Sentinel  ·  Perceptual Macroeconomics  ·  Milestone Planning and Research
Artificial Intelligence · Research · Specialty AI Products · Training & Coaching

We convert the narrative
into actionable
business decisions.

Most organizations ignore their most abundant data source: the stream of language flowing through headlines, reports, and communications every day. We show what becomes possible when you don't. Milestone Planning and Research conducts AI research, builds specialty AI products, and delivers training and coaching that help organizations improve productivity and competitive advantage. Start with the weekly S&P 500 forecast — driven entirely by textual data, 93% directional hit rate across 907 trading days.

93.0%
Directional hit rate · text data only
907
Trading days · expanding-window backtest
23.5
Annualised Sharpe · M1 champion
185
Topic series · all from headline text

You're in. Scroll down to read this week's forecast.

Free. Published every Monday. No sales pitch — just the forecast and the method behind it.

The textual data advantage — demonstrated

Most organizations discard
their most powerful signal.

Language — in headlines, reports, customer communications, and internal documents — is the largest and most ignored data source in most enterprises. Sentinel is a live demonstration of what structured textual analysis produces when applied rigorously: a forecast driven entirely by word frequency data, with performance that outpaces conventional models by a wide margin. The same methodology scales to any domain where text carries signal.

The demonstration
S&P 500 · driven by text alone

185 headline topic series, scored daily by word frequency. A principal component index derived from that language explains 88% of S&P 500 variance across 899 trading days. No price data enters the model. The forecast is built entirely from what people are writing about.

The principle
Narrative and price share a common trend

The PC-1 narrative composite and the S&P level are formally co-integrated — confirmed by Engle-Granger and Johansen tests. They are the same thing expressed in different units. When they diverge, the gap closes. The ECT z-score measures that gap every week.

The architecture
Statistical model first. LLM second.

Stata runs four do-files, estimates five competing models, and locks all quantitative outputs before Claude sees them. Claude receives those fixed values and interprets them — it cannot change a single number. Auditable, reproducible, no black box.

The output
A briefing, not a chart

Each weekly issue contains the signal dashboard, 20-day forecast cone with Jeffreys bands, the narrative topic table, model competition rankings, and a five-layer Claude analysis — plus a human analyst context note on signals the model cannot yet see.

The broader point
This method applies to your organization

Sentinel uses financial headlines. The same architecture works on customer communications, project status reports, regulatory filings, competitive intelligence, or any domain where language volume and sentiment carry information about future states. That is the service we offer.

The body of work
Six books. Four AI systems. 30 years.

John Aaron has published six books advancing this framework — from AI governance and hybrid cognition to Bayesian project risk and competitive strategy. PRIMMS-GPT, Sentinel, IDS-X, and the PRIMMS Portfolio System are the operational systems that put the theory to work.

93.0%
Directional accuracy
h=1 · M1 champion model
907
Trading days
Aug 2022 — Apr 2026
23.5
Annualised Sharpe
Long/short strategy · M1
236
Calmar ratio
PnL / max drawdown
88%
Variance explained
PC-1 narrative → S&P level
185
Headline topic series
Daily WORDSTAT frequency
Desktop product · Available for license

PRIMMS‑GPT

Project Risk Identification Measurement
and Mitigation System + LLM

Every project management tool on the market does the same thing: it records what has happened. None of them can tell you what is structurally happening inside a project — the confidence erosion, the optimism bias compressing the recovery window, the failure archetype taking shape weeks before it appears on a dashboard.

PRIMMS-GPT closes that gap. A compiled MATLAB application computes Bayesian weight-of-evidence signals from your schedule data, team sentiment, and project documents. Claude receives those locked signals and runs a five-layer Cortical Hierarchy analysis — producing a sponsor-ready governance document in a single workflow. The machine orients. The humans decide.

17+
Years operational
65.5 dB
Illustrative jeopardy signal
4-day
Recovery window surfaced
24
ICT projects calibrated
Contact Inquire about licensing Learn more →
Six-posture jeopardy classification
1–2 Dark/Light Green3 Amber4 Orange5–6 Red

Bayesian weight-of-evidence accumulation in decibans. Signal strength is auditable and reproducible — not an AI opinion.

Six failure archetypes identified
01 Execution Collapse
02 Perception Distortion
03 Scope Drift
04 Governance Gap
05 Data Dependency Lock
06 Resource Fracture · Premature Closure
What it produces
Trajectory projections at +2, +4, and +8 weeks — with named leading indicators and watch thresholds
Six-option COA matrix — each with owner archetype, decision timing, and measurable gate trigger
Sponsor-ready governance document — jeopardy banner, signal table, recovery options, three leadership decisions with triggers
Two products. One architecture. Sentinel applies Bayesian signal processing and LLM interpretation to financial narrative forecasting. PRIMMS-GPT applies the same architecture to project risk. Both are produced by John Aaron at Milestone Planning and Research.
Sentinel → Forecasting
PRIMMS-GPT → Project Risk
The methodology

Text in. Decisions out.
Nothing black-boxed.

Sentinel is built on a principle that transfers to any domain: textual data, structured rigorously, produces quantitative signals that outperform conventional models. The architecture is explicit and auditable at every step.

Channel 1 · Stata
Deterministic · locked first
The model computes
  • Five competing models re-estimated across the full expanding-window backtest
  • ECT z-score — distance of current price from the narrative equilibrium anchor
  • Regime detection: NORMAL, SURPRISE, or GEO
  • 20-day forward price cone with Jeffreys confidence bands
  • Consensus signal and WoE increment in decibans
Channel 2 · Claude
Interpretive · cannot override channel 1
The model interprets
  • Signal validation — are the Stata outputs internally consistent?
  • Dominant narrative identification in plain language
  • Situation awareness linking current headlines to model signals
  • Dörner audit for reasoning errors and overconfidence
  • Three cone-anchored scenarios with named confirmation conditions
"Asset prices are determined by the shadows, not directly by the objects. The narrative composite and the price level share a common stochastic trend — they are, in the long run, the same thing expressed in different units." Aaron (2026) · Perceptual Macroeconomics: Narrative Co-Integration and Asset Price Forecasting

The same relationship exists in your organization between the language in your communications, reports, and external environment — and the outcomes you are trying to predict or control. Surfacing that signal is what we do. The S&P forecast is the live demonstration. The services below are how we bring it to your domain.

Sample output

What the briefing looks like

This is the signal dashboard from the April 10, 2026 issue — the week PC-1 reached its highest reading in the full 907-day sample.

Sentinel · Week of April 10, 2026 · GEO Regime Active
6,831
S&P 500 · Apr 10
Intraday anchor
8,096
20-day base target
+18.5% · M1
93.0%
M1 hit rate · h=1
907 trading days
6.06
PC-1 composite
Record high in sample
Who builds Sentinel

Research, not software.
Judgment, not automation.

Sentinel originates from a 30-year body of work in applied economics, Bayesian analytics, and enterprise transformation — and is maintained by the same people who consult on applying probabilistic reasoning inside organizations.

JA
John M. Aaron
Founder · Principal architect
Milestone Planning and Research, Inc.
Chicago, Illinois · Est. 1995
Credentials
PhD Economics · Univ. of Illinois Chicago
MS Organizational Development
PMP Certified #1018 (PMI · 1993)
Electronics & avionics background

John Aaron is the founder of Milestone Planning and Research and the intellectual architect of the platform behind Sentinel, PRIMMS-GPT, and The Probability Advantage framework. Over more than three decades he has led complex enterprise initiatives — including SAP implementations and S/4HANA upgrades — while designing the specialty analytic and AI systems that now form the operational core of this work.

His academic foundation spans econometrics, time series analysis, and organizational development, combined with hands-on electronics and field engineering experience. That breadth allows him to move across executive strategy, software architecture, Bayesian analytics, cybersecurity diagnostics, and delivery execution — and to translate advanced analytic ideas into practical tools rather than leaving them at the level of theory.

Systems built
Sentinel forecasting system
PRIMMS-GPT (hybrid cognition)
PRIMMS® Portfolio System
IDS-X Intrusion Detection
Books published
The Probability Advantage
The Inductive Enterprise
Machine-Assisted Perception
Human Relevance in an Age of Induction
+ 2 further volumes
30+
Years in practice
2,000+
PMs trained since 1989
6
Books in the series
SD
Shouvik Dutta
Enterprise AI · CIO-level advisory

Senior AI and digital transformation executive specializing in enterprise-scale AI deployment. Former CIO across multiple industries. Works with PE-backed and growth-stage companies on AI strategy through to production capability — from data platform architecture and ERP modernisation to LLM integration and AI governance. The Probability Advantage thesis is the operational anchor of his advisory practice.

TR
Timothy Ricordati, Ed.D.
Executive coach · AI capability

Executive coach and organizational development consultant specializing in LLM integration and structured decision-making under uncertainty. Partner, George Washington Street Partners. His workshop and coaching engagements embed the same hybrid cognition discipline Sentinel embodies — machine-assisted perception, human judgment in command — directly into how leadership teams operate.

From forecast to enterprise capability

The method behind this forecast
is available to your organization.

Six service areas — all grounded in the six-book Inductive Enterprise series — covering enterprise AI deployment, project risk intelligence, narrative signal mining, competitive strategy, AI capability coaching, and enterprise transformation. Products include Sentinel (this forecast) and PRIMMS-GPT for project risk.

John Aaron
Inductive enterprise consulting

AI deployment into existing process architecture — structured and unstructured data, AI PMO design, five governance policy instruments.

John Aaron · PRIMMS-GPT
Machine-assisted project risk

Bayesian WoE + Claude for ERP and transformation programmes. Detects structural deterioration before dashboards do.

John Aaron · Sentinel
Narrative intelligence & text mining

Custom textual signal systems for forecasting, risk, or competitive intelligence — applied to your domain, not just financial markets.

John Aaron
AI strategy & bold-play advisory

Competitive positioning under AI convergence. When front-end analysis becomes commodity, advantage migrates to commitment and governance.

Tim Ricordati, Ed.D. · The Probability Advantage
AI capability coaching & workshops

Seven strategic domains from Book 5: AI vs analytics, process optimization, LLM productivity, hybrid cognition, governance, AI PMO, infrastructure. Operational engagements, not events.

Contact Enquire with Tim →
Shouvik Dutta
Enterprise AI transformation

CIO-level advisory: AI strategy to production, data platform, ERP modernisation, LLM integration, intelligent automation, and governance at scale.

Contact Enquire with Shouvik →
Contact [email protected]  ·  milestoneplanning.net
Free · Published every Monday

See what textual data
can actually do.

Enter your email to access this week's Sentinel briefing — a live demonstration of narrative-driven forecasting. Future issues arrive every Monday, free, with no obligation.

You're in. Scroll down to read this week's forecast.

No spam. No sales pitch. One structured briefing, weekly. Unsubscribe at any time.