Order book information is crucial for traders, but it can be complex. With the numbers of stocks listed in stock exchanges, it is impossible to track all the available information for the human mind. Therefore, the order flows could be an interesting dataset for machine learning models. The novel research of Kolm, Turiel and Westray (2021) utilizes deep-learning for high-frequency return forecasts for 115 NASDAQ stocks based on order book information at the most granular level.

The paper has several key contributions. Firstly, it does not forecast one single return but rather a whole vector of returns – a term structure consisting of mid-price return forecasts at a specified horizon. The forecasted term structure provides essential information about the most optimal execution algorithms (or a trading strategy). According to the authors, forecasts have an „accuracy peak“ at two price changes, after which the accuracy declines. Secondly, the paper compares several methods: autoregressive model with exogenous inputs, MLP, LSTM, LSTM-MLP, stacked LSTM, and CNN-LSTM. Therefore, the article could also serve as a horse race across several possible forecasting methods. Lastly, using more traditional statistical approaches, the authors have identified a better forecasting performance in more information-rich stocks. As a result, this novel research could benefit many areas such as high-frequency trading (but trading costs must be considered), optimal execution strategy, or market-making.

At the beginning of October, we plan to introduce for our Quantpedia Pro clients a new Quantpedia Pro report dedicated to clustering methods in portfolio management. The theory behind this report is more extensive; therefore, we have decided to split the introduction into our methodology into three parts. We will publish them in the next few weeks before we officially unveil our reporting tool. This first short blog post introduces three clustering methods as well as three methods that select the optimal number of clusters. The second blog will apply all three methods to model ETF portfolios, and the final blog will show how to use portfolio clustering to build multi-asset trading strategies.

Nowadays, it is a standard that fillings such as 10-Ks and 10-Qs are analyzed with machine learning models. ML models can extract sentiment, similarity metrics and many more. However, words are not everything, and we humans also communicate in other forms. For example, we show our emotions through facial expressions, but the research on this topic in finance is scarce. Novel research by Banker et al. (2021) fills the gap and examines the CEOs facial expressions during CNBC’s video interviews about corporate earnings.

Nowadays, we see an increasing number of machine learning based strategies and other related financial analyses. But can the machines replace us? Undoubtedly, AI algorithms have greater capacities to “digest” big data, but as always in the markets, everything is not rational. Cao et al. (2021) dives deeper into this topic and examines the stock analysts. Target prices and earnings forecasts are crucial parts of the investing practice and are frequently used by traders and investors (and even ML-based strategies). The novel research examines and compares the abilities of human analysts versus the AI algorithm in forecasting the target price. As a whole, AI-based analysts, on average, outperforms human analysts, but it is not that straightforward. While AI can learn from large datasets, humans do not seem to be replaced soon. There are certain fields where human uniqueness is valuable. For example, in illiquid and smaller firms or firms with asset-light business models. Moreover, it seems that rather than competing with each other, AI and human analysts are complementary. The novel technology can be used with great success to help us in areas where we lag, and the combined knowledge and forecasts of AI and humans outperform the AI analyst in each year.

Quantitative based community is represented by the Quantconnect – Algorithmic Trading Platform, where quants can research, backtest and trade their systematic strategies. Additionally, similar to Seeking Alpha, there is a possibility to follow other quants/analysts through the open free market – Alpha Market.
To our best knowledge, the literature on community/social media alpha is scarce, and this paper aims to fill this gap. In the first part, we evaluate the benchmark strategy that consists of all strategies in the alpha market that are equally weighted. Moreover, through multidimensional scaling and clustering analysis, we examine how well can significantly lower amount of strategies track the aforementioned benchmark. This could solve the problem of costly and inconvenient following of every strategy in the market. Overall, this approach can lead to a strategy that follows the benchmark with drastically reduced costs, and these strategies can be even more profitable and less volatile.

Stay tuned for the 2nd, 3rd and 4th part of this series, where we will step on the gas and explore factor meta-strategies built on top of the QuantConnect’s Alpha Market.