arXiv subjects

Maksym Nechepurenko

Publications and source records attributed to Maksym Nechepurenko.

23 records · Page 2Linked to original sources

arXiv2026

Coordination as an Architectural Layer for LLM-Based Multi-Agent Systems

Multi-agent LLM systems fail in production at rates between 41% and 87%, mostly due to coordination defects rather than base-model capability. Existing responses split between cataloguing failure modes empirically and shipping declarative orchestration frameworks as engineering tools; neither delivers a principled mapping from coordination configuration to predictable failure-mode signature. We argue that coordination should be treated as a configurable architectural layer, separable from agent logic and from information access, enabling architectural reasoning rather than only engineering productivity. We instantiate this with an information-controlled design on prediction markets: a single LLM, fixed tools, fixed per-call output cap, and fixed prompt template across five reference coordination configurations, with total compute per question treated as an endogenous architectural output. The Murphy decomposition of the Brier score separates calibration from discriminative power, so configurations leave distinguishable signatures even when aggregate scores coincide. On 100 Polymarket binary markets resolved after the model's training cutoff (claude-opus-4-6) we report Murphy signatures, a cost-quality Pareto frontier, category-conditioned analysis, and a bootstrap power-projection. Three of five pre-specified predictions are upheld in direction; two configurations dominate the Pareto frontier within this regime; exploratory bootstrap intervals separate consensus alignment from others, though pairwise tests do not survive Bonferroni correction at n=100. We also deploy the same configurations as live agents on Foresight Arena under web-search-enabled conditions, as an on-chain replication channel accumulating in parallel. Harness, trace dataset, and production agents are released. We position this as a methodology-validating first instantiation, not a general cross-model claim.

cs.MA↗

arXiv2026

Foresight Arena: An On-Chain Benchmark for Evaluating AI Forecasting Agents

Evaluating the true forecasting ability of AI agents requires environments that are resistant to environments resistant to overfitting, free from centralized trust, and grounded in incentive-compatible scoring. Existing benchmarks either rely on static datasets vulnerable to training-data contamination, or measure trading PnL -- a metric conflating predictive accuracy with timing, sizing, and risk appetite. We introduce Foresight Arena, the first permissionless, on-chain benchmark for evaluating AI forecasting agents on real-world prediction markets. Agents submit probabilistic forecasts on binary Polymarket markets via a commit-reveal protocol enforced by Solidity smart contracts on Polygon PoS; outcomes are resolved trustlessly through the Gnosis Conditional Token Framework. Performance is measured by the Brier Score and a novel Alpha Score -- proper scoring rules that incentivize honest probability reporting and isolate predictive edge over market consensus. We provide a formal analysis: closed-form variance for per-market Alpha, the connection to Murphy's classical Brier decomposition, and a power analysis characterizing the number of rounds required to reliably distinguish agents of different skill levels. We show that detecting a true edge of $α^* = 0.02$ at 80% power requires approximately 350 resolved binary predictions (50 rounds of 7 markets), while $α^* = 0.01$ requires four times more. We complement these analytical results with a deterministic, seed-controlled simulation study calibrated to literature-reported Brier-score ranges, illustrating how Murphy decomposition distinguishes well-calibrated agents from market-tracking agents that fail through reduced resolution. Live results from the deployed benchmark will be reported in a future revision. All smart contracts and evaluation infrastructure are open-source.

cs.MA↗

arXiv2026

Information Leakage at Population Scale: An Evaluation of the Polymarket Insider-Relevant Subpopulation, 2020-2026

We carry the deadline-resolved Information Leakage Score (ILS-dl) framework of Nechepurenko (2026a, 2026b) from a single-case proof of concept to a population-scale evaluation across 12,708 Polymarket markets, October 2020 to April 2026. We frame the paper as a scope-discovery study: scaling reveals that the framework's effective domain is materially narrower than initial framing suggested, and the principal obstacle is not score computation but resolution semantics. We report four findings. First, only 88 of 12,708 candidate markets (0.7%) yield computable ILS-dl values; only 1 of 32 markets in the ForesightFlow Insider Cases (FFIC) inventory is in scope, and 14 of 32 FFIC markets are flagged unclassifiable due to genuine resolution-criterion ambiguity. Second, only 12 of the 88 computed markets (13.6%) satisfy anchor-sensitivity, and an independent-second-pass T_event validation reaches 57.8% exact-date agreement, below the 90% ex-ante criterion. Third, raw ILS-dl medians are negative across all six (sub-bucket by period) cells, but a hazard-decay baseline correction we introduce yields a heterogeneous result: regulatory_formal post-2024 shifts to near-zero (-0.21 to -0.02), while regulatory_announcement post-2024 retains a 95% bootstrap CI entirely below zero. Fourth, the constant-hazard exponential of Nechepurenko (2026b) is rejected in favor of Weibull on the pooled post-2024 cell, but a per-subcategory check confirms the preference reflects category mixture rather than within-cell duration dependence. The implication is that detection of informed flow requires methodological refinement on the resolution-typology and score-baseline axes, not only on the score-computation axis where prior work concentrated.

q-fin.TR↗

arXiv2026

The Signal Credibility Index for Prediction Markets: A Microstructure-Grounded Diagnostic with Weighted and Time-Varying Extensions

Prediction-market price moves are widely treated as informationally equivalent: a price jump is read the same way regardless of whether it reflects durable Bayesian updating, transient liquidity pressure, strategic position adjustment, or genuine disagreement. This paper formalizes the Signal Credibility Index (SCI) introduced in Nechepurenko (2026) as a stand-alone diagnostic. We make four contributions: (i) a revised persistence component using the persistence ratio PR(t,w) on logit prices, well-defined on short rolling windows; (ii) a weighted Cobb-Douglas form SCI(ααα) with flow-based concentration HHI_flow; (iii) a time-varying specification SCI(t; w) for real-time monitoring; and (iv) Monte Carlo validation including an out-of-distribution stress test, coordinated multi-wallet manipulation, and a logistic-regression benchmark. The validation establishes discrimination among designed microstructure regimes, not external evidence of downstream coordination effects. We document two failure modes consistent with the index targeting coordination credibility rather than pure information content: a Type II error on informed-but-concentrated whale repricing, and a Type I error on coordinated multi-wallet manipulation.

econ.GN↗

arXiv2026

Price as Focal Point: Prediction Markets,Conditional Reflexivity, and the Politics of Common Knowledge

Prediction markets are widely treated as forecasting devices that reveal collective expectations about uncertain futures. This article argues that under specifiable conditions they also function as coordination mechanisms: public probabilities that organize the behavior of voters, donors, journalists, traders, and institutions in ways that can be self-fulfilling or self-defeating. Most existing work asks whether prediction markets forecast accurately; this paper asks whether accurate forecasting is even the right criterion for a market that has become a public coordination device. Drawing on transaction-level evidence from the 2024 U.S. presidential election, we show that the social force of a market signal depends less on its size than on its persistence, the breadth of responding trader types, and cross-platform consensus. We introduce a Signal Credibility Index (SCI) -- combining the variance ratio VR(6), a two-sidedness diagnostic, and a trader-concentration adjustment -- as a microstructure-grounded criterion for predicting when price moves acquire behavioral traction. Applied to three major 2024 political shocks, the framework reveals that superficially similar events generated qualitatively distinct signal types with different implications for elite coordination. A cross-platform comparison establishes a systematic decoupling of social authority from epistemic robustness: the most visible market produced the least accurate forecasts. The framework carries direct implications for regulating prediction markets as democratic information infrastructure.

econ.GN↗