Three papers — on narrative co-integration and asset price forecasting, machine-assisted project risk management, and the structural case for human relevance under AI — available for download. Enter your details below to unlock all three.
Enter your details once to unlock all three downloads. No subscription — just the research.
Your details are used only to send you relevant research updates from Milestone Planning and Research. No third-party sharing.
Papers unlocked. All three downloads are now active — click any paper below to download the PDF.
Introduces Perceptual Macroeconomics — a framework in which the dominant composite of daily headline narrative and the S&P 500 price level constitute co-integrated time series sharing a common stochastic trend across 893 trading days (August 2022 – March 2026). A principal component narrative index derived from approximately 185 topic-level word-frequency series explains 88% of the variance in S&P 500 levels, outperforming the University of Michigan Consumer Sentiment Index, Federal Reserve policy variables, and the Baker-Wurgler sentiment index in a direct horse race. Formal Engle-Granger and Johansen co-integration tests confirm a long-run equilibrium relationship. The paper extends the framework to organizational application: any organization with internally generated text data can apply the same measurement and forecasting architecture to their own domain outcomes. A five-model forecasting competition evaluates performance on directional accuracy, long/short P&L, Sharpe ratio, maximum drawdown, and Kelly-optimal position sizing. M1 achieves 92.9% directional accuracy and an annualized Sharpe ratio of 23.0 across 689 out-of-sample observations. This is the empirical and theoretical foundation of the Sentinel Forecast System.
Introduces PRIMMS-GPT — Project Risk Identification Measurement and Mitigation System, augmented by a Large Language Model — as a hybrid cognition architecture designed to close the intelligence gap between what project dashboards record and what is structurally happening inside a project. PRIMMS-GPT combines a compiled MATLAB application that computes deterministic Bayesian weight-of-evidence signals from schedule data, Voice of Team distributed sentiment, and documentary artifacts, with a five-layer Cortical Hierarchy prompt architecture that routes those signals through an LLM for interpretation, pattern recognition, trajectory projection, and executive briefing generation. Classifies projects against six empirically grounded failure archetypes, produces recovery trajectory projections at +2, +4, and +8-week horizons, and generates a sponsor-ready governance document — while preserving an inviolable human authority boundary at every decision point. Retrospective application to completed enterprise transformation programs demonstrates decisive evidence classifications and identifies actionable recovery windows not surfaced by conventional monitoring. An illustrative example produces a schedule jeopardy classification at 65.5 dB and identifies a four-day actionable recovery window missed by the PMO dashboard in use at the time.
The sixth and final volume in the Inductive Enterprise series. Where the preceding five volumes argued from institutional design, organizational economics, and competitive strategy, this volume argues from mathematical logic, theoretical computer science, computability theory, and political economy that human agency is not merely currently valuable but permanently and structurally necessary. The convergence of Gödel's incompleteness theorems, Turing's halting problem, Hayek's dispersed knowledge argument, Arrow's impossibility theorem, and Marks's no-free-lunch theorem establishes that no sufficiently expressive inductive system can self-regulate, self-verify, or substitute for human judgment at the level of institutional accountability. This is not a claim about current AI capability limitations — it is a structural proof about what any inductive system can in principle achieve. The paper identifies the Gödel-Turing constraint and the Zero constraint as the two foundational limits that make human authority over AI systems not merely ethically preferable but architecturally necessary.
The papers above are the publicly available research outputs of a larger six-volume body of work — The Inductive Enterprise — that develops a unified framework for understanding the relationship between human judgment and machine induction.
Each volume addresses the same central question from a different direction: how should organizations deploy AI to gain competitive advantage while preserving the human authority that makes AI deployment legitimate and sustainable? The answer developed across the series is architectural, not philosophical — specific governance structures, institutional designs, and system architectures that keep human judgment in command while extending human perception through machine intelligence.