Journal of Investment Strategies

Ali Hirsa
Professor, Columbia University & Managing Partner, Sauma Capital LLC

Welcome to the third issue of the thirteenth volume of The Journal of Investment Strategies, which contains three research papers.

In the first paper in the issue, “Assessing the efficiency of pure-play internet banks in South Korea, Japan and China with data envelopment analysis”, Hüseyin Öcal, Erdem Bağcı and Anton Abdulbasah Kamil examine the efficiency of the leading pure-play internet banks KakaoBank, K Bank, Sony Bank, Jibun Bank and WeBank between 2017 and 2020 using a profit-oriented data envelopment analysis (DEA) model. They also analyze the year-on-year changes and use fixed cross-section regression to explore the relationship between net profit and the input and output variables. Their results reveal varying levels of efficiency over the study period. In 2017, Sony Bank operated efficiently, while the other four banks were inefficient. By 2018, WeBank and Sony Bank had become efficient and the efficiency of KakaoBank and Jibun Bank improved significantly, but K Bank remained inefficient. In 2019, KakaoBank, WeBank and Sony Bank achieved full efficiency, Jibun Bank was somewhat efficient, but K Bank continued to lag. In 2020, KakaoBank and Sony Bank maintained efficiency, K Bank remained inefficient and Jibun Bank’s efficiency actually declined slightly. The authors’ regression analysis showed the strongest relationship was between net profit and noninterest expense, underscoring the need to manage these expenses closely. The relationship with noninterest income was also significant, emphasizing the importance of balancing income and expenses to improve efficiency. Contrary to expectations, pure-play internet banks showed lower profitability than traditional banks, largely due to challenges in managing operating expenses. This suggests that such banks need more time to achieve economies of scale and operational maturity. Öcal et al recommend that less efficient internet banks focus on reducing noninterest expenses and increasing noninterest income to enhance both their efficiency and their profitability.

In the issue’s second paper, “Advanced visualization for the quant strategy universe: clustering and dimensionality reduction”, Boubacar Sidibe, Christophe de La Bastide and Florian Peres present a two-dimensional quantitative visualization of 5000 quantitative investment strategies developed by the top 18 global investment banks. This visualization was created by applying uniform manifold approximation and projection (UMAP) dimension reduction to processed returns, then examining the clustering based on the resulting embedding. Their results demonstrate the robustness of this methodology, as evidenced by the preservation of the original data structure and the accurate grouping of strategies with similar risk factor exposures. The applications of this visualization model go beyond dimension reduction and clustering. Its ability to identify nonlinear relationships and shared patterns across strategies in the quantitative investment strategies (QIS) universe allows the effective assessment of risk factor exposure. Sidibe et al’s findings show that equity strategies with similar factor properties, Sharpe ratios or geographical coverage are clustered together, offering deeper insights into the positioning of each strategy within its peer group. These insights can also be extended to entire portfolios. The advanced visualization model enables the identification of diversifying strategies within the extensive universe of quantitative strategies. Clustering analysis further supports this by projecting and visualizing clusters that can complement an existing portfolio setup, ultimately targeting greater diversification.

Our third and final paper, “Using option prices to trade the underlying asset” by Johannes Hendrik Venter, Pieter Juriaan de Jongh and Eduard Pieterse, examines how information extracted from option chain data can be used for effective trading in the underlying asset. Metrics such as expected profit and loss are condensed into a risk–reward ratio, which guides algorithmic trading rules to determine when and how to trade. The authors’ empirical results using data from the Standard & Poor’s 500 (ticker: SPX) and SPDR S&P 500 ETF Trust (ticker: SPY) demonstrate that their methodology improves both the frequency and the size of profit events compared with loss events, and it also enhances realized risk–reward performance. Additional results for the SPDR Dow Jones Industrial Average ETF Trust (ticker: DIA) and Invesco QQQ Trust (ticker: QQQ) further support these findings. The methodology of Venter et al presents several opportunities for extension. Trading rules could be refined by making batch sizes data-dependent; for example, larger batches might be advantageous on highly favorable days. Similarly, risk tolerance thresholds could adapt to market conditions, such as ongoing trends, to favor long or short trades. Determining optimal adjustments remains an open research question. On a statistical level, using a geometric Brownian motion process to model asset price dynamics may be limiting; alternatives include parametric non-Gaussian models or nonparametric methods that estimate the cumulative distribution function of asset prices directly from options data. Exciting directions for future research include exploring the impact of these models on the quality of trade information or extending the methodology to incorporate simultaneous trading in the underlying asset and options or across portfolios of multiple assets using their combined option chains.

The editorial board and I extend our gratitude to you, our valued readers, for your unwavering support and interest in our journal.We welcome the submission of practical papers from both academic and industry experts on diverse topics relating to modern investment strategies, including, perhaps, responses to the open questions referred to above.