NEXUM TRADING | Devpost

Concept

Everything in our world is interconnected it is a rule of nature, history, and the markets. "Nexum Trading" philosophy is to use the interconnection between global traditional financial markets and cryptocurrency markets as a basement for algorithmic trading.

"Nexum" is a Latin word meaning "Connection".

Assumptions

• Cryptocurrencies prices are led by the changes in equities prices, therefore there should be a time lag. • Correlation between equities and crypto is highly different in different market conditions. • The highest correlation occurs during periods of high volatility periods in equities, but not depend on the cryptocurrency volatility. • Crypto prices can change not affecting equities, but any significant change in equities price will affect crypto prices.

Research steps

• Data collection. Download E-mini-S&P 500 (ES) futures 1-minute candlestick data and Binance ADAUSDT perpetual futures data candlestick as an example. • Calculate each minute price change for both assets. • Calculate standard deviation as volatility measurement for the S&P 500. • Calculate the correlation between changes in prices for the different volatility levels. • Make an output with the conclusion. • Define, what volatility levels are responsible for the highest correlation levels. • Derive trading logic for the high volatility periods from the previous research steps. • Setup backtest environment • Test various inputs for the trading logic

According to the correlation results, we can see that • Our hypothesis is accepted. In the 2021 period, one can see high correlation between ADA and SPY, also considering the certain periods of high volatility we could see a chance to use the bot and profit. With more research we would be able to get more interesting and robust results

Following steps

• Figuring out the enter/exit volatility threshold, either from empirical evidence or own’s ML models

Our achievements

• Research relatively new investment concepts • Data collection and editing (i.e matching 24/7 markets data with US equities etc.) • First hackathon for some of us • Using each one’s specialty and skills

Our mistakes

• Underestimating the nuances of strategy and the time to clean the dataset, and implement the code (or in general, time management) • Not backtesting the strategy enough • We need to define tasks much better (ambiguous task splitting) • Not generating enough valid questions