The Alpha Formula
- Our favorite trading-related book of 2019
- Includes 4 stand-alone strategies
- Combines strategies to portfolio for improved risk-adjusted performance
The Alpha Formula is the fall 2019 publication in a long list of top-selling books on market strategies and volatility trading by Connors Research. The founder of Connors Research is Larry Connors, an industry veteran with more than 35 years of experience in the financial markets. In the past, Larry Connors has been the primary author of Connors Research's books. The Alpha Formula is the first book primarily authored by Chris Cain, a senior quantitative researcher at Connors Research and an expert on trading system design and development.
It seems that shifting primary authorship from Larry Connors to Chris Cain has significantly changed the nature of this book compared to previous releases. We always liked publications by Connors Research, namely Short Term Trading Strategies That Work, High Probability ETF Trading, and Buy the Fear, Sell the Greed. At the same time, we had some complaints about these books regarding the transparency and depth of their content. The Alpha Formula addresses all these complaints and is, in our opinion and at the time of this writing, hands-down the best book ever published by Connors Research.
A New Style of Writing
Strategies by Connors Research have always been effective, easy to follow, and typically simple to implement. The Alpha Formula follows that lead, even though its strategies are more complex than those from previous books.
With earlier books by Connors Research, we disliked that they disclosed backtesting results only in tables with key metrics. The Alpha Formula addresses this criticism in multiple ways. For once, the book presents more metrics for the backtests. Further, the book now also shows charts with cumulative returns and drawdowns. Also, the book discusses the backtested results, points out key findings, and puts the strategies into a broader investment context. We like the new format a lot and feel that it is a significant improvement: It is more instructional for readers and makes it much easier for developers to validate their implementation.
Previous books by Connors Research provided only minimal background. They focused on presenting the strategies and showing a few selected example trades. With Buy the Fear, Sell the Greed, we first saw an increased focus on behavioral finance background. But The Alpha Formula takes this to an entirely new level. About a third of the book describes the question of why prices behave the way they do and why trend following and mean reversion work. New is the use of First Principles. While these principles are not ground-breaking, they are used throughout the book to provide context for the strategies. In our opinion, Chris Cain has found an excellent balance between the strategies as the book's main deliverable and the background required for a more thorough understanding.
What makes The Alpha Formula genuinely outstanding is its portfolio approach. While previous books by Connors Research were merely a collection of loosely related strategies, The Alpha Formula takes a new route. The four strategies presented by Chris Cain are intentionally diversified in multiple ways: the assets traded, the methodology, and the targeted First Principle. And while the strategies can all be traded stand-alone, they are meant to be used in unison, creating a portfolio with improved diversification and robust returns in almost any market environment. We have seen books by other authors taking a similar approach, but in our opinion, Chris Cain has managed to present the most convincing solution yet.
The book's first strategy is a long-only trend-following strategy, trading a universe of 12 ETFs covering various markets and asset classes. We have seen plenty of these strategies before, yet Rising Assets has more robust returns than most of them. We attribute these improved returns to the strategy's inverse volatility weighting.
The effect of this weighting clearly shows in the Monte-Carlo simulation. While other momentum strategies tend to amplify the standard deviation of returns, Rising Assets reduces volatility compared to a passive 60/40.
The strategy determines momentum over multiple time frames. This method might be an improvement over a naïve momentum calculation. Still, we believe the strategy could do even better if it used a more sophisticated momentum estimation with better noise rejection. Regardless, we like Rising Assets and consider it one of the better simple trend-following strategies we have in our arsenal.
Implementing the strategy for TuringTrader was straightforward, and our results match those published in the book closely.
Weekly Mean Reversion
The next strategy attempts to buy pullbacks in an overall bullish market, trading individual stocks with top liquidity. Weekly Mean Reversion's core logic is very similar to the various mean-reversion strategies published in previous books by Larry Connors. However, Chris Cain put a slightly new spin on it. Our primary complaint with previous mean-reversion strategies was their underutilization of available capital. Weekly Mean Reversion addresses this issue twofold: First, the strategy invests in 10 of the 500 most liquid stocks, providing more opportunities to place a trade. Second, Weekly Mean Reversion allocates idle cash to U.S. Treasuries.
We always liked mean-reverting strategies, primarily because of their high probability of winning and the resulting improved Sharpe Ratios. Weekly Mean Reversion demonstrates the best of this approach. The Monte-Carlo simulation shows that Weekly Mean Reversion, compared to holding the S&P 500, has much-improved returns while at the same time cutting the downside in half. However, it is worth noting that the tail risk of the strategy is similar to holding the S&P 500, as it invests up to 100% of the available capital in the stock market.
Weekly Mean Reversion has quickly become one of our all-time favorite strategies. In addition to its risk and return properties, the following aspects make Weekly Mean Reversion unique: First, the strategy only trades once per week, keeping maintenance requirements low. And second, Weekly Mean Reversion uses a stop-loss to exit positions should the market move against us.
Implementing Weekly Mean Reversion for TuringTrader was straightforward. The only slightly more tricky aspect of the strategy is it operating on both weekly and daily bars. Our results match those published in The Alpha Formula closely. The universe selection can explain the difference in results: While the book uses a universe of the 500 most liquid stocks, we use a survivorship-bias free S&P 500 universe for our testing. For a strategy trading as frequently as the Weekly Mean Reversion, we should always consider trading commissions. Unfortunately, The Alpha Formula doesn't do so and assumes zero commissions and fees instead.
U.S. Treasuries have proven to be an essential portfolio component in times of market stress, mainly because of their negative correlation to the stock market. Compared to a static allocation, Dynamic Treasuries attempts to provide value by dynamically adjusting the treasuries' duration to optimize returns.
When Chris Cain wrote The Alpha Formula, there was much concern that the U.S. Federal Reserve would raise interest rates, and with that, smash the total return of longer-term bonds for years to come. However, the Fed has changed course since, and it looks like there won't be any rate hike before the end of 2021. In this unforeseen new context, the returns of Dynamic Treasuries are underwhelming. Not only does the strategy offer no edge over buying and holding medium-term Treasuries, but it also does so with higher volatility.
We agree with using U.S. Treasuries to reduce portfolio volatility, and we see value in adjusting the duration of the bonds held. However, we believe the approach taken with Dynamic Treasuries is too simplistic. For a menu of Treasuries with varying durations, the strategy measures their momentum over multiple time frames. It then weighs the assets depending on the number of time frames with positive returns. This method puts longer-term Treasuries at a disadvantage because of their higher volatility. In our opinion, a viable strategy should determine the current rate environment through the Federal Funds Rate instead of the bond's momentum. Further, an improved approach should extend duration in times of market stress while reducing duration in times of recovery.
Unlike the book's other strategies, Chris Cain fails to prove Dynamic Treasuries' value, as yields continuously fell throughout the backtested period. Instead of using proxies for the ETFs and testing Dynamic Treasuries before 1980, Chris reverts to voicing his mere opinion. "We believe it is" [worth to run this strategy] are words we should not find in a book about quantitative trading.
Implementing the strategy for TuringTrader was very simple, and our results match the book closely. However, we believe Dynamic Treasuries to be the weakest strategy in the book and have only minimal interest in trading it.
The last stand-alone strategy in The Alpha Formula is called ETF Avalanches and aims to profit from global bear markets. The strategy uses a universe of 40 ETFs covering various countries, sectors, and markets. ETF Avalanches uses mean-reversion and sells assets showing short-term strength while being extremely volatile and in an overall downtrend. As an additional improvement, the strategy puts idle cash into short-term Treasuries.
In the long term, and as expected, ETF Avalanches produces returns similar to those of short-term Treasuries. This result is much better than many other short-position strategies, which continuously lose money during bull markets. During times of market stress, ETF Avalanches picks up substantially, helping to counteract losses from assets with positive market correlation.
The strategy works as advertised, and we generally like it a lot. However, we see this strategy as a dynamic portfolio component. As ETF Avalanches provides only meager returns outside of market stress, we feel that ETF Avalanches creates too much drag on overall portfolio performance if used as a static component.
Implementing ETF Avalanches for TuringTrader was straightforward. However, our backtested results differ substantially from the data published in The Alpha Formula. In particular, we see our implementation profit more than the book in fall 2011, while the book did better in summer 2015. So far, we have not been able to identify the root cause of these differences.
The Alpha Portfolio
While we certainly enjoyed the strategies presented in the book, what sets this book apart is something else: The Alpha Portfolio. About a third of the book is devoted to combining the four individual strategies into a portfolio and reviewing the resulting meta-strategy's characteristics. The Alpha Portfolio is diversified across asset classes, geographical regions, strategy styles, and time horizons, leading to smooth returns and low volatility in many different market environments. The portfolio combines high returns with low volatility and a low correlation to the stock market. With these properties, the Alpha Portfolio addresses almost all needs investors are looking for and rivals many ETFs and mutual funds available to retail investors. We have not seen a strategy of this quality published in a book before.
We have only very little criticism for the Alpha Portfolio. Except for the Dynamic Treasuries, all portfolio components are solid. However, we feel that ETF Avalanches is a drag on the overall portfolio performance. We believe it would be helpful to dynamically shift exposure between Rising Assets and ETF Avalanches, based on sentiment and possibly volatility of the stock market.
Understandably, a portfolio of this complexity comes at a price. It requires solid programming skills to implement the various components of the Alpha Portfolio. A reliable data feed is required to access the quotes for the many instruments. Further, readers need a survivorship-bias-free universe for Weekly Mean Reversion. And finally, investing in the Alpha Portfolio requires significantly more maintenance than most other portfolios.
In summary, The Alpha Formula appeals equally to readers looking for a strategy they can use out of the box, as it does to quant developers looking to jump-start their own development. We consider The Alpha Formula a must-read for anybody interested in quantitative trading and the best trading-related book we read in 2019.
Even more importantly, The Alpha Formula has been the motivation to launch our All-Stars. This family of meta portfolios offers the same benefits of combining uncorrelated strategies, but with TuringTrader's ease of use.