Quantpedia Premium Update – 17th December 2021

New strategies:

Period of rebalancing: Monthly
Markets traded: bonds
Instruments used for trading: bonds
Complexity: Simple strategy
Backtest period: 2002-2020
Indicative performance: 3%
Estimated volatility: 4.29%

Source paper:

Kaufmann, Hendrik and Messow, Philip: Putting Credit Factor Investing into Practice
https://ssrn.com/abstract=3791602
Abstract:
Leveraged ETFs and market makers who are active in option markets must adjust imbalances arising from market movements. Establishing delta-neutrality may cause either return momentum or reversal depending on the sign and size of the imbalance vis-a-vis market prevailing liquidity. We ﬁnd that a large and negative (positive) aggregated gamma imbalance, relative to the average dollar volume, gives rise to an economically and statistically signiﬁcant end-of-day momentum (reversal). We compare this channel to the rebalancing of leveraged ETFs and ﬁnd that the eﬀect generated by leveraged ETFs is economically larger. Consistent with the notion of temporary price pressure, the documented eﬀects quickly revert at the next day’s open. Information-based explanations are unlikely to cause the results, suggesting a non-informational channel through which leveraged ETFs and option markets aﬀect underlying stocks towards the market close.

#695 –Size Factor in Corporate Bonds

Source paper:

#696 –Fair Spread Value Factor in Corporate Bonds

Period of rebalancing: Monthly
Markets traded: bonds
Instruments used for trading: bonds
Complexity: Very complex strategy
Backtest period: 2002-2020
Indicative performance: 2.1%
Estimated volatility: 2.1%

Source paper:

#697 –Multifactor Corporate Bond Strategy

Source paper:

#698 – Liquidity Volatility in Stocks

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Moderately complex strategy
Backtest period: 1990-2015
Indicative performance: 3.86%
Estimated volatility: 3.72%

Source paper:

Frank Yulin Feng, Wenjin Kang, Huiping Zhang, Liquidity Shocks and the Negative Premium of Liquidity Volatility Around the World
https://ssrn.com/abstract=3930591
Abstract:
We find that liquidity volatility negatively predicts stock returns in global markets. This relationship holds for different liquidity measures and cannot be explained by the idiosyncratic volatility effect. This puzzle can be explained by the asymmetric impact of liquidity increase and decrease on expected returns. Since the price decline following liquidity decrease outweighs the price appreciation after liquidity increase, high-liquidity-volatility stocks, which are more likely to experience large liquidity changes in either direction, tend to have negative returns on average. We find that including liquidity decrease explains the negative premium of liquidity volatility, while including liquidity increase does not.

#699 – Stock and Bond Returns Predict Currency Returns

Period of rebalancing: Monthly
Markets traded: currencies
Instruments used for trading: futures, forwards, CFDs
Complexity: Moderately complex strategy
Backtest period: 1989-2019
Indicative performance: 8.47%
Estimated volatility: 3.88%

Source paper:

Phylaktis, Kate and Yamani, Ehab Abdel-Tawab, Foreign Currency Forecasting: What Can Stock and Bond Markets Tell Us?
https://ssrn.com/abstract=3775939
Abstract:
This paper provides a comprehensive statistical and economic evidence on the forecasting power of local-currency equity and bond returns in predicting exchange rate returns. We first construct out-of-sample (OOS) forecasts using various model specifications of equity and bond returns, and assess their statistical accuracy against the naïve random walk (RW) model. Next, we test their economic value by designing a trading strategy based on the sign of our OOS forecasts. Using a sample of 28 countries, we find three key results: (1) our in-sample forecasts show that all the slope estimates of our predictive regressions are significant, indicating that the RW model is misspecified; (2) the statistical accuracy criteria suggest that our OOS forecasts outperform the RW model; and (3) our trading strategies, that use stock and bond returns, generate high and significant Sharpe ratios which are uncorrelated with the traditional risk factors and other popular determinants of currency returns.

#700 –Expected Options Return Predictability Using Machine Learning

Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: options
Complexity: Very complex strategy
Backtest period: 2001-2020
Indicative performance: 31.52%
Estimated volatility: 15.29%

Source paper:

Bali, T. G., Beckmeyer, H., Moerke, M., & Weigert: Option Return Predictability with Machine Learning and Big Data
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3895984
Abstract:
We investigate the cross-sectional return predictability of delta-hedged equity options using machine learning and big data. Drawing upon more than 12 million observations over the period from 1996 to 2020, we find that allowing for nonlinearities significantly increases the out-of-sample performance of option and stock characteristics in predicting future option returns. Besides statistical significance, the nonlinear machine learning models generate economically sizeable profits in the long-short portfolios of equity options even after accounting for transaction costs. Although option-based characteristics are the most important standalone predictors, stock-based measures offer substantial incremental predictive power when considered alongside option-based characteristics. Finally, we provide compelling evidence that option return predictability is driven by informational frictions, costly arbitrage, and option mispricing.

#701 – Rebalancing Premium in Cryptocurrencies

Period of rebalancing: Daily
Markets traded: cryptos
Instruments used for trading: cryptos
Complexity: Simple strategy
Backtest period: 2018-2021
Indicative performance: 7.65%
Estimated volatility: 2.61%

Source paper:

Hanicova, Daniela and Vojtko, Radovan, Rebalancing Premium in Cryptocurrencies
http://ssrn.com/abstract=3982120
Abstract:
This paper analyzes the rebalancing premium in cryptocurrencies. Rebalancing premium is defined as the premium an investor gains from periodically rebalancing their portfolio. Many papers examine this effect; however, very few to none study crypto markets.
In the first section of this article, we generate 30 vectors, each with the mean daily return equal to 0 and daily volatility set to 7.5%, to simulate cryptocurrencies. Firstly, we created two portfolios, a buy and hold portfolio and a daily rebalanced portfolio. Multiple improvements can be observed when the portfolio is rebalanced periodically, including improved cumulative return and lowered volatility, resulting in a better Sharpe ratio. Moreover, the drawdowns are not as significant when applying daily rebalancing.
The second section focuses on real data. We analyzed a portfolio consisting of 27 cryptocurrencies and compared daily- and monthly-rebalanced portfolios with benchmark buy-and-hold portfolio. We also compare the daily-rebalanced long-only portfolio to long-short portfolio with various weights on the short side.
Moreover, in this section, we also consider the combination of cryptos with bonds. We assume that the volatility of cryptocurrencies combined with low volatile assets such as bonds could raise the overall performance while keeping the volatility at a reasonable level.
Overall, we can say that there is a significant rebalancing premium in cryptocurrencies. Moreover, the crypto portfolio can also improve the performance of less volatile assets, such as bonds, when combined in a periodically rebalanced portfolio. Lastly, the functionality is based on a proposition that no single asset consistently outperforms the others. The rebalancing might not be as profitable in such a case.

New research papers related to existing strategies:

#70 – Combining Post-Earnings Announcement Drift with Accrual Anomaly

Ballas, Apostolos A. and Hevas, Dimosthenis (Dimosthenes) L. and Karampinis, Nikolaos I. and Vlismas, Orestes: Strategy and Earnings Quality
https://ssrn.com/abstract=3938309
Abstract:
This study explores the relationship between strategy and earnings quality. We focus on firms classified as prospectors or defenders, and we analyze the effects of different strategic choices on various attributes of earnings quality: (accrual and real) earnings management, conditional conservatism, and earnings persistence. Our sample comprises 30 countries for the period 2005–2019 and includes 120,205 firm-year observations. It seems that business strategy is significant-ly linked to earnings quality metrics. Specifically, prospectors systematically exhibit higher val-ues of absolute discretionary accruals, lower abnormal operating cash flows, lower abnormal production costs, and higher discretionary expenses. Moreover, prospectors are more condition-ally conservative and have more sustainable earnings.

#77 – Betting Against Beta Factor in Stocks

Drobetz, Wolfgang and Hollstein, Fabian and Otto, Tizian and Prokopczuk, Marcel and Prokopczuk, Marcel: Estimating Security Betas via Machine Learning
https://ssrn.com/abstract=3933048
Abstract:
This paper evaluates the predictive performance of machine learning techniques in estimating time-varying betas of US stocks. Compared to established estimators, tree-based models and neural networks outperform from both a statistical and an economic perspective. Random forests perform the best overall. Machine learning-based estimators provide the lowest fore-cast errors. Moreover, unlike traditional approaches, they lead to truly ex-post market-neutral portfolios. The inherent model complexity is strongly time-varying. The most important predictors are various historical betas as well as fundamental turnover and size signals. Compared to linear regressions, interactions and nonlinear effects enhance the predictive performance substantially.

#77 – Betting Against Beta Factor in Stocks

Hamaui, Andrea and Jaffard, Pierre: Chasing Beta, Losing Alpha
https://ssrn.com/abstract=3816944
Abstract:
In this paper, we tackle the Beta anomaly, namely the fact that high-Beta assets tend to be associated with lower risk-adjusted returns than low-Beta assets, and connect it to mutual funds’ expectations. We present a model with two types of investors, mutual funds and hedge funds, with heterogeneous market expectations and margin constraints. We show that the Beta anomaly is especially present for stocks purchased by over-optimistic mutual funds. On the empirical side, we first introduce a mutual fund-level measure of market expectations. Then, portfolio analyses and regressions confirm the model’s prediction. The results are robust to alternative definitions of the mutual funds’ market beliefs variable that correct for stock picking, and carry predictive power for mutual funds’ returns.

#685 – Boosted Trees and Cryptocurrency Return Prediction

Cocco L, Tonelli R, Marchesi M.: Predictions of bitcoin prices through machine learning based frameworks
https://doi.org/10.7717/peerj-cs.413
Abstract:
The high volatility of an asset in financial markets is commonly seen as a negative factor. However short-term trades may entail high profits if traders open and close the correct positions. The high volatility of cryptocurrencies, and in particular of Bitcoin, is what made cryptocurrency trading so profitable in these last years. The main goal of this work is to compare several frameworks each other to predict the daily closing Bitcoin price, investigating those that provide the best performance, after a rigorous model selection by the so-called k-fold cross validation method. We evaluated the performance of one stage frameworks, based only on one machine learning technique, such as the Bayesian Neural Network, the Feed Forward and the Long Short Term Memory Neural Networks, and that of two stages frameworks formed by the neural networks just mentioned in cascade to Support Vector Regression. Results highlight higher performance of the two stages frameworks with respect to the correspondent one stage frameworks, but for the Bayesian Neural Network. The one stage framework based on Bayesian Neural Network has the highest performance and the order of magnitude of the mean absolute percentage error computed on the predicted price by this framework is in agreement with those reported in recent literature works.

#685 – Boosted Trees and Cryptocurrency Return Prediction

Thomas E. Koker and Dimitrios Koutmos: Cryptocurrency Trading Using Machine Learning
https://www.mdpi.com/1911-8074/13/8/178/pdf
Abstract:
We present a model for active trading based on reinforcement machine learning and apply this to five major cryptocurrencies in circulation. In relation to a buy-and-hold approach, we demonstrate how this model yields enhanced risk-adjusted returns and serves to reduce downside risk. These findings hold when accounting for actual transaction costs. We conclude that real-world portfolio management application of the model is viable, yet, performance can vary based on how it is calibrated in test samples.

#685 – Boosted Trees and Cryptocurrency Return Prediction

Helder Sebastião and Pedro Godinho: Forecasting and trading cryptocurrencies with machine learning under changing market conditions
https://jfin-swufe.springeropen.com/track/pdf/10.1186/s40854-020-00217-x.pdf
Abstract:
This study examines the predictability of three major cryptocurrencies—bitcoin, ethereum, and litecoin—and the proftability of trading strategies devised upon machine learning techniques (e.g., linear models, random forests, and support vector machines). The models are validated in a period characterized by unprecedented turmoil and tested in a period of bear markets, allowing the assessment of whether the predictions are good even when the market direction changes between the validation and test periods. The classifcation and regression methods use attributes from trading and network activity for the period from August 15, 2015 to March 03, 2019, with the test sample beginning on April 13, 2018. For the test period, fve out of 18 individual models have success rates of less than 50%. The trading strategies are built on model assembling. The ensemble assuming that fve models produce identical signals (Ensemble 5) achieves the best performance for ethereum and litecoin, with annualized Sharpe ratios of 80.17% and 91.35% and annualized returns (after proportional round-trip trading costs of 0.5%) of 9.62% and 5.73%, respectively. These positive results support the claim that machine learning provides robust techniques for exploring the predictability of cryptocurrencies and for devising proftable trading strategies in these markets, even under adverse market conditions.

#626 – Image Recognition in Stock Price Charts Predicts Stock Returns

Komori, Yohei: Convolutional Neural Network for Stock Price Prediction Using Transfer Learning
https://ssrn.com/abstract=3756702
Abstract:
The goal of this paper is to build a trading algorithm by applying image recognition neural network – Convolutional Neural Network(CNN) – to the 2D technical candle stick charts. First, this paper shows a research survey of the previous paper. Second, this paper explains the basic theory of CNN model and how it can works on chart images. Next, this project performs an experimental study of CNN on S&P 500 index from January 1, 1985 to June 30, 2020. The CNN model structure used in this paper is transferred from inception v3 with three additional layers, and the technical indicators used in the input chart image are simple moving average (25 days). The label data used in the model are categorical – either up, flat, or down. The model has 50% accuracy on the test set when conducting three-days ahead forecast, which is higher than the simple momentum strategy and contrarian strategy, indicating its high alpha generating potential. One-day ahead forecast and five-days ahead forecast have lower accuracy than the three-days forecast. This means you might have the best performance when you close your position at T + 3.

#459 – Machine Learning and Currency Carry Strategy

Deniz Can Yıldırım: FORECASTING DIRECTIONAL MOVEMENT OF FOREX DATA USING LSTM WITH TECHNICAL AND MACROECONOMIC INDICATORS
https://etd.lib.metu.edu.tr/upload/12623270/index.pdf
Abstract:
Foreign Exchange is known as Forex or FX is a financial market where currencies are bought and sold simultaneously. Forex is the largest financial market with more than $5 trillion volume. It is a decentralized market that is operational 24 hours in a day other than weekends which makes different from other markets. Fundamental and Technical Analysis are the two techniques that are commonly used in predicting the future prices in Forex. Fundamental Analysis concentrates on the economical, social and political factors that can cause to price moving higher, lower or staying the same. Technical analysis, on the other hand, is based on only the price to predict the future price movements. It studies the effect of the price movement by using technical indicators. In this thesis, a model that uses LSTM with technical and macroeconomic indicators is proposed to forecast directional movement of Forex data. It is based on the two LSTM models that learn the effects of both indicators individually. The predictions of two LSTM models are combined according to the predefined set of rules in order to determine the final decision. The experiments are conducted on EUR/USD currency pair to forecast 1-day, 3-days and 5-days ahead and promising results are succeeded.

#669 – Volatility Risk Premium in Currencies 2

Bruno Feunou, Ricardo Lopez Aliouchkin, Roméo Tédongap and Lai Xu: Variance Premium, Downside Risk, and Expected Stock Returns
http://wp.lancs.ac.uk/fofi2018/files/2018/03/FoFI-2018-0150-Rom%C3%A9o-Tedongap.pdf
Abstract:
We decompose total variance into its bad and good components and measure the premia associated with their fluctuations using stock and option data from a large cross-section of firms. The total variance risk premium (VRP) represents the premium paid to insure against fluctuations in bad variance (called bad VRP), net of the premium received to compensate for fluctuations in good variance (called good VRP). Bad VRP provides a direct assessment of the degree to which asset downside risk may become extreme, while good VRP proxies for the degree to which asset upside potential may shrink. We find that bad VRP is important economically; in the cross-section, a one-standard-deviation increase is associated with an increase of up to 13% in annualized expected excess returns. Simultaneously going long on stocks with high bad VRP and short on stocks with low bad VRP yields an annualized risk-adjusted expected excess return of 18%. This result remains significant in double-sort strategies and cross-sectional regressions controlling for a host of firm characteristics and exposures to regular and downside risk factors.

#228 – Google Search Volume Combined with Extent of Press News Predicts Stocks’ Returns

Kuhlen, Nikolas and Preston, Andrew: News Entropy
https://ssrn.com/abstract=3820449
Abstract:
We introduce the concept of ‘news entropy’ to characterise the relationship between news coverage and the economy. Intuitively, news entropy decreases as the news focus on a smaller set of pressing topics. We observe that news entropy exhibits clear negative spikes close to important economic, financial, and political events. Investigating the effect of changes in news entropy, we find that decreases are associated with two key features: an increase in uncertainty measures and a macroeconomic contraction. The variable is priced in the cross-section of stock returns and low news entropy is associated with increased stock price volatility at the firm level.

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

Synthetic Lending Rates Predict Subsequent Market Return

It is indisputable that the data are changing financial markets – computing power has increased, allowing to rise the trends of ML/AI and big data (number of possible predictors or granularity) or HFT strategies. Indeed, not all the datasets are worth the time of academics, investors or traders, but we are always keen to analyze the novel and unique datasets. Of course, if we believe that the analysis is worthy of sharing, we are happy to do so. This post offers a shorter version of our newest research about Synthetic lending rates and subsequent market return. We hope that you find it enriching; enjoy the reading!

Estimating Rebalancing Premium in Cryptocurrencies

Our new article investigates “rebalancing premium” or “diversification return” in cryptocurrencies which can be achieved by periodically rebalancing portfolios. We analyze whether the daily/ monthly rebalanced portfolios outperform a simple buy-and-hold portfolio of cryptocurrencies and under which conditions. Additionally, we also look at the various combinations of volatile cryptocurrency portfolios with low-risk bonds.

NFTs: Important Preliminary Risk and Return Analysis

NFTs are taking the cryptocurrency trading world by storm. NFTs stand for the non-fungible tokens which have emerged as another possible usage of blockchain technology. NFT can be used to record/verify/track the ownership of a unique – hence the non-fungible asset. Commonly, NFTs are connected with art (visual art, music, etc.), but there are also several decentralized finance or gaming-related projects.
Same as for the other blockchain-related projects, the critics are easy to find, so a research paper with hard data concerning the NFTs can be of great importance. The research paper by Mazur (2021) studies the NFT startups traded in the crypto markets. Therefore, the paper does not analyze the individual NFTs (such as some piece of art), but rather the whole projects and their tokens traded on the Binance crypto exchange.