About

Eugene Pinsky is an Associate Professor of the Practice in the Department of Computer Science at Boston University Metropolitan College. He serves as the Coordinator of Software Development. His role involves overseeing aspects of software development education and practice within the college. The page does not provide specific details about his research focus, background, or key contributions beyond his titles and administrative roles.

Research topics

• Artificial Intelligence
• Computer Science
• Machine Learning
• Natural Language Processing
• Engineering
• Operations research
• Intensive care medicine
• Surgery
• Medicine
• Internal medicine
• Management science
• Embedded system
• Medical physics

Selected publications

• Evaluating Simple Strategies with Mutual Funds and ETFs to Outperform the China’s Shanghai Composite Index (SCI)

  Journal of risk and financial management · 2026-03-28

  articleOpen accessSenior authorCorresponding

  This paper examines several portfolio rules for comparing performance against the Shanghai Composite Index. The investor can use mutual funds or sector-based Exchange-Traded funds (ETFs). Five different approaches are applied for analysis. Two core approaches are discussed in detail and compared to passive investing: The top-N strategy and the sector rotation strategy. The Top-N strategy shifts capital each period into the last period rank-N fund, and the sector rotation strategy ranks funds based on their performance in the preceding investment period, forming three baskets: “Winners”, “Median”, and “Losers”. Extensive statistical tests on more than 300 equity mutual funds are performed for the top-N strategy to evaluate performance persistence using quintile sorts, winner–loser spreads, and transition tests. In contrast, the sector-rotation strategy and a holdings-based replication strategy (constructed from annual reports and sector funds) are implemented as case studies using the ten largest funds. Their performance is evaluated using multiple return and risk metrics.

  PublisherDOI

• Clustering analysis for predictive maintenance of oil wells in Kazakhstan

  Proceedings of OilGasScientificResearchProjects Institute SOCAR · 2026-01-01

  articleSenior author

  This study explores the use of unsupervised machine learning to optimize operational efficiency and reduce downtime in oil well production through clustering analysis. Using production data from a major Kazakhstani oil operator, the study applies K-means clustering to categorize oil wells based on output patterns. The dataset includes more than three million entries over a two-year period, capturing daily values for oil volume, liquid volume, and water cut percentage. Data preprocessing involved normalization, outlier handling, and correlation analysis to ensure robust clustering performance. The optimal number of clusters (K = 10) was selected using the Elbow method. Cluster interpretations revealed distinct operational profiles, including stable, declining, and volatile well behavior. In particular, more than 70% of wells remained in their assigned clusters over time, suggesting temporal stability and operational consistency. By identifying risk-prone and underperforming wells, the results support proactive maintenance, informed resource allocation, and long-term planning. The study demonstrates how unsupervised learning techniques can support data-driven decision-making in the petroleum industry without reliance on expensive sensor infrastructure. This research provides a replicable framework for oilfield analysis and highlights the value of behavioral clustering as a predictive tool for operational optimization. Our methodology integrates unsupervised learning, dimensionality reduction techniques, and visual analytics to provide interpretable cluster groupings. The enhanced predictive model demonstrates the potential to support data-driven decision-making and resource allocation in oilfield operations. Keywords: machine learning; K-means clustering; predictive maintenance; oil well monitoring; statistical analysis.

  PublisherDOI

• Sector Rotation Strategies in the TSX 60: A Comprehensive Analysis of Risk-Adjusted Returns, Machine Learning Applications, and Out-of-Sample Validation (2000–2025)

  Journal of risk and financial management · 2026-01-15

  articleOpen accessSenior author

  We investigate the profitability of systematic sector rotation strategies in the Canadian equity market using TSX 60 constituents (2000–2025). Testing 72 distinct strategies across three theoretical frameworks—momentum, mean-reversion, and balanced approaches—with varying rebalancing frequencies, we identify that median-performer selection combined with quarterly rebalancing generates statistically significant risk-adjusted returns (Sharpe ratio 0.922 versus 0.624 for equal-weighted buy-and-hold). Our primary contributions include rigorous out-of-sample validation, demonstrating performance persistence from 2020 to 2025, machine learning regime classification with 72.7% accuracy, and a comprehensive transaction cost analysis. Results support intermediate-horizon mean reversion in sector returns and challenge strict efficient market hypothesis interpretations in concentrated markets. Findings inform tactical asset allocation practices and contribute to the momentum-reversal literature by documenting conditions under which rotation strategies generate economically meaningful alpha.

  PublisherOA PDFDOI

• Algorithmic Insights into Human Irrationality: Machine Learning Approaches to Detecting Cognitive Biases and Motivated Reasoning

  Machine Learning and Knowledge Extraction · 2026-04-11

  articleOpen accessSenior authorCorresponding

  This study illuminates fundamental questions in behavioral science through advanced machine learning methodologies applied to large-scale public opinion data. Drawing on Kahneman and Tversky’s dual-process theory and Sunstein’s nudge architecture, we employ hierarchical unsupervised clustering and supervised predictive models to detect cognitive biases—loss aversion, availability heuristic, and partisan motivated reasoning—embedded within a nationally representative survey of 5022 American respondents. Our primary methodological contribution is a hierarchical two-stage clustering framework that uncovers latent opinion structures without imposing a priori partisan categories, permitting discovery of cross-cutting cleavages invisible to conventional survey analysis. Three principal findings emerge: (1) loss aversion is empirically confirmed in prospective economic perception, with pessimists outnumbering optimists at a 1.14:1 ratio even among respondents rating current conditions positively; (2) partisan motivated reasoning produces a 13.15 percentage-point perception gap among individuals with identical financial circumstances; and (3) multi-platform digital engagement is associated with reduced partisan bias, providing evidence that challenges simple echo chamber assumptions. Crime safety perception emerges as the strongest predictor of economic bias, surpassing party affiliation, and substantiating availability heuristic dominance in political cognition. These findings carry implications for democratic accountability, platform governance, and the ethics of AI-augmented behavioral analysis in an era of affective polarization.

  PublisherOA PDFDOI

• The Carbon Cost of Intelligence: A Domain-Specific Framework for Measuring AI Energy and Emissions

  Energies · 2026-01-26 · 1 citations

  articleOpen accessSenior author

  The accelerating energy demands from artificial intelligence (AI) deployment introduce systemic challenges for achieving carbon neutrality. Large language models (LLMs) represent a dominant driver of AI energy consumption, with inference operations constituting 80–90% of total energy usage. Current energy benchmarks report aggregate metrics without domain-level breakdowns, preventing accurate carbon footprint estimation for workloadspecific operations. This study addresses this critical gap by introducing a carbon-aware framework centered on the carbon cost of intelligence (CCI), a novel metric enabling workload-specific energy and carbon calculation that balances accuracy and efficiency across heterogeneous domains. This paper presents a comprehensive cross-domain energy benchmark using the massive multitask language understanding (MMLU) dataset, measuring accuracy and energy consumption in five representative domains: clinical knowledge (medicine), professional accounting (finance), professional law (legal), college computer science (technology), and general knowledge. Empirical analysis of GPT-4 across 100 MMLU questions, 20 per domain, reveals substantive variations: legal queries consume 4.3× more energy than general knowledge queries (222 J vs. 52 J per query), while energy consumption varies by domain due to input length differences. Our analysis demonstrates the evolution from simple ratio-based approaches (weighted accuracy divided by weighted energy) to harmonic mean aggregation, showing that the harmonic mean, by preventing bias from extreme values, provides more accurate carbon usage estimates. The CCI metric, calculated using weighted harmonic mean (analogous to P/E ratios in finance, where A/E represents accuracy-to-energy ratio), enables practitioners to accurately estimate energy and carbon emissions for specific workload mixes (e.g., 80% medicine + 15% general + 5% law). Results demonstrate that the domain workload mix significantly impacts carbon footprint: a law firm workload (60% law) consumes 96% more energy per query than a hospital workload (80% medicine), representing 49% potential savings through workload optimization. Carbon footprint analysis using US Northeast grid intensity (320 gCO2e/kWh) shows domain-specific emissions ranging from 0.0046–0.0197 gCO2 per query. CCI is validated through comparison with simple weighted average, demonstrating differences up to 12.1%, confirming that the harmonic mean provides more accurate and conservative carbon estimates essential for carbon reporting and neutrality planning. Our findings provide a novel cross-domain energy benchmark for GPT-4 and establish a practical carbon calculator framework for sustainable AI deployment aligned with carbon neutrality goals.

  PublisherOA PDFDOI

• A Comparative Analysis of Overnight vs. Daytime Static and Momentum Strategies Across Sector ETFs

  Risks · 2026-04-08

  articleOpen accessSenior authorCorresponding

  This study examines overnight vs. daytime static and momentum strategies applied to ten sector Exchange-traded funds (ETFs) over a 27-year period from 1999 to 2025. Our findings reveal that several such strategies, particularly reversal strategies, consistently outperform static and buy-and-hold strategies. This outperformance decreases significantly when transaction costs are taken into account. We consider two transaction-cost scenarios (1 bps vs. 2 bps), which are industry standards for institutional and retail investors, respectively. We provided a detailed analysis of volatility and drawdowns. Our results indicate that by considering night and daytime separately, it is possible to outperform passive strategies for most sector ETFs.

  PublisherOA PDFDOI

• Duration Rotation in U.S. Treasury Fixed-Income ETFs: Evidence for a “Median” Strategy

  FinTech · 2026-04-07

  articleOpen accessSenior authorCorresponding

  We examine a simple duration-rotation strategy applied to six U.S. Treasury ETFs spanning the full maturity spectrum, using data from 2007 to 2025. At each semi-annual rebalancing date, ETFs are ranked by prior-period return and divided into three equal groups—Winners, Median, and Losers. Contrary to conventional momentum logic, the middle group consistently outperforms. The Median strategy grows USD 100 to USD 199.90 by end-2025, a CAGR of 3.79% against 2.17% for the passive benchmark, with a higher Sharpe ratio (0.606 vs. 0.494) and a shallower maximum drawdown (−11.6% vs. −14.4%). Newey–West HAC and Lo (2002) tests confirm statistical significance (p=0.031 and p=0.014), and an expanding-window walk-forward procedure yields p=0.0005 across 27 out-of-sample evaluations from 2012 to 2025. The result is robust to calendar alignment, evaluation endpoint, lookback window, and execution timing, and survives transaction costs by a wide margin. The strategy requires no interest rate forecasts, no proprietary data, and is implementable with standard ETF brokerage access.

  PublisherOA PDFDOI

• Comparing “Winners,” “Median,” and “Losers” Rotation Strategies with S&P 500 Sector ETFs

  The Journal of Investing · 2026-02-12

  articleSenior author

  This article proposes a simple periodic ETF sector rotation strategy for S&P 500. Unlike most ETF rotation strategies, the proposed strategy does not predict economic cycles. The strategy involves periodic ranking of component sectors based on sector ETF returns and reinvesting funds equally into the middle (median) three sector ETFs. We show that for the S&P 500, the proposed “median” ETF rotation strategy with monthly rebalancing is better than focusing on the winners- or the losers-sector ETFs with different frequencies of rebalancing. Using historical data from 2000 to 2024, we show that the proposed median monthly rotation strategy significantly outperforms the other two strategies and passive index investment in terms of total return, volatility, and maximum drawdown. An average investor can easily implement the proposed ETF rotation strategy.

  PublisherDOI

• Estimation of distribution parameters by mean absolute deviations of a truncated distribution using quantile functions

  Statistical Papers · 2026-02-18

  article1st authorCorresponding
  PublisherDOI

• Resilient Anomaly Detection in Ocean Drifters with Unsupervised Learning, Deep Learning Models, and Energy-Efficient Recovery

  Oceans · 2026-01-06

  articleOpen accessSenior author

  Changes in climate and ocean pollution has prioritized monitoring of ocean surface behavior. Ocean drifters, which are floating sensors that record position and velocity, help track ocean dynamics. However, environmental events such as oil spills can cause abnormal behavior, making anomaly detection critical. Unsupervised learning, combined with deep learning and advanced data handling, is used to detect unusual behavior more accurately on the NOAA Global Drifter Program dataset, focusing on regions of the West Coast and the Mexican Gulf, for time periods spanning 2010 and 2024. Using Density-Based Spatial Clustering of Applications with Noise (DBSCAN), pseudo-labels of anomalies are generated to train both a one-dimensional Convolutional Neural Network (CNN) and a Long Short-Term Memory (LSTM) network. The results of the two models are then compared with bootstrapping with block shuffling, as well as 10 trials with bar chart summaries. The results show nuance, with models outperforming the other in different contexts. Between the four spatiotemporal domains, a difference in the increasing rate of anomalies is found, showing the relevance of the suggested pipeline. Beyond detection, data reliability and efficiency are addressed: a RAID-inspired recovery method reconstructs missing data, while delta encoding and gzip compression cut storage and transmission costs. This framework enhances anomaly detection, ensures reliable recovery, and reduces energy consumption, thereby providing a sustainable system for timely environmental monitoring.

  PublisherOA PDFDOI

Frequent coauthors

• Saveli Goldberg

  Massachusetts General Hospital

  9 shared

• Adrian E. Conway

  Assured Information Security (United States)

  9 shared

• Lev Salnikov

  6 shared

• Sarthak Pattnaik

  Boston University

  5 shared

• Kathleen Park

  Center for Global Development

  5 shared

• Wei-hsing Wang

  Princeton Public Schools

  3 shared

• Sidney Klawansky

  Harvard University

  3 shared

• Maryan Rizinski

  Boston University

  3 shared

Education

• Ph.D.

  Columbia University

• B.A.

  Harvard University