Quantpedia Premium Update – 1st May

New strategies:

#743 – Gordon Growth Fair Value Model

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Complex strategy
Backtest period: 2003-2018
Indicative performance: 14.63%
Estimated volatility: 11.89%

Source paper:

Rasmussen, Henrik and Thormann, Andre, The Discounted Cash Flow Terminal Value Model As an Investment Strategy
https://ssrn.com/abstract=3396505
Abstract
To determine a company’s intrinsic value, equity analysts make forecasts several years into the future, although the value generated in the explicit forecast period only represents a small part of the total valuation. Instead, most of a company’s value is determined in the terminal period – a perpetuity that includes all years after the explicit forecast. The literature states that the forecasts of analysts are too optimistic, and their assumptions in the terminal period does not reflect a normalized level in the company’s business cycle. This thesis develops several investment strategies based on estimates of the terminal value by using past financials with the goal of eliminating optimism and bias. Our valuations utilize the Gordon Growth and value driver formulas – both of which are commonly used in discounted cash flow models. We present several variations to estimate the fundamental components in the two formulas and test whether these variations are robust. The strategies are applied to the 727 non-financial stocks that have been a part of the S&P 500 index from 2003 to 2018. The results are benchmarked against the index and the performance of stock recommendations from Morningstar’s independent equity analysts. By investing in stocks trading below their terminal value, the strategies produce average annualized returns between 12.6% and 17.6% in excess of the risk-free rate, while the S&P 500 index excluding financials has generated 12.5% annualized. A portfolio of 4- and 5-star rated stocks gave 13.1% yearly but with a relatively higher risk. The Gordon Growth and value driver strategies have similar returns and risk, but an important finding is that they invest in stocks with very different characteristics. The Gordon Growth strategy invests in stocks that appear cheap on price multiples and have higher debt, lower margins, and lower returns on invested capital. In contrast, the value driver strategy invests in cheap stocks with higher quality on these parameters. Both strategies favour stocks in the healthcare and consumer defensive sectors but often find technology stocks to be overvalued. If the strategies are used to both buy cheap and short-sell expensive stocks, they continue to deliver strong risk-adjusted returns, but the different variations of the strategies are less consistent – particularly the value driver strategies. The results are robust across many variations of the Gordon Growth and value driver strategies, but if the assumptions to growth and discount rates become too conservative, the amount of undervalued stocks is reduced, which makes the strategies more concentrated and riskier. The discount rate also has significant influence on which sectors appear most attractive. The results indicate, that our strategies represent a better alternative to the traditional quantitative value factor of Fama & French that evaluates stocks on their book-to-market ratios. The value factor has performed poorly in the past decade and appear redundant in the face of more recent factors of profitability and investment.

#744 – The Halloween Effect Within Long-term Reversal

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Simple strategy
Backtest period: 1931-2021
Indicative performance: 7.26%
Estimated volatility: 12.16%

Source paper:

King Fuei Lee: An Anomaly within an Anomaly: The Halloween Effect in the Long-term Reversal Anomaly (2021)
https://ssrn.com/abstract=3973224
Abstract
In this study, we investigated the presence of the Halloween effect in the long-term reversal anomaly in the US. After examining the cross-sectional returns of loser-minus-winner portfolios formed on prior returns over the period of 1931–2021, we found evidence of stronger returns during winter months versus summer months. Specifically, the effect appeared to be driven by a significant winter-summer seasonality in the portfolio of small-capitalisation losers and a lack of the Halloween effect in the portfolio of large-capitalisation winners. This study’s results were found to be robust with respect to alternative measures of the long-term reversal effect, differing sub-periods, the inclusion of the January effect and outlier considerations, as well as regarding small- and large-sized companies.

#745 – Value and Profitability in Chinese Equities

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Moderately complex strategy
Backtest period: 1999-2020
Indicative performance: 8.73%
Estimated volatility: 9.9%

Source paper:

Feng, Frank Yulin and Kang, Wenjin and Liu, Shuyan and Zhang, Huiping and Zhang, Kang: Is Value Strategy Still Alive? Evidence from the Chinese A-Share Market
https://ssrn.com/abstract=4029336
Abstract
The value investment strategy has been in the limelight since the 1930s but has faced challenges recently. We find that, while the last two decades have seen vanishing profits from value investing in the U.S., the value strategy remains successful in China’s A-share market, the second-largest stock market in the world. The factor combining various valuation and profitability metrics generates highly positive returns over time in China. The differing performance of value investing in the two markets is partly driven by the short leg of the relative strategy and poses an investment dilemma: Short selling stocks of growth firms contributes to the value-investing profit in China but generates paper money only, while in the U.S., short sales of such stocks are doable but produce no alpha. To circumvent this difficulty, we propose a new value investing mindset to buy value firms only in both markets. Due to the extremely low return correlation of the two country portfolios, such an international strategy achieves a high Sharpe ratio and delivers significantly positive risk-adjusted returns.

#746 – Combined Value and Profitability in US and Chinese Equities

Source paper:

New research papers related to existing strategies:

#657 – Portfolio Optimization with Nonlinear Risk Budgeting using Neural Network
#542 – Committee Portfolio Selection

Chevalier, Guillaume and Coqueret, Guillaume and Raffinot, Thomas: Supervised Portfolios
https://ssrn.com/abstract=3954109
Abstract:
We propose an asset allocation strategy that engineers optimal weights before feeding them to a supervised learning algorithm. In contrast to the traditional approaches, the machine is able to learn risk measures, preferences and constraints beyond simple expected returns, within a flexible, forward-looking and non-linear framework. Our empirical analysis illustrates that predicting the optimal weights directly instead of the traditional two step approach leads to more stable portfolios with statistically better risk-adjusted performance measures.

#423 – Industry Herding and Momentum

Radi, Sherrihan and Alexandridis, Antonios and Pappas, Vasileios, Asymmetric and Cross-Asset Herding: Evidence from Bond and Equity Markets
https://ssrn.com/abstract=4050374
We examine herding in US corporate bond and equity markets between 2008-2018. Conditioning on market liquidity and volatility, we find significant asymmetric herding behavior in both markets. More specifically, we document a positive relationship between herding and market liquidity, with the level of volatility intensifying the observed herding effects. Herding is more pronounced in corporate bonds in comparison to equities. Further, we find cross-asset herding spillovers from the corporate bonds to their respective equities. The direction of the cross-asset herding effect holds during the 2008 global financial crisis, but switches post-crisis from equity to corporate bonds.

#245 – Post-Split Drift Combined with PEAD Anomaly
#315 – Stock Splits Strategy Based on Earnings Management

Liu, Mark H. and Sheather, Simon J., Predicting Stock Splits Using Ensemble Machine Learning and SMOTE Oversampling
https://ssrn.com/abstract=4033250
Abstract:
This study predicts stock splits using two ensemble machine learning techniques: gradient boosting machines (GBMs) and random forests (RFs). The goal is to form implementable portfolios based on positive predictions to generate abnormal returns. Since splits are rare events, we use SMOTE oversampling to synthesize new observations of splits in the sample to improve predictions. When predicting stock splits in the next quarter, GBM and RF achieve area under the receiver operating characteristic curve (AUC) scores of around 0.86 and 0.87, respectively. GBM and RF predictions generate monthly five-factor alphas (Fama and French, 2015) of 0.26% and 0.95% among stocks in the smallest size quintile. Three important features for predicting stock splits in both ensemble ML methods are current price levels, the ratio of current price to the price at last split, and stock returns in the past twelve months. When predicting stock splits in the next year, GBMs generate monthly five-factor alphas of 0.38% among small stocks.

#58 – VIX Predicts Stock Index Returns

Bangsgaard, Christine and Kokholm, Thomas, The Lead-Lag Relation between VIX Futures and SPX Futures
https://ssrn.com/abstract=4003464
Abstract:
We analyze the lead-lag relationship between VIX futures and SPX futures on a sample of high-frequency data. We find that the two futures markets are weakly connected when market volatility is low. In contrast, when volatility is high, their prices are highly negatively correlated and with the VIX futures leading the SPX futures. We study the determinants of the lead-lag relation, and find that an improvement in the relative liquidity of one market strengthens the lead of that market. In addition, we compute a measure of cross-market activity and find that days of high activity are associated with a strengthened VIX futures leadership. The results provide some indication that VIX futures hedging activities of dealers impact the lead-lag relation. We also document that, when dealers at an aggregate level are in a negative gamma position, an increase in SPX option dealers’ rebalancing activities further strengthens an existing VIX futures leadership.

#392 – Intraday Market-Wide Ups/Downs and Returns
#569 – Intraday Time-series Momentum in Chinese Futures
#608 – Intraday Reversal in China

Zhang, Ting and Xie, Chi and Wang, Gang-Jin and Zhou, Wei-Xing, The Speed of Convergence to Market Efficiency: New Evidence from the Chinese Stock Market
https://ssrn.com/abstract=4059042
Abstract:
We study the speed of convergence to market efficiency by examining the short-term return predictability using the high-frequency data of Chinese stocks. The empirical results reveal that it takes more than 30 minutes but less than 60 minutes for these stocks to accommodate the price pressures that arise from the persistent order imbalances, and a reversal caused by the overreactions of unsophisticated investors follows within 120 minutes. In addition, we find that large stocks converge faster than small stocks. Finally, we demonstrate that high liquidity is associated with the reduced market efficiency.

And several interesting free blog posts have been published during last 2 weeks:

The Price of Transaction Costs

Capturing the systematic premia is the main aim of many quantitative traders. However, investors tend to overlook an important factor when backtesting. Trading costs are an essential part of every trade, and yet even when we consider them, we only use an approximation. The recent article from Angana Jacob (SigTech) looks into how heavily trading costs affect the overall return of various strategies and analyzes multiple ways of implementing trading costs into the trading rules themselves.

How Does Weighting Scheme Impacts Systematic Equity Portfolios?

How often do you think about the weights of the assets in your portfolio? Do you weigh your assets equally, or do you prefer value-weighting? The researchers behind a recent research paper analyzed various weighting schemes and examined their effect on factor strategy return. They studied five weighting schemes that ignore prices: equal weighting, rank weighting, z-score weighting, inverse volatility weighting, and fundamental weighting, and three price-based weighting schemes: Rank x mcap (rank-times-mcap), Z-score x mcap (z-score-times-mcap), and Integrated core.

They found that schemes that are not based on price can inflate turnover and costs. However, the weighting schemes based on price are the most practical to target multiple premiums, provide robust risk control, and decrease turnover and expenses.