in hope that the assets’ prices will decrease in
the future
• Definition: Selling a borrowed asset to profit from a price decrease.
• Example: Sell at $100 now, buy back at $70 in the future; profit =
$30.
• Risk: Potential losses are unlimited (if the asset's price rises).
2, What are Realized Volatility(RV) and Implied Volatility(IV)
Volatility attempts to measure magnitude of price movements that a
financial instrument experiences over a certain period of time. The more
dramatic the price swings are in that instrument, the higher the level of
volatility, and vice versa.
Realized Volatility (RV): Price changes that really happened, and traders
will try to predict this.
• Definition: The actual historical volatility of an asset over a
specific period, based on observed price movements.
Implied Volatility (IV): Prices changes expected by the market (the
masses) using models such as Black Scholes, not always accurate.
• Definition: The market's expectation of the asset's future volatility,
derived from option prices.
• The premium for an option contract is heavily affected by this index.
3, How short/long positions are decided by volatility to capitalize on
options and futures markets
As we already acknowledged, IV reflects the current view of the market of
the future, and determines the premium fee of the option, heavily affecting
out options prices. RV is what will happen, and we will attempt to predict
this.
High IV: Options’ price will be higher for derivative contracts.
Low IV: Options’ price will be lower for derivative contracts, as there will
be low fluctuations, therefore less risk.
Our job is to predict a model that carefully and accurately predicts the
volatility of the asset prices better than IV, therefore effectively
outperforming the market. There are several cases in which we can win:
Case 1: RV > IV (Market Underpricing Volatility)
The market is underestimating future volatility, making options cheap
relative to expected price movements.
• Scenario:
• Realized Volatility (RV) is forecast by our models to be 25%,
but the market's Implied Volatility (IV) predicts only a 15%.
• Therefore, IV heavily determines the price options to be
cheaper than the actual price.
• Strategy:
• Get into the Long position and exit when maximum profit is
determined.
• Avoid getting into Short position because the option’s price is
higher in the future, rendering a loss for the trader when he is
obliged to buy back the options.
Case 2: RV < IV (Market Overpricing Volatility)
The market is overestimating future volatility, making options
expensive now, but will be cheaper in the future.
• Scenario:
• Implied Volatility (IV) is 25%, but your model predicts
Realized Volatility (RV) to be only 15%.
• Therefore, IV heavily determines the price options to be higher
than the actual price.
• Strategy:
• Get into the Short position to profit from the market’s
overpricing.
• Determine the optimal time to buy back the options within
expiration period, ideally when IV falls closer to RV.
• Avoid getting into Long positions, because the profits are very
likely to be minimized and cannot cover market fees.
Specifically, according to Sinclair (2020), a number of trading strategies
evolve around identifying situations where this volatility mismatch occurs:
///////////////////////////
P/L = Vega*abs(sigma(implied)-sigma(realized))
//////////////////////////
in which:
• Vega is the measurement of an option's price sensitivity to changes in the
volatility of the underlying asset.
• Sigma(implied) and sigma(realized) are IV and RV, respectively.
=> If traders can produce accurate predictions about the upcoming
state of the market (predicting accurate RV) better than the masses
(shown by IV), they can successfully develop a trading strategy which
will profit them significantly.
