Are There Any Simple Calendar Effects in Bitcoin Market? Wednesday, 24 January, 2018

Is Bitcoin market efficient ? A new research study analyzes simple calendar effects:

Authors: Baur, Cahill, Godfrey, Liu

Title: Bitcoin Time-of-Day, Day-of-Week and Month-of-Year Effects in Returns and Trading Volume

Link: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3088472

Abstract:

There is a large literature that reports time-specific anomalies in equity markets such as the Monday effect, the January effect and the Halloween effect. This study is the first to report intra-day time-of-day, day-of-week, and month-of-year effects for Bitcoin returns and trading volume. Using more than 15 million price and trading volume observations from seven global Bitcoin exchanges reveal time-varying effects but no consistent or persistent patterns across the sample period. The results suggest that Bitcoin markets are efficient.

Notable quotations from the academic research paper:

"The aim of this paper is to analyze market efficiency by testing for time-dependent anomalies in returns and trading volume. For example, do Bitcoin investors trade differently when major stock exchanges are open compared to when they are closed? Do they trade less on weekends and during the Northern hemisphere summer months (i.e., June - August)? To answer these types of questions, we analyze time-of-the-day (ToD), day-of-the-week (DoW) and month-of-the-year (MoY) patterns in Bitcoin returns and trading volume.

Given the conflict between efficient pricing and the seemingly irrational demand for Bitcoin, we look for any consistent patterns in returns that would contradict the efficient market hypothesis. Since Bitcoin is a relatively new and unregulated asset, it is possible that the market has been dominated by retail investors. This suggests that we can expect to find inefficiencies and return anomalies in Bitcoin pricing. The fact that Bitcoin is continuously and globally traded makes an analysis of time-of-day, day-of-week, and month-of-year effects particularly interesting to study. Moreover, since Bitcoin is traded in different currencies and in different geographic locations, the market provides an additional layer of complexity similar to foreign exchange and commodities.

Our analysis of time-of-day, day-of-week and month-of-year patterns shows evidence of time-specific anomalies such as a lower weekend volume effect and a higher Monday return effect which are more consistent with currency markets.

We find increased trading activity on Bitcoin exchanges at times when U.S stock exchanges are open and lower trading activity between midnight and the early morning on most exchanges. Bitcoin exchanges denominated in USD display stronger patterns compared to exchanges denominated in Japanese yen and Chinese yuan. We use heatmaps to illustrate patterns in returns and trading volume both across time and across exchanges.

The results support the view that Bitcoin markets are weak-form efficient because we do not see any consistent price pattern that could be exploited based on historical information. We also use statistical tests to check the robustness of the heatmap analysis and to determine the statistical significance of the effects."


Are you looking for more strategies to read about? Check http://quantpedia.com/Screener

Do you want to see performance of trading systems we described? Check http://quantpedia.com/Chart/Performance

Do you want to know more about us? Check http://quantpedia.com/Home/About

Crash Sensitivity Explains the Momentum Effect in Stocks Wednesday, 17 January, 2018

Related mainly to equity based momentum strategies:

Authors: Ruenzi, Weigert

Title: Momentum and Crash Sensitivity

Link: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3092546

Abstract:

This paper proposes a risk-based explanation of the momentum anomaly on equity markets. Regressing the momentum strategy return on the return of a self-financing portfolio going long (short) in stocks with high (low) crash sensitivity in the USA from 1963 to 2012 reduces the momentum effect from a highly statistically significant 11.94% to an insignificant 1.84%. We find additional supportive out-of sample evidence for our risk-based momentum explanation in a sample of 23 international equity markets.

Notable quotations from the academic research paper:

"Although momentum is widely documented on fi nancial markets, there is still an active ongoing debate about its main drivers and determinants. While most studies advocate a behavioral explanation for the eff ect, i.e., momentum is driven by either overreaction or underreaction of investors, recent studies point out the riskiness of momentum strategies.

Motivated by these recent results, we investigate whether momentum profi ts are driven by exposure of the momentum strategy to a systematic crash risk factor. Speci cally, we regress the momentum portfolio long-short return (UMD) on the return of a self- financing portfolio that buys stocks with high crash sensitivity and sells stocks with low crash sensitivity (CRASH) on the US stock market in the period from 1963 to 2012. The crash sensitivity of individual stocks is measured based on the lower tail dependence of their return time series with the market return time series (see Chabi-Yo, Ruenzi, and Weigert, 2017). Our results indicate that the momentum strategy loads signi ficantly positive on the crash sensitivity factor. While simultaneously controlling for the Fama and French (1993) factors, we show that including the crash sensitivity factor as an explanatory variable for the momentum return reduces its annualized alpha from a statistically signifi cant 11.94% to an insigni ficant 1.84%, i.e., a percentage decrease of almost 85%.

As an out-of-sample check we also examine the relationship between the momentum return and the crash sensitivity factor on 23 international equity markets. We find that in 22 countries (i.e., in all countries except of Singapore) momentum loads positively on systematic crash sensitivity with corresponding statistical signifi cance (at least on the 10% level) in 13 countries. Including the crash sensitivity factor as an explanatory variable in the regression setup lowers the alpha of momentum returns in 22 countries and enhances the adjusted R-square in 20 countries of our international sample. Overall, our findings show that at least a substantial part of U.S. and international momentum pro fits represents a risk premium for the exposure of the strategy to systematic crash risk."


Are you looking for more strategies to read about? Check http://quantpedia.com/Screener

Do you want to see performance of trading systems we described? Check http://quantpedia.com/Chart/Performance

Do you want to know more about us? Check http://quantpedia.com/Home/About

Summing-Up Insights into Momentum Strategies Saturday, 13 January, 2018

Related to all momentum based strategies:

Authors: Roncalli

Title: Keep Up the Momentum

Link: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3083921

Abstract:

The momentum risk premium is one of the most important alternative risk premia alongside the carry risk premium. However, it appears that it is not always well understood. For example, is it an alpha or a beta exposure? Is it a skewness risk premium or a market anomaly? Does it pursue a performance objective or a hedging objective? What are the differences between time-series and cross-section momentum? What are the main drivers of momentum returns? What does it mean when we say that it is a convex and not a concave strategy? Why is the momentum risk premium a diversifying engine, and not an absolute return strategy?

The goal of this paper is to provide specific and relevant answers to all these questions. The answers can already be found in the technical paper "Understanding the Momentum Risk Premium" published recently by Jusselin et al. (2017). However, the underlying mathematics can be daunting to readers. Therefore, this discussion paper presents the key messages and the associated financial insights behind these results.

Among the main findings, one result is of the most importance. To trend is to diversify in bad times. In good times, trend-following strategies offer no significant diversification power. Indeed, they are beta strategies. This is not a problem, since investors do not need to be diversified at all times. In particular, they do not need diversification in good times, because they do not want that the positive returns generated by some assets to be cancelled out by negative returns on other assets. This is why diversification may destroy portfolio performance in good times. Investors only need diversification in bad economic times and stressed markets.

This diversification asymmetry is essential when investing in beta strategies like alternative risk premia. On the contrary, this diversification asymmetry is irrelevant when investing in absolute return strategies. However, we know that generating performance with alpha strategies is much more difficult than generating performance with beta strategies. Therefore, beta is beautiful, but convex beta is precious and scarce. Among risk premia, momentum is one of the few strategies to offer this diversification asymmetry. This is why investing in momentum is a decision of portfolio construction, and not a search for alpha.

Notable quotations from the academic research paper:

"Key Takeaways:

The performance of momentum strategies depends on three main parameters:
   - The absolute value of Sharpe ratios
   - The correlation matrix of asset returns
   - The moving average duration to estimate the trends

Time-series momentum likes zero-correlated assets. This is why time-series momentum makes sense in a multi-asset framework.

Cross-section-momentum likes highly correlated assets. This is why cross-section momentum makes sense within a universe of homogenous assets, e.g. a universe of stocks that belong to the same region.

Short-term momentum is more risky than long-term momentum. Therefore, the cross-section dispersion of short-term momentum returns is broader than the cross-section dispersion of long-term momentum returns.

The Sharpe ratio of long-term momentum is higher than the Sharpe ratio of short-term momentum.

The choice of the moving average estimator is more crucial for short-term momentum than for long-term momentum.

Too much leverage can be harmful for the strategy, since momentum portfolios are not homothetic transformations with respect to the portfolio's leverage.

The payoff of a trend-following strategy is a long straddle option profi le. Therefore, trend-following strategies exhibit a convex payoff .

Trend-following portfolios are not absolute return strategies. In the long-run, trend-following strategies present a low moderate correlation with traditional asset classes. However, it is an illusion due to long-term averaging, since they present either a high positive or a high negative beta.

The main motivation of momentum investing is diversi fication, not performance. The convexity of trend-following strategies mitigates the risk of diversi fied portfolios in bad times. This is why momentum strategies must be located in diversifying buckets, and not in absolute return buckets. Therefore, analysing the risk/return trade-off of momentum strategies on a standalone basis does not make sense.

It follows that momentum risk premium is key for building an alternative risk premia portfolio.

"


Are you looking for more strategies to read about? Check http://quantpedia.com/Screener

Do you want to see performance of trading systems we described? Check http://quantpedia.com/Chart/Performance

Do you want to know more about us? Check http://quantpedia.com/Home/About

Deep Learning Insights for Factor Investing Wednesday, 3 January, 2018

Deep learning is a very popular area of research and is used in a lot of industries. We link to a new paper which gives interesting insights about equity factor investing:

Authors: Messmer

Title: Deep Learning and the Cross-Section of Expected Returns

Link: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3081555

Abstract:

Deep learning is an active area of research in machine learning. I train deep feedforward neural networks (DFN) based on a set of 68 firm characteristics (FC) to predict the US cross-section of stock returns. After applying a network optimization strategy, I find that DFN long-short portfolios can generate attractive risk-adjusted returns compared to a linear benchmark. These findings underscore the importance of non-linear relationships among FC and expected returns. The results are robust to size, weighting schemes and portfolio cutoff points. Moreover, I show that price related FC, namely, short-term reversal and the twelve-months momentum, are among the main drivers of the return predictions. The majority of FC play a minor role in the variation of these predictions.

Notable quotations from the academic research paper:

"This study includes cross-sectional stock data from the CRSP/Compustat database from 1970-2014. In total I use 68 published FC, constructed based on accounting and market data. The focus of the analysis lies solely on large and mid cap stocks, to prevent a potential contamination arising from economically unimportant small and micro-cap stocks.

Training a deep neural network for stock picking, is at least partly motivated by a recent important contribution to the open source software community. The so called ”Tensorflow” library provides a highly scalable and flexible machine learning framework, allowing an efficient usage of DL networks and is the core implementation of Google’s artificial intelligence (AI), unit which is under active development. Moreover, the US cross-section of returns is a relative data rich environment. In this application, roughly 2.1 million observations provide a fertile ground for these parameter rich networks.

On the other hand, it is hard to assess ex-ante if these methods are suitable for predicting stock returns, due to the inherently different statistical character they possess — for example, the signal-to-noise ratio of a stock return process is a tiny fraction compared to the processes typically encountered in computer science.

The main contribution of this work lies in the investigation if recent developments in artificial intelligence are of any use predicting cross-sectional stock returns. Applying artificial neural networks (ANN) in finance is not new. Hence, past attempts have to be distinguished. This study can be seen as an extension to earlier attempts of applying ANN to predict stock returns, with the difference of having access to additional regularization techniques, better computational resources and more data. Additionally, it is, to best of my knowledge, the first study which investigates the cross-section and its relation to a rich set of published FC by exploiting a purely data-driven algorithm without any prior assumption on the functional relation between FC and expected return spreads.

Precisely, this paper aims at answering the following main three research questions: First, how can one efficiently employ a DFN framework for the purpose of return predictions? Second, do DFN based predictions add additional economic value compared to a parsimonious linear approach utilizing the same information set? Third, which set of FC drive the prediction results and how far do they differ from recent findings in the literature of FC selection.

The first question is related to the selection problem of the optimal DFN design for this exercise. I address this question by stating the problem as an outer optimization problem. This computational intensive task is tackled by utilizing a random search algorithm as proposed in Bergstra and Bengio (2012) in combination with a one-dimensional grid search for learning rate tuning. The procedure reveals that many network designs fail to deliver reasonable numerical behavior. Despite a relatively high failure rate, I identify architectures which show promising improvements compared to the linear benchmarks based on a validation data set.

The short answer whether economically measurable improvements can be achieved, is yes. I find significant and robust factor alpha ’s, which are consistently higher compared to the parsimonious linear benchmark. In many (but not in all) cases I document significant higher Sharpe Ratios (SR). No specification favors the linear model, irrespectively of which performance measure is considered. However, a naive strategy is sensitive to trading cost adjustments for both approaches. Nonetheless, I show that a simple rebalancing frequency adjustment leads to stark improvements. An explicit rebalancing optimization is not carried out and can be seen as a limitation of this work. Over the sample period, I document that DFN based portfolios perform much weaker during high volatility periods compared to times of calmer markets, a phenomena which is characteristical for momentum based strategies. Controlling for momentum exposure during these times, levels the alpha’s significantly into the positive domain.

The answer to the question which FC drive the predictions points unambiguously at price based information, predominantly short-term reversal (providing an explanation for the turnover intensity) and the twelve-months momentum. However, I study the impact purely by looking at prediction changes arising from variation in the input data. As a result, it can not be seen as a perfect measure, but a computational trivial way in gaining model insights at this stage.

"


Are you looking for more strategies to read about? Check http://quantpedia.com/Screener

Do you want to see performance of trading systems we described? Check http://quantpedia.com/Chart/Performance

Do you want to know more about us? Check http://quantpedia.com/Home/About

Persistence in Cryptocurrencies Saturday, 30 December, 2017

Cryptocurrencies are at the moment very popular, therefore we decided to point on a short research paper showing interesting cryptos' characteristic - persistence:

Authors: Caporale, Gil-Alana, Plastun

Title: Persistence in the Cryptocurrency Market

Link: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3084023

Abstract:

This paper examines persistence in the cryptocurrency market. Two different longmemory methods (R/S analysis and fractional integration) are used to analyse it in the case of the four main cryptocurrencies (BitCoin, LiteCoin, Ripple, Dash) over the sample period 2013-2017. The findings indicate that this market exhibits persistence (there is a positive correlation between its past and future values), and that its degree changes over time. Such predictability represents evidence of market inefficiency: trend trading strategies can be used to generate abnormal profits in the cryptocurrency market.

Notable quotations from the academic research paper:

"One of the key issues yet to be analysed is whether the dynamic behaviour of cryptocurrencies is predictable, which would be inconsistent with the Efficient Market Hypothesis. Long-memory techniques can be applied for this purpose.

The present study carries out a more comprehensive analysis by considering four main cryptocurrencies (the most liquid ones: BitCoin, LiteCoin, Ripple, Dash) and applying two different long-memory methods (R/S analysis and fractional integration) over the period 2013-2017 to investigate their stochastic properties. Moreover, it also examines the evolution of persistence over time (by looking at changes in the Hurst exponent). Any predictable patterns could of course be used as a basis for trading strategies aimed at making abnormal profits in the cryptocurrency market.

pic1

As can be seen, the series do not follow a random walk, and are persistent, which is inconsistent with market efficiency. The most efficient cryptocurrency is Bitcoin, which is the oldest and most commonly used, as well as the most liquid. Degree of persistence varies over the time, and fluctuates around its average.

"


Are you looking for more strategies to read about? Check http://quantpedia.com/Screener

Do you want to see performance of trading systems we described? Check http://quantpedia.com/Chart/Performance

Do you want to know more about us? Check http://quantpedia.com/Home/About