There are many known relationships among macroeconomic variables in economics, while some of them are even presented as “laws”—for example, money supply and inflation or benchmark interest rates and inflation. However, the well-known economic models usually utilize only a small amount of variables. Nowadays, with the advances in machine learning and big data fields, these established models might be improved. A possible solution is presented in the research paper of Yang et al. (2020). The authors construct knowledge graphs where they connect widely recognized variables such as GDP, inflation, etc., with other more or less known variables based on the massive textual data from financial journals and research reports published by leading think tanks, consulting firms or asset management companies. With the help of advanced natural language processing, it is possible to basically “read “all the relevant published research and find the relationships among the macroeconomic variables. While this task could take years for human readers, the machine learning method can go through these texts in a much shorter time.

It is striking how cryptocurrencies are both similar and dissimilar to the more established asset classes at the same time. On the one hand, many findings from traditional asset classes also apply to this novel class. On the other hand, this “new” world with its own characteristics brings many novel “problems” that attract researchers. This week’s blog presents several research papers connected to the pump and dump schemes in cryptos. These pumps and dumps are nothing new, and we already know them from the stock market. However, there are some notable differences…

Various research papers show that market sentiment, also called investor sentiment, plays a role in market returns. Market sentiment refers to the general mood on the financial markets and investors’ overall tendency to trade. The mood on the market is divided into two main types, bullish and bearish. Naturally, rising prices indicate bullish sentiment. On the other hand, falling prices indicate bearish sentiment. This paper shows various ways to measure market sentiment and its influence on returns.

Additionally, we take a look at an overnight anomaly in combination with three market sentiment indicators. We analyse the Brain Market sentiment indicator in addition to VIX and the short-term trend in SPY ETF. Our aim is not to build a trading system. Instead, it is to analyze financial markets behaviour. Overall the transaction costs of this kind of strategy would be high. However, more appropriate than using this system on its own would be to use it as an overlay when deciding when to make trades.

Although the models cannot entirely capture the reality, they are essential in the analysis and problem solving, and the same could be said about asset pricing models. These models had a long journey from the CAPM model to the most recent Fama French five-factor model. However, the asset pricing models still rely on fundamentals, and as we see in the practice every day, the financial markets or investors are not always rational, and prices tend to deviate from their fundamental values. Past research has already suggested that the assets are driven by both the fundamentals and sentimen. The novel research of Koeppel (2021) continues in the exploration of the hypothesis mentioned above and connects the sentiment with the factors in Fama´s and French´s methodology. The most interesting result of the research is the construction of the sentiment risk factor based on the direct search-based sentiment indicators. The data are sourced by the MarketPsych that analyze information flowing on social media. For comparison, public news is not a source of such exploitable sentiment indicator.

The sentiment score extracted from social media can be exploited to augment the Fama French five factors model. Based on the results, this addition seems to be justified. Adding the sentiment to the pure fundamental model explains more variation and reduce the alphas (intercepts). Moreover, the factor is unrelated to the well-known and established risk factors utilized in the previous asset pricing models, including the momentum. Finally, the sentiment factor seems to be outperforming several other factors, even those established as the smart beta factors.

While there may be debates about passive and active investing, and even blogs about the numbers of active funds that were outperformed by the market, the history taught us that the outperformance of active or passive investing is cyclical. As a proxy for the active investing, the new Quantpedia’s research paper examines factor strategies and their smart allocation using fast or slow time-series momentum signals, the relative weights based on the strength of the signals and even blending the signals. While the performance can be significantly improved, using those smart approaches, the factors still got beaten by the market in both US and EAFE sample. However, the passive approach did not show to be superior. The factor strategies and market are significantly negatively correlated and impressively complement each other. The combined Smart Factors and market portfolio vastly outperforms both factors and market throughout the sample in both markets. With the combined approach, the ever-present market falls can be at least mitigated or profitable thanks to the factors.

Our own research paper ESG Scores and Price Momentum Are More Than Compatible utilized the Knapsack problem to make the ESG strategies more profitable or Momentum strategies significantly less risky. The implementation of the Knapsack problem was created in R, using slightly modified Simulated annealing optimization algorithm. Recently, we have been asked about our implementation and the code. The code is commented and probably could be implemented more efficiently (in R or in another programming language). For example, R is more efficient with matrices, but the code would not be that “straightforward”. Lastly, the most important tuning parameter is the temperature decrease (the probability of accepting a new solution is falling with the rising number of iterations).