arXiv subjects

Levin David Schwab

Publications and source records attributed to Levin David Schwab.

2 recordsLinked to original sources

arXiv2026

Market Completeness and Optional Projections under Restricted Information

In a finite discrete-time market, trading decisions may be predictable with respect to a filtration that does not adapt asset prices. The first fundamental theorem then characterizes absence of arbitrage by measures under which the optional projection of discounted prices is a martingale. We examine the corresponding completeness question. For a fixed projection, every claim measurable with respect to its terminal price history is attainable precisely when all equivalent martingale measures of that fixed process agree on the claim sigma-field. We give the finite-dimensional proof, retaining the distinction between the trading filtration and the claim sigma-field. If the projection is common to all optional martingale measures of the original prices, projected completeness implies uniqueness of their restrictions, but the converse fails even in a three-state model with a unique optional martingale measure. For the binomial model under its product martingale measure, a delay of $k$ periods yields a complete projected market with effective horizon $(T-k)^+$. An explicit replication construction and finite examples distinguish this completeness from replication at the original prices.

q-fin.MF↗

arXiv2026

Marginal Log-Likelihood Increments under Dirichlet-Smoothed Markov Estimation

For a Dirichlet-smoothed transition model, the effect of adding one workflow trace to the training archive is an exact change in reference-weighted log likelihood. We derive that change and show that it is a weighted reduction of Kullback--Leibler divergence between the reference conditionals and the model. From this form we obtain an upper bound on the gain available to any acquisition, which expresses a millinat difference as a share of what is attainable, an exact covariance identity for the effect of the reference weighting, and a sign criterion for the interaction between two candidates, from which the batch objective is neither submodular nor supermodular. A case study on the BPI Challenge 2012 loan-application log measures all three and finds a positive selection result in one of the four combinations of reference weighting and budget unit. There, of two regressors fitted to identical descriptors and identical labels, the one that predicts individual increments far more accurately, median $R^2$ 0.87 against 0.62, realizes the smaller share of the attainable gain, 61 against 69 per cent, so ranking accuracy for individual traces is neither necessary nor sufficient for batch quality.

cs.LG↗