Curated items for this scope are still being collected. Below are the latest related updates; items meeting the curation criteria will be marked separately.
This study focuses on the allocation of bank default losses without external capital injection, proving that for all non-decreasing concave terminal payoffs, the "tax the richest" truncation rule is optimal. The study constructs a continuous-time optimal solution through time discretization, characterizes the optimal flow as a novel reflected free boundary problem, and proves the positive time smoothness of the free boundary and the density of surviving banks.
The paper proposes using the half-width of conformal prediction intervals as a scaling measure for fractional Kelly position sizing with a leverage cap, and tests this rule under a pre-registered protocol. During the development phase, an autonomous LLM agent searched approximately 200 configurations on a 2016-2021 sample of 8 U.S. equity ETFs, with subsequent data sealed before the search, and the final configuration, 9 comparison portfolios, and conclusion mapping all registered before unsealing.
This research in quantitative financial risk management proves that when estimates remain unchanged after replicating all scenarios, binary estimators can accurately describe the same population risk and recover the population dual set from finite models. The study derives Wasserstein, Kolmogorov, and total variation norms under bounded support based on Bernoulli loss risk, characterizes uniform estimability, provides impossibility bounds for arbitrary stochastic estimators, and further derives uniform regret bounds for data-driven risk-averse optimization, with applications in stop-loss retention scenarios.
For a discrete-time financial market model with trades having instantaneous price impact, the existence of a solution to the corresponding utility maximization problem is proven. This study removes the unnatural restrictions on market depth and resilience processes present in earlier work, with the non-standard feature of this problem being that the set of attainable portfolio values may not satisfy convexity.
A new arXiv paper uses Gaussian surrogate models to precisely calculate the finite-sample risk of candidate scoring decision models, analyzing the boundaries of the role of three training elements: teacher soft targets, frozen pre-trained backbones, and temperature calibration. The paper points out that after a student model inherits teacher bias, more student data cannot eliminate that bias; fixed representations only reduce risk below an explicit sample size threshold; under the probit model, exact recalibration maps to temperature scaling, and can only eliminate all excess risk when the direction of the teacher's coefficients is correct.
This arXiv paper tests the degree to which 4 AI trading agents adhere to their own strategies and risk rules, based on 2627 closed trades from February 12 to August 27, 2026. 18% of the trades were live orders on Hyperliquid, while the rest were simulated at real-time prices, with a uniform round-trip fee of 0.09%.
This arXiv paper proposes a conditional weight of evidence method to measure the marginal impact of a single credit feature on the log odds of default, given the rest of an applicant's credit information. The same feature value can yield different evaluations for applicants with different credit backgrounds.
This study is the first to prove the validity of dynamic programming equations for Value-at-Risk Markov Decision Processes (VaR MDPs) in continuous spaces, covering both finite-horizon and infinite-horizon scenarios. Previously, related dynamic programming equations had only been derived and verified in finite state and action spaces, leaving the continuous-space problem unresolved for a long time. This framework can be used to characterize optimal dynamic strategies for optimal execution problems with VaR objectives, solving an open problem in the field.
This arXiv paper proposes constructing probability weighting functions from changes in return distributions rather than parameter families, with the formula w(p)=F_post(F_prior^{-1}(p)). Both the prior and posterior are normal tempered-stable distributions, and the posterior differs from the prior along three independent channels—scale, skewness, and tail thickness—under the premise of fixed variance.
The paper proposes a Neural Filter Estimator (NFE), trained on simulated paths of mean-reverting processes, that outputs the posterior distribution of the implied state for each spread in the cross-section in a single forward pass, incorporating estimation uncertainty into trading decisions. On simulated paths with known ground truth, the method achieves accuracy comparable to other practical estimators, with its advantage lying in correctly characterizing its own uncertainty.
This paper proposes AddMart, a few-shot implied volatility surface reconstruction framework that embeds no-arbitrage structure into the architecture to achieve global static no-arbitrage, enabling continuous and valid surface generation from a small number of option quotes. Experiments on 10 real-world option datasets show that the framework delivers competitive reconstruction accuracy with no significant static arbitrage violations, and can be directly applied to VaR, CVaR and crash probability estimation.
This paper addresses the stationary quadratic trading problem with transient price impact, deriving optimal trading strategies using predictive theory tools, expressed as a combination of conditional predictions and explicit causal linear filters. The paper recovers the Gârleanu-Pedersen target portfolio under the exponential elasticity assumption, derives fractional trading filters under power-law impact decay, and clarifies the mechanisms determining the net realizable value of predictive signals under different trading cost scenarios.
