Introduction
I want to share a methodology I have spent the last eight months developing, validating, and writing up — and I would genuinely value the technical scrutiny of this community before I take it further into academic submission.
The problem it addresses is one I suspect everyone here has lived:
A static risk register sums independent expected values. It treats fifteen risks as fifteen separate coin flips. It produces a number. The Project Director nods at the number. The project begins.
The number is wrong. Not because the risks are wrong. Because they are not independent.
On a highways Survey and Investigation project I got onto in September 2025, the static register produced an expected value of £628,168. Standard Monte Carlo — independent risks, no dependency structure — gave a P90 of £1,397,284.
Probabilistic Chain Analysis (PCA) with Bayesian Network-coupled Monte Carlo gave a different answer entirely.
Not just a different number. A different conversation.
What PCA does that standard Monte Carlo cannot:
Standard MC samples each risk independently on every iteration. It cannot know that when Contractor Mobilisation Delay fires, Key Subcontractor Default becomes four times more likely to fire. It has no chain. It has no memory of cause.
PCA maps the dependency structure explicitly — a directed acyclic graph of cause and effect — and couples it to the Monte Carlo engine. Every iteration now carries the chain. When N4 fires in iteration 3,742, the engine elevates N5's probability from 15% to 64%. The chain runs. The tail gets heavier. And the output is honest.
Key results from the UK highways validation:
- P90 cost reduced by £250,000 through a single targeted pre-mobilisation intervention
- CVaR(90%) reduced by £101,025 versus the standard Monte Carlo baseline
- Cascade uplift of 3.3×–4.1× identified on six nodes the static register treated as independent
- 16.7× return on £15,000 of targeted management action
- Intelligence Log posteriors updated across 12 risk events — calibrating the model for the next project
The six-step methodology:
- Identify — every risk as a chain entry point, not a standalone event
- Map — trace how Risk A activates Risk B and B compounds into C
- Quantify — Monte Carlo simulation with correlated variables and Bayesian lift factors
- Analyse — cascade probability across the full network
- Optimise — break the chain at its highest-leverage node, not everywhere
- Monitor — continuously, not quarterly, as T1 assumptions shift to T2
The Intelligence Log — the thing that changes everything:
Every commercial risk tool does the same thing at project close: nothing. The posterior beliefs that reality handed you — the overestimated impacts, the underestimated probabilities, the dependencies nobody modelled — are reset to zero. The next project starts from scratch.
The Intelligence Log carries corrected beliefs forward using Bayesian inference. Every risk that fires is logged against its prior. Every risk that does not fire is logged too. The next project starts from evidence, not from the same industry defaults that have been recycling optimism bias for decades.
RiskPulseV12 — the engine that implements PCA — is calibrated across 16,000+ infrastructure projects spanning 8 sectors, sourced entirely from public databases including SECI, MNRE, UK National Highways outturn data, NAO reports, and IPA annual reports.
Why I am posting here:
Planning Planet has the most technically rigorous planning and controls community I know of. Before I submit this to PMI Project Management Journal and Construction Management and Economics, I want the methodology stress-tested by practitioners who will find the gaps I have not found myself.
The schedule model failed in the case study. I have documented that honestly in Section 3.7. A whitepaper that only reports wins is a marketing document. This is a technical contribution — and I want it treated as one.
The full whitepaper is attached.
Canonical citation and DOI: https://zenodo.org/records/19436269
I welcome technical challenge, methodological critique, and practical pushback from anyone who has run QSRA or probabilistic schedule risk analysis in live delivery environments.
Time, Cost and Scope don't run your project. The invisible chains of risk associated with them do. To manage the project, you must manage the cascade.
Nikhil Dhand, PMP Senior Project Management Consultant | Creator, RiskPulseV12
