In briefNordic electricity price links and price information diverge after the energy crisis
SummaryAfter the 2021-2022 energy crisis, day-ahead electricity prices in Nordic bidding zones stopped moving together. The study measures how Norwegian and Swedish zones co-moved with Germany-Luxembourg from 2018 to 2025 and forecasts 2025 prices with one model. Southern zones stayed strongly linked to Germany-Luxembourg, while northern zones linked weakly. However, the usefulness of price information did not follow this split, and it differed mainly between national generation systems.
My Thi Diem Phan, Tuyen Trung Truong, Hoai Phuong Ha et al.
In briefDynamic scenario-based portfolio optimization for multi-asset ETFs: how trading constraints shape outcomes
SummaryThis paper compares historical moving-window returns with conditional scenarios generated by ARMA-GARCH models and a Student-t copula for 30 U.S.-listed ETFs. Mean-variance and conditional value-at-risk portfolios are built under long-only and limited long-short constraints and evaluated over 946 sessions. Historical portfolios showed higher observed reward-risk ratios in most comparisons, but no paired difference stayed significant after multiple-testing adjustment, and conditional scenarios did not consistently reduce realized downside risk. The authors separate target-weight changes from rebalancing trades and find that relaxing the target-adjustment limit affected exposures, turnover, and some returns more than changing scenario count or random seed.
Abigail A. Mensah, Ali M. Jaffri, Nicholas Appiah W. Brent Lindquist et al.
In briefTAPES: Offline Time-Grid Scheduling to Reduce Numerical Error in Financial Diffusion Samplers
SummaryFinancial diffusion samplers used with few function evaluations can have time-grid numerical error that distorts terminal quantities such as covariances and tails. The paper proposes TAPES, which estimates local errors via Richardson comparisons, propagates them to the terminal report using an adjoint, and optimizes the time grid by minimizing the squared norm of their sum. Across a regime-switching simulation and Mag7 data, TAPES gives the lowest design and audit report errors in all 24 tested rows. It changes only the offline schedule and keeps the deployment number of function evaluations unchanged.
In briefAccuracy Does Not Equal Welfare: Testing AI Shopping Agents in a Simulated Two-Seller Market
SummaryThe paper evaluates price-comparison pipelines for AI buying agents in a synthetic duopoly, using both a fixed menu and empirical correlated-equilibrium games. Calculator assistance chose the optimal offer 84.00% of the time, compared with 68.57% under verified-total disclosure, yet disclosure produced a consumer-worst equilibrium surplus 20.14 units higher. The authors argue that accuracy, loss severity, seller incentives and welfare incidence must be measured separately.
In briefHuman Capital in Context: Employment Survival and Worker-Firm Match Quality
SummaryThe paper proposes a measure of worker-firm match quality that treats fit as a property of the pair, learned from how long employment relationships survive. A two-tower neural network embeds workers and firms into a common space, with match quality given by the cosine similarity of the two vectors. Estimated on over a billion person-periods, a 0.1 increase in match quality is associated with a 48% reduction in the odds of separation.
WhyThe authors treat worker-firm match quality as a pair property learned from employment survival with a two-tower network. They report estimates on over a billion person-periods and links to wages, fir
In briefGlobal Seasonality Factor Investing: a study using calendar patterns to predict risk-adjusted returns in global futures markets
SummaryThis paper treats seasonality as a factor for predicting risk-adjusted returns across global futures in bond, commodity and equity markets. The authors fit nonlinear seasonality models for each individual market and test them with walk-forward out-of-sample evaluation. Fast-moving patterns such as day-of-week and day-of-month deliver strong gross returns but turn negative after transaction costs due to very high turnover. Slower patterns such as day-of-quarter, day-of-half-year and day-of-year remain positive net of costs, except for day-of-quarter. Adding seasonality to a global factor portfolio raises its out-of-sample Sharpe ratio by about 30%.
In briefMarket-specific effects in global time series momentum
SummaryThis paper studies time series momentum, the tendency for a market's recent average return to positively predict future returns. Building on a global nonlinear factor model, the authors fit models to individual markets and find their response curves differ from the global model. They propose a framework that lets each feature-market pair express its own predictive behaviour. Empirically, nonlinear models fitted to single markets, despite much smaller samples, are robust and can marginally improve out-of-sample performance.
In briefCompact yet flexible parameter designs for models of realized covariance matrices
SummaryThis paper proposes a new class of parameterizations for covariance models that describe how conditional covariance matrices of financial asset returns evolve over time. Each variance and covariance receives its own time-varying coefficient while the total number of parameters stays limited. The authors derive sufficient conditions that guarantee the resulting covariance matrices are positive definite. The approach is assessed through Monte Carlo simulations and an empirical application, showing it can capture complex dependence structures while remaining computationally tractable.
