3 Strategies

Here we will list some strategies you can try out. The main goal is to provide a starting point for you to learn how to do research part of trading strategydevelopemnt or/and trading strategies in backtesting engine (Quantconnect in our case). We are providing very shoer description of the strategy and link to original source of the strategy. You can choose one or more strategies and try to reproduce them. You can analyse some additional aspect of strategy not already mentioned in the source document.

3.1 Where to look for ideas?

SSRN - SSRN is a repository of academic research papers in finance and economics. Look at G1, C58 JEL codes.
QuantConnect - QuantConnect has list of trading ideas
Robot Wealth blog and bootcamp. Robot Wealth is a blog and online course platform that provides resources on quantitative trading strategies.
Quant roadmap from X user quant_arb. This roadmap provides detailed information on how to get started with quantitative trading, including the tools and resources you need to succeed. It also provides resources to dive into some aspects of trading like momentum, seasonality, volatility etc.
Quantpedia - Quantpedia is a great resource for quantitative trading strategies. They have a large database of strategies that are well-researched and backtested. You can search for strategies based on different criteria, such as asset class, strategy type, and performance metrics.
Substack - Substack is a platform for independent writers to publish newsletters. Many traders and researchers share their trading ideas and strategies on Substack. You can subscribe to newsletters that focus on quantitative trading and receive regular updates on new strategies and research. Some popular newsletters include: QuantSeeker have weekly recaps (free) where he summarize new quant research papers on daily basis. Machine Learning & Quant Finance
Vertox posts can be downloaded here
Social media - Twitter and reddit.

3.2 Sources for stats and econometrics for quant finance

3.2.1 Portfolio optimization

Portfolio optimization by Daniel P. Palomar. It-s free! The book provides a comprehensive introduction to the field of portfolio optimization, with a focus on modern techniques derived from convex optimization. The book covers a wide range of topics, including mean-variance optimization, robust optimization, and factor models. It also includes practical examples and exercises to help readers apply the concepts to real-world problems.
Quantitative Portfolio Management: The Art and Science of Statistical Arbitrage by Michael Isichenko - This book provides a comprehensive introduction to the field of quantitative portfolio management, with a focus on statistical arbitrage.

3.2.2 Financial econometrics

Financial Econometric Modeling Stan Hurn, Vance L. Martin, Peter C.B. Phillips, Jun Yu - This book provides a comprehensive introduction to the field of financial econometrics, with a focus on practical applications in finance. The book covers a wide range of topics, including time series analysis, volatility modeling, and factor models. It also includes practical examples and exercises to help readers apply the concepts to real-world problems.
Analysis of Financial Time Series by Ruey S. Tsay. This book provides a comprehensive introduction to the field of financial econometrics, with a focus on time series analysis. The book covers a wide range of topics, including volatility modeling, factor models, and forecasting. It also includes practical examples and exercises to help readers apply the concepts to real-world problems.
Nonlinear Time Series Analysis by Ruey S. Tsay and Rong Chen. This book provides a comprehensive introduction to the field of nonlinear time series analysis, with a focus on practical applications in finance.
Ben Lambert’s Econometric Course - this is general introduction to econometrics. Emphasize is not on financial econometrics.

3.2.3 Financial machine learning

Advances Financial Machine Learning by Marcos Lopez de Prado. This book provides a comprehensive introduction to the field of financial machine learning, with a focus on practical applications in quantitative trading. The book covers a wide range of topics, including feature engineering, model evaluation, and portfolio construction. It also includes practical examples and exercises to help readers apply the concepts to real-world problems.
Machine Learning for Algorithmic Trading: Predictive models to extract signals from market and alternative data for systematic trading strategies with Python by Stefan Jansen. This book provides a comprehensive introduction to the field of machine learning for algorithmic trading, with a focus on practical applications in quantitative trading. The book covers a wide range of topics, including data preprocessing, feature engineering, and model evaluation. It also includes practical examples and exercises to help readers apply the concepts to real-world problems.

3.2.4 Quantitative Trading Strategies

Quantitative Trading: How to Build Your Own Algorithmic Trading Business by Ernie Chan - This book provides a comprehensive introduction to the field of quantitative trading, with a focus on building and testing algorithmic trading strategies. The book covers a wide range of topics, including backtesting, risk management, and execution. It also includes practical examples and exercises to help readers apply the concepts to real-world problems.

3.3 Intraday trading strategies

3.3.1 The Opening Range Breakout

The paper of the strategy can be found here.

The Opening Range Breakout (ORB) strategy is a day trading approach that focuses on trading within the first few minutes after the market opens. The strategy works as follows:

First, it identifies the opening range, which is defined by the highest and lowest prices within the first few minutes of the trading session, commonly five minutes. This range serves as the basis for setting entry points. A trade is initiated if the price breaks out of this opening range. For example, if the price moves above the high of the opening range, a long (buy) position is taken. Conversely, if the price drops below the low of the opening range, a short (sell) position is taken.

The actual entry into the trade is made at the open of the second 5-minute candle in the same direction as the breakout. If the first candle is bullish, meaning it closes higher than it opened, a long position is taken at the open of the second candle. If the first candle is bearish, closing lower than it opened, a short position is taken at the open of the second candle.

Risk management is a crucial aspect of the ORB strategy. For a long trade, the stop loss is set at the low of the first 5-minute candle, and for a short trade, the stop loss is set at the high of the first 5-minute candle. This stop loss helps manage risk by limiting potential losses if the trade moves against the expected direction. The profit target is typically set at 10 times the risk, labeled as $R. If the price reaches this target, the trade is exited. If the target is not reached by the end of the day, the position is closed at market closure.

Leverage and position sizing are also important components of the strategy. The strategy assumes a maximum leverage of 4x, as allowed by most US brokers. The trading size is calibrated so that if a stop loss is hit, it results in a loss of 1% of the trading capital. The formula for calculating the number of shares to trade takes into account the account size, the risk per trade, and the leverage allowed.

For example, suppose the market opens and the first 5-minute candle forms with a high of $100 and a low of $95. If the second candle opens at $101, indicating a bullish breakout, a long position is entered at $101 (the open of the second candle). The stop loss is set at $95 (the low of the first candle), and the profit target is set at $161 (10 times the risk of $6).

The ORB strategy was backtested from January 1, 2016, to February 17, 2023, using the QQQ ETF. The results showed that the ORB strategy significantly outperformed a passive buy-and-hold strategy on QQQ. The introduction of leveraged ETFs like TQQQ allowed traders to overcome leverage constraints and achieve higher returns.

Key takeaways from this strategy include trading in the direction of the breakout from the opening range, effective risk management through the use of stop losses and profit targets, and the use of leveraged ETFs to enhance returns. The strategy proved effective across different market conditions, including both bull and bear markets. By adhering to these principles, the ORB strategy aims to capitalize on early market volatility to achieve significant returns.

3.3.2 Intraday momentum

The paper of the strategy can be found here.

The paper investigates the profitability of an intraday momentum strategy applied to the SPY ETF, which tracks the S&P 500. This strategy, unlike others that limit trading to the last 30 minutes of the session, initiates trend-following positions as soon as there is an indication of an abnormal demand/supply imbalance in intraday price action.

The strategy works as follows: The data used covers SPY and VIX from May 2007 to April 2024, using 1-minute OHLCV (Open, High, Low, Close, and Volume) data from IQFeed. The Noise Area represents a region where prices are expected to oscillate under conditions of balanced buying and selling pressures. It is defined using the average movements over the previous 14 days for each time-of-day interval. The Upper Boundary and Lower Boundary of the Noise Area are calculated based on the average movement from the open, adjusted for overnight gaps. Positions are initiated when the price crosses the Noise Area boundaries. A crossing above the Upper Boundary indicates a long position, while a crossing below the Lower Boundary indicates a short position. Positions are unwound at market close or if there is a crossover to the opposite boundary of the Noise Area. Trading is restricted to bi-hourly intervals (e.g., HH:00 and HH:30) to avoid overtrading due to short-term market fluctuations.

The strategy uses dynamic trailing stops to mitigate downside risks while allowing for unlimited upside potential. The trailing stops are set at the opposite side of the Noise Area. The number of shares traded is calculated based on the equity available at the beginning of each trading day, with adjustments made for market volatility. The strategy was backtested with an initial capital of $100,000, incorporating transaction costs and slippage. The equity curve of the strategy showed a total return of 1,985%, with an annualized return of 19.6% and a Sharpe Ratio of 1.33. The strategy adjusts daily exposure based on recent market volatility, targeting a daily market volatility of 2%.

3.3.3 Ensemble trend following

Trend following strategies are based on the idea that assets that have performed well in the past will continue to perform well in the future. Trend following strategies are usually based on moving averages, breakouts, and other technical indicators. They are usually simple to implement and have been shown to work in different markets and asset classes. They are not the same as momentum strategies. While trend following strategies analyse time series dimension (e.g. assets is on ATH), momentum strategies analyse cross-section dimension (e.g. returns of assets in point in time).

There are various approaches to measure trend in the market. We can’t know ex ante (through experience or current research) which approach is best. So we can try different approaches and combine them in an ensemble An we want to estimate is statisticaly and economicaly significant difference between individual indicators and ensemble

In this little project our goal is to calculate different trend following indicators on dynamic asset stock universe and combine them in an ensemble. We recommend following steps:

Go to Robot Wealth bootcamp and open the video on trend following and momentum strategies.You can watch the whole video, but if you already know basics of trend following and momentum trading go to 42:00 and watch till the
Ensure you have data for the project. You need daily stock OHLCV prices. Optionally, if you wan to try on crypto data, you should have daily crypto prices too.
Filter dynamic sock universe. For the beginning, choose 10 stocks with highest market capitalization or highest dollar volume by month. This means, that you have monthly rebalancing of the portfolio.
Calculate trend following indicators for all stocks in the universe. Use the indicators mentioned on the dashboard in the RW video.
Calculate ensemble for all indicators.
Visualize equity curve for trading strategy that buys stocks when it is in uptrend.

3.3.4 ML market regimes

The paper of the strategy can be found here.

The authors develop a hybrid model combining unsupervised learning (for regime identification) and supervised learning (for regime forecasting). Traditional portfolio strategies often depend on generalized market conditions, which may not adequately capture the nuances of individual assets. By focusing on asset-specific regimes, the proposed framework can respond more dynamically to market changes, leading to potentially enhanced investment performance. The authors demonstrate that this approach results in better portfolio outcomes compared to more conventional methods, particularly in complex and rapidly changing market environments.

3.3.5 Price-Path Convexity and Short-Horizon Return Predictability

The paper explores the relationship between the curvature of stock price paths, termed “price-path convexity,” and future returns. The authors find that convexity negatively predicts short-horizon returns at both the aggregate and firm levels, with its predictive power outperforming many traditional return predictors. The study identifies this relationship as driven partly by behavioral biases, particularly overextrapolation, where investors overreact to recent price trends, leading to reversals. Convexity provides unique insights distinct from cumulative returns or other known factors, remaining robust across various time periods, horizons, and even after accounting for liquidity and risk considerations.

tags: intraday, mean reversion

3.3.6 Buy the dip strategy

The one variant of this strategy can be found here.

If the day’s close is near the bottom of the range, the next day is more likely to be an upwards move. It can also be labeled as mean reversion strategy. The strategy is based on the idea that if the price of an asset falls significantly in a short period, it is likely to rebound in the near future.

3.4 Options and volatility trading strategies

3.4.1 Short volatility

The one variant of this strategy can be found here.

This trading strategy focuses on capitalizing on the variance risk premium (VRP) by systematically comparing implied volatility (IV) to realized volatility (RV) over time. The process begins with creating a time series that tracks IV against RV. To clarify, implied volatility is a measure derived from options prices that reflects the market’s expectations for future volatility. Realized volatility, on the other hand, is the actual volatility that occurs over a given time period. The strategy involves comparing these two forms of volatility over a four-year period to determine the average difference between them. This difference is what the strategy refers to as the variance risk premium. A positive VRP indicates that the market has consistently priced options with an expectation of higher volatility than what actually occurs, suggesting that there may be an opportunity to profit by selling these options.

Once the VRP is identified, the next step is to backtest whether this risk premium can actually be monetized. The backtest involves a specific trading strategy: selling 30-day at-the-money (ATM) straddles. A straddle is an options strategy where both a call option (which benefits from price increases) and a put option (which benefits from price decreases) are sold at the same strike price and expiration date. The ATM straddle specifically refers to selling options where the strike price is very close to the current price of the underlying asset. The backtest also involves delta hedging, which is the process of adjusting the position daily to remain neutral to price changes in the underlying asset. Delta hedging helps manage risk by ensuring that the overall portfolio does not have directional exposure to the underlying asset’s price movements. To make the backtest realistic, the strategy assumes that traders must cross the bid-ask spread when entering trades, which accounts for slippage, or the small loss that occurs when executing a trade at a less favorable price.

The ultimate goal of this backtest is to see if the identified VRP can be converted into profits after accounting for trading costs and the daily adjustments required by delta hedging. If the strategy is profitable under these conditions, it suggests that the VRP is not just a theoretical construct but something that can be exploited in practice.

However, the strategy does not stop at simply identifying profitable trades. It also incorporates a risk management layer by considering the implied volatility percentile. The IV percentile measures how high the current implied volatility is relative to its historical levels. The strategy sets a threshold where only trades with an IV percentile below 90% are considered. This is because while higher volatility might seem to offer more opportunities, it also increases the variability of outcomes, leading to more unpredictable profits and losses (PNL). By filtering out trades where the IV percentile is too high, the strategy aims to avoid situations where the potential for higher gains is outweighed by the increased risk.

Trades that meet these criteria—having a positive VRP and an acceptable IV percentile—are categorized into a “universe of good trades.” The strategy aims to take as many of these trades as possible, though additional rules may help in selecting the best opportunities within this universe. Notably, the strategy does not attempt to time the risk premium, meaning it does not try to predict when volatility will be particularly favorable for trading. Instead, it relies on the law of large numbers, which suggests that over a sufficiently large number of trades, the positive VRP will result in consistent profits, even if individual trades vary in outcome.

Finally, this strategy is applied exclusively to exchange-traded funds (ETFs) rather than individual stocks. ETFs, which are funds that hold a diversified portfolio of assets and trade like a stock, tend to exhibit a more reliable and favorable risk premium compared to individual stocks. Stocks often have more idiosyncratic risks, or risks unique to that specific stock, which can make the VRP less predictable and harder to monetize. By focusing on ETFs, the strategy aims to exploit the more stable and consistent risk premium that these instruments offer.

3.4.2 Short VIX

The one variant of this strategy can be found here, page 81.

The implied volatility term structure can be a powerful predictor in financial markets, particularly when dealing with futures contracts. Implied volatility, a measure reflecting the market’s expectations for future price swings, is derived from the prices of options rather than being directly observable. The term structure of these futures, which maps out the relationship between prices of contracts of various maturities, plays a crucial role in predicting future returns. Typically, this term structure can exhibit two primary conditions: contango and backwardation. In contango, longer-dated futures contracts are more expensive than their shorter-dated counterparts (or spot price), often due to factors like storage costs and insurance. Conversely, backwardation occurs when longer-dated futures are cheaper, which can happen when there’s strong current demand or expected future supply shortages.

The VIX, also known as the “fear gauge,” measures market expectations of volatility for the next 30 days and is derived from S&P 500 index options. VIX futures often trade in contango when the VIX index is low and in backwardation when the index is high. This term structure anomaly conflicts with the concept of mean reversion, which suggests that prices eventually return to their historical average levels. For instance, if the VIX is unusually low, mean reversion would predict an increase in the VIX, while the contango in the term structure might suggest the opposite. Despite this conflict, the term structure usually serves as a stronger predictor of future returns, with exceptions primarily occurring when the VIX is at very high levels.

In practical trading terms, the strategy would involve selling VIX futures or index options when the term structure is in contango, as these prices are expected to decline. Conversely, buying VIX futures or index options when the term structure is in backwardation could be profitable, given the expectation that these prices will rise. However, it is worth noting that the term structure’s predictive power does not apply uniformly across all markets. For example, in the context of equity options, the presence of extreme cases within the stock universe can cause mean reversion to dominate, leading to different outcomes. Nevertheless, the general principle remains that the term structure can provide valuable insights into future market movements, guiding informed trading decisions.

3.5 Other (clasify if found some paper on ssrn or arhivx)

3.5.1 Risk leverage trading

This is my trading strategy idea (haven’t yet found paper that implement similar logic). The idea is to use leverage in trading based on the risk of the trade. The risk of the trade is calculated with risk indicators. We want to use small numbers of potentially orthogonal (different) indicators. I already have 2 indicators I used (MinMax and PRA), but the goal is to use test the strategy with 2 to 10 indicators and see performance. The leverage used is function of active risk indicators. For example, if we use tow indicatora I mentioned above (MinMAx and PRA), we can invest in SPY with following logic: 1) invest x2 in SPY (or in SSO ETF that invest n SPY with 2x leverage) if both indicators are bullish. 2) invest x1 in SPY if one indicator is bullish and one is bearish. 3) invest x0 in SPY if both indicators are bearish (try also with short).

The goal is to test the strategy on SPY ETF and compare it with buy and hold strategy. The strategy should be tested on as much data possible. The strategy should be tested with:

different number of indicators (2, 3, 4, 5, 6, 7, 8, 9, 10)
different leverage (1x, 2x, 3x, 4x, 5x, 6x, 7x, 8x, 9x, 10x)
different risk indicators (MinMax, PRA, RSI, MACD, Bollinger bands, etc.)
different time frames (daily, hourly).

and the performance should be compared with buy and hold strategy.

We want to test the strategy only with SPY in the beginning.

3.6 Crypto trading strategies

3.6.1 A Perpetual Futures Basis Strategy

Strategy is explained in detailed on Robot Wealth bootcamp. Note there is an notebook with R code on the bottom of the document.

3.7 Event based strategies

3.7.1 IPO ATH

The idea is to stocks after IPO if they are at all time high in 90 days from the IPO. The strategy is explained in detail in the substack post.