Erik Postnieks

I study how autonomous agents, firms, and institutions optimize when consequential system effects remain outside their objective functions. My work combines economic theory, reproducible computational experiments, open governance standards, and working software.

The current flagship project reproduces the Calvano et al. algorithmic-pricing environment and instruments autonomous pricing decisions with structured, pre-decisional records. The experiment separates record filing, welfare-adjusted rewards, consequential auditing, rejected-alternatives review, and dose-matched random interruption. The paper reports the nulls alongside the effects that survive: filing alone stays near baseline; sufficiently frequent consequential reading changes conduct; and rejected alternatives add little deterrence in this particular laboratory because outcome evidence already catches almost every failed audit.

I also developed the Decision Record and Control Standard, a 17-field record covering evidence, authority, uncertainty, predictions, rejected alternatives, affected stakeholders, system welfare, and reconsideration conditions. Steadfastly is an independent Microsoft 365 reference implementation; live tenant validation remains pending.

My comparative advantage is experimental production: translating a difficult economic or governance question into a reproducible environment, documented result, and inspectable artifact quickly and inexpensively. The current Calvano package exposes the simulation code, saved treatment cells, pooled estimates, figure and table builders, robustness checks, compute audit, and checksums.

I am most useful where economic reasoning and research engineering have to live in the same person: specifying the question, building the testbed, finding the right controls, preserving failed hypotheses, and turning the result into a paper and reusable evaluation asset.

Erik Postnieks holds a B.A. in Economics from Cornell University and an M.B.A. from the Amos Tuck School of Business at Dartmouth College (1993). At Cornell he studied under Richard Thaler and, at twenty-two, co-authored "The Smart Crash of October 19th" (Harvard Business Review, 1988) with Avner Arbel and Steven Carvell.¹

His career spans quantitative finance and technology investing. At Bankers Trust Company in New York, he structured derivatives and co-managed a $2 billion derivatives-product portfolio. He later founded Parametric Capital Management, a multi-strategy quantitative investment manager that reached approximately $440 million in peak assets and produced a KPMG-audited performance record.

He subsequently founded Wooster Asset Management and managed the Portage family of G10 currency funds. External capital was returned when major-currency interest-rate differentials converged and the underlying anomaly materially diminished. The discipline of recording what was known, predicted, rejected, and accepted before outcomes became part of the intellectual foundation for Decision Accounting.

I am seeking full-time research, research-engineering, applied-science, or economic-research work involving multi-agent systems, AI evaluations, agentic markets, computational economics, or auditable decision infrastructure. I am comfortable operating inside an established research agenda, sharing intellectual ownership with collaborators, and reporting to an executive or research leader.

I am open to relocating to the Bay Area, Boston, New York, or Washington, DC. I am particularly interested in roles where rapid computational experimentation can turn an open question into evidence that changes a product, evaluation program, market design, or institutional decision.