Abstract

During the initial outbreak of the COVID March 2020 the safety that the 60-40 stock-bonds portfolio offered seemed to break down, leading investors to seek new uncorrelated assets to hedge portfolios in times of crisis. This micro-study aims to determine the new 60-40 portfolio, as the interest from idle cash starts to diminish. It uses machine learning to select and weight portfolio assets based on the magnitude of the predicted returns. The strategy uses machine learning and economic factors to manage a portfolio of risk-on and risk-off assets. The algorithm rebalances the portfolio at the start of every month. During each rebalance, it allocates a portion of the portfolio to each asset the regression model predicts will have a positive return over the following month, scaling the positions based on the magnitude of the predicted returns.

Abstract

This research explores Bitcoin's role as a leading indicator for US Equity market turbulence. Bitcoin, classically a risk-on asset that trades 24/7, can signal crises in other markets due to its liquidity and volatility. The study demonstrates a trading strategy using the LEAN engine, rotating capital between US Equities and cash based on Bitcoin's price action. When Bitcoin drops two standard deviations below its two-year moving average, the strategy shifts to cash, enhancing risk-adjusted returns for long-term investors.

Abstract

This research explores a trading algorithm that mimics trades made by U.S. Congress members, leveraging their privileged access to market-moving information. The Stop Trading on Congressional Knowledge (STOCK) Act mandates disclosure of such trades, enabling public access. Using the Quiver Quantitative dataset, the algorithm employs an inverse-volatility weighting scheme to balance risk across assets, limiting individual asset exposure to 10% to mitigate concentration risk. By forming a portfolio based on these disclosures, the strategy aims to capitalize on the informational advantage indirectly.

Abstract

The Wheel strategy is a popular options trading approach that generates steady income from equities intended for long-term holding. It involves selling cash-secured puts and covered calls. Initially, out-of-the-money (OTM) puts are sold until shares are assigned. Once shares are held, OTM covered calls are sold until exercised. This strategy generates income through premiums from option sales. The underlying equity should be one the trader is comfortable owning. For implementation, SPY was used as the underlying asset, chosen for its stability and long-term hold potential. The strategy offers built-in risk management and downside protection by effectively managing option assignments and sales.

Abstract

The Probabilistic Sharpe Ratio (PSR) is a method for evaluating investment performance that takes into account the non-normality of returns. The traditional Sharpe ratio assumes that returns are normally distributed, which can lead to misleading results for strategies with non-normal returns. The PSR addresses this limitation by considering the distribution of returns and estimating the probability that a given Sharpe ratio is a result of skill rather than luck. This provides a more accurate measure of a strategy's performance and allows for better comparisons between different strategies. The PSR is particularly useful for strategies with non-normal returns, as it takes into account the impact of skewness and kurtosis on the statistical significance of the observed Sharpe ratio.

Abstract

Abstract: This tutorial explores a sector rotation strategy based on news sentiment using the LEAN algorithmic trading engine and datasets from the QuantConnect Dataset Market. The strategy involves monitoring the news sentiment for 25 different sector Exchange Traded Funds (ETFs) and periodically rebalancing the portfolio to maximize exposure to sectors with the highest public sentiment. Backtesting results demonstrate that the strategy consistently outperforms benchmark approaches. The tutorial provides details on universe selection, implementation, and presents equity curves and Sharpe ratios for different versions of the strategy and benchmarks. To replicate the results, users are encouraged to clone and backtest each algorithm.

Abstract

Abstract: This tutorial explores four alternative data strategies that utilize the US Regulatory Alerts dataset to make trading decisions. The strategies include capitalizing on movement in the healthcare sector in response to FDA announcements, capturing momentum in the Bitcoin-USD trading pair based on new Crypto regulations, exploiting trading patterns in the SPY based on specific regulatory alerts, and a country rotation strategy using NLP to detect sentiment in country ETFs. The results show that all four strategies outperform their respective benchmarks. The tutorial also discusses NLP and its role in trading strategies, as well as the implementation of the four strategies using the LEAN algorithmic trading engine.

Abstract

Abstract: This tutorial focuses on utilizing natural language processing (NLP) to detect impactful news in ETF constituents. Building upon a previous NLP strategy, we monitor the Tiingo News Feed to determine intraday news sentiment of the largest constituents in the Nasdaq-100 index, while avoiding look-ahead bias. The results indicate that this strategy has experienced lower risk-adjusted returns compared to the QQQ ETF over the past two years. The tutorial discusses the implementation of this strategy as a framework algorithm using the LEAN trading engine, including universe selection and portfolio construction. Backtesting results show a Sharpe ratio of -0.659, with comparisons to other benchmarks provided.

Abstract

This discussion focuses on the detection of the head and shoulders pattern in technical analysis. While technical analysis traders commonly use graphical patterns to identify trading opportunities, quant traders tend to overlook them due to subjectivity and difficulty in accurate detection. However, this tutorial presents a method to programmatically detect the head and shoulders pattern in an event-driven trading algorithm. The algorithm achieves greater risk-adjusted returns than the benchmarks during the backtesting period. The head and shoulders pattern consists of two shoulders, a tall head, and a neckline. It is believed to signal a bullish-to-bearish trend reversal. Further research can include testing other technical patterns, adjusting algorithm parameters, exploring new position sizing techniques, implementing different exit strategies, and incorporating risk management for corporate actions.

Abstract

This research focuses on Futures Fast Trend Following strategies that can be applied to both long and short positions, taking into account the strength of the trend. The purpose of the research is to explore the effectiveness of these strategies and their potential implications for trading in the futures market. The research utilizes various methods to analyze historical data and identify trends, and the key findings highlight the profitability and consistency of the trend following strategies. The implications of the research suggest that these strategies can be valuable tools for traders seeking to capitalize on trends in the futures market.

Abstract

This research is a re-creation of strategy #11 from Advanced Futures Trading Strategies (Carver, 2023) that combines carry and trend strategies in futures trading. The algorithm incorporates exponential moving average crossover (EMAC) trend forecasts and carry forecasts to form a diversified portfolio. The results show that using both styles of strategies can improve risk-adjusted returns. Additionally, the research provides a background on how carry returns are calculated for different asset classes and how the strategy calculates and smooths carry from different future contracts.

Abstract

QuantConnect provides trading infrastructure and data for quants to develop and deploy algorithmic trading strategies. They offer the Alpha Streams platform for quants to license their proprietary signals to investors. To assist investors in analyzing the performance of these signals, QuantConnect has released a new notebook that determines the optimal portfolio weights for each alpha, maximizing the portfolio's Sortino ratio. The Sortino ratio measures the strategy's average daily return in excess of a risk-free rate, divided by the standard deviation of negative daily returns. The notebook uses a walk-forward approach to avoid bias and overfitting, and the optimization is done on a rolling monthly basis.

Abstract

Residual momentum is a strategy where stocks with higher monthly residual returns outperform those with lower returns. It has been found to have less exposure to Fama-French factors, higher Sharpe ratios, and better out-of-sample performance compared to total return momentum strategies. Residual momentum is also more stable throughout the business cycle and tends to underperform during trending periods but outperform during reverting periods. This strategy is less concentrated in small-cap stocks, leading to lower trading costs and minimizing the impact of tax-loss selling. The algorithm imports custom data, selects a universe of stocks based on fundamental data and market cap, and rebalances the portfolio monthly by longing the top 10% and shorting the bottom 10% of stocks based on their scores.

Abstract

This discussion focuses on the momentum in mutual fund returns and the use of net asset value (NAV) as a predictor of future returns. The study suggests that historical returns and the proximity of NAV to a previous high can provide predictive power. The tutorial uses asset management firms' share prices as a proxy for fund performance and NAV. The performance of this trading strategy is compared to buying-and-holding the S&P 500 index ETF. The strategy generally has a lower Sharpe ratio than the benchmark, except during the 2020 stock market crash where it significantly outperformed with a Sharpe ratio of 9.3. The strategy also demonstrates more consistent returns compared to the benchmark across different testing periods.

Abstract

This tutorial implements an intraday momentum strategy for trading actively traded ETFs. The strategy predicts the sign of the last half-hour return based on the return generated in the first half-hour of the trading day. The algorithm is a recreation of the research conducted by Gao, Han, Li, and Zhou (2017), which found that this momentum pattern is statistically and economically significant. The tutorial provides background information on the characteristics of the opening and closing periods of trading, as well as the selection of ETFs for the strategy. The conclusion states that the momentum pattern produces lower returns compared to the S&P 500 benchmark, but outperforms the benchmark during the downfall of the 2020 crash.

Abstract

Abstract

Abstract

This tutorial focuses on training a Gradient Boosting Model (GBM) to forecast intraday price movements of the SPY ETF using technical indicators. The implementation is based on research by Zhou et al (2013), who found that a GBM produced a high annualized Sharpe ratio. However, the tutorial's research shows that the model underperforms the SPY with its current parameter set during a 5-year backtest. The tutorial concludes by suggesting potential areas of further research to improve the model's performance. The GBM is trained by iteratively building regression trees to predict pseudo-residuals and making predictions based on the learning rate and regression tree outputs. Technical indicator values are used as inputs, and the mean squared error loss function is used to assess the model's performance.

Abstract

Abstract: This tutorial explores the use of news sentiment to predict the price direction of drug manufacturers. By implementing an intraday strategy, we aim to capitalize on the upward drift in stock prices following positive news releases. Our findings show that combining this effect with the day-of-the-week anomaly can lead to profitable trading during the 2020 stock market crash. However, our algorithm underperforms the S&P 500 market index ETF, SPY, during the same period. The algorithm is inspired by the work of Isah, Shah, & Zulkernine (2018). We conclude that while the sentiment analysis strategy may not provide accurate results in the US drug manufacturing industry, profitability can be achieved by restricting trading to the most profitable day of the week. The strategy produces a negative Sharpe ratio of -1.

Abstract

Abstract: This discussion focuses on the Gaussian Naïve Bayes (GNB) model and its application in forecasting the daily returns of stocks in the technology sector. The GNB model is trained using historical returns of the sector and compared to the performance of the SPY ETF over a 5-year backtest and during the 2020 stock market crash. The implementation of the GNB model shows a higher Sharpe ratio and lower variance compared to the SPY ETF. The algorithm used in this discussion is based on previous research and follows the principles of Naïve Bayes models. The GNB model assumes independence and normal distribution of feature vectors.