Study guide
6.3 Risk Outputs, Confidence Dates, Sensitivity, and Contingency
AACE Planning and Scheduling Professional PSP
Track your progress as you read
Finished sections get a check mark.
1 Chapter 0: Introduction & AACE PSP Exam Structure
0.1 Current AACE PSP Exam, Eligibility, and Certification Facts 0.2 Planning and Scheduling Professional Roles and Competencies 0.3 Cost Engineering and Scheduling Terminology per RP 10S-90
2 Chapter 1: Scope Definition & Planning Fundamentals
1.1 Work Breakdown Structure (WBS) Architecture & RP 33R-15 1.2 Organization & Resource Breakdown Structures (OBS, RBS, RAM) 1.3 Project Planning and Activity-Logic Development 1.4 Contract Requirements and Delivery Constraints 1.5 Stakeholder Identification, Needs, and Plan Review 1.6 Constructability Methods and Value Engineering 1.7 Resource Identification and Project Variables 1.8 Scope, Goals, Project Plan, and Phase Definition 1.9 Cost Breakdown Structure and Cost Estimate Development 1.10 Phase Relationships, Baseline Plan, Forecasts, and Risk/Recovery Planning
3 Chapter 2: Critical Path Method Mechanics & Network Logic
2.1 Activity Definition and Logic Relationships 2.2 Activity Durations, Productivity, and Work Calendars 2.3 Forward and Backward Pass Mathematical Mechanics 2.4 Float Calculations and Critical-Path Identification
4 Chapter 3: Advanced Network Techniques & Schedule Constraints
3.1 Schedule Constraints and Their Impact on Dates and Float 3.2 Advanced Logic Modeling: Ladders and Fragnets 3.3 Schedule Levels of Detail and Schedule Basis Documentation 3.4 Schedule Scope and Schedule Specification 3.5 Breakdown-Structure Relationships and the Cost Estimate Model 3.6 Stakeholder Feedback and Types of Schedules 3.7 Activities, Milestones, Schedule Quality, and Basis Documentation
5 Chapter 4: Resource Management & Leveling
4.1 Resource Planning, Allocation, and Histograms 4.2 Resource Leveling vs. Resource Smoothing 4.3 Cumulative Resource S-Curves & Productivity Tracking
6 Chapter 5: Baseline Management & Schedule Updating
5.1 Baseline Establishment, Forecasts, and Controlled Revision 5.2 Progress Measurement and Data Date Mechanics 5.3 Out-of-Sequence Logic: Progress Override vs. Retained Logic 5.4 Schedule Update Review and Quality Verification per RP 53R-06 5.5 Cost/Resource Management, Schedule Change, and Feedback 5.6 Acceleration and Recovery Scheduling per RP 54R-07 5.7 Control Schedules, Variance Analysis, Forecasts, and Management Summaries
7 Chapter 6: Schedule Risk Analysis & Monte Carlo Simulation
6.1 Schedule Risk Principles, Risk Drivers, and Three-Point Estimates 6.2 Monte Carlo Simulation, Criticality, Correlation, and Merge Effects 6.3 Risk Outputs, Confidence Dates, Sensitivity, and Contingency
8 Chapter 7: Earned Value & Earned Schedule Management
7.1 Earned Value Foundations: PV, EV, AC, and Variances 7.2 Traditional EV Performance Indices & Late-Project Quirks 7.3 Earned Schedule (ES) Theory, SV(t), and SPI(t) 7.4 Independent Schedule Forecasting & IEAC(t)
9 Chapter 8: Forensic Schedule Analysis & Delay Claims
8.1 Delay Classification, Criticality, and Contract Questions 8.2 RP 29R-03 Taxonomy, Data Validation, and Method Selection 8.3 TIA, Windowed Analysis, and Subtractive Models 8.4 Concurrency, Pacing, and Float: Technical Analysis Before Entitlement
10 Chapter 9: Executive Communication & Practical Scenario Applications
9.1 Official PSP Memo Writing Framework 9.2 Structured Technical Problem-Solving for PSP Scenarios 9.3 Current Exam Strategy and Core Calculation Review
6.3 Risk Outputs, Confidence Dates, Sensitivity, and Contingency
Key Takeaways Explain This Topic AI
A percentile date is conditional on the risk model and indicates the proportion of simulated outcomes at or before that date.
AACE guidance does not make P80 a universal commitment; decision-makers select confidence in light of risk tolerance and consequences.
Schedule contingency is the time allowance between a deterministic target and a selected risk-informed commitment or reserve point.
Criticality index, correlation, and risk-driver contribution are complementary diagnostics and should not be treated as interchangeable.
Sensitivity testing should challenge durations, correlations, logic, calendars, and risk occurrence assumptions before management relies on results.
Last updated: October 2026
6.3 Risk Outputs, Confidence Dates, Sensitivity, and Contingency
A schedule risk simulation produces a distribution , not a promised finish date. If the model’s P50 completion is October 20, about half of the modeled iterations finish on or before that date and half finish later. If the P80 completion is November 14, about 80% finish on or before November 14—subject to the model’s logic, calendars, duration distributions, risk drivers, and correlations.
Percentiles are not universal policy
Percentiles express confidence within a model. They do not prescribe which date must become the contract date or management commitment. A sponsor may select a higher percentile for a safety-critical shutdown or a lower percentile for an internal stretch target. The decision should reflect consequence, risk appetite, commercial allocation, and the maturity of the underlying plan.
This distinction matters on exam scenarios. If a question says management selected P80 and provides Day 300 deterministic and Day 345 P80 results, the gap is 45 days. If the question merely asks for the “required” contingency without giving a selected confidence policy, there is not enough information to declare P80 mandatory.
Contingency, float, and padding
These concepts are related but not identical:
- Concept Purpose
- Total float Calculated flexibility created by network logic and constraints
- Schedule contingency Time allowance associated with identified uncertainty and a chosen confidence objective
Management reserve A governed allowance outside detailed control-account baselines under the organization’s control framework
Activity padding Hidden or explicit allowance embedded in individual durations
Contingency should be visible enough to govern. That does not mean one terminal buffer is always the only acceptable design. A project may place allowances at phase interfaces, use explicit risk activities, hold a management reserve, or incorporate weather calendars, depending on contract and control policy. What matters is that the method is documented and avoids double-counting the same risk in durations, calendars, and a separate buffer.
- Reading sensitivity outputs
- Criticality index
The criticality index is the percentage of iterations in which an activity is critical under the model’s criticality rule. A high value highlights frequent critical-path participation. It can miss activities that are nearly critical in most iterations but exactly critical in few.
Correlation or tornado chart
A rank-correlation tornado chart shows association between an input and the completion output. A strong positive Spearman correlation means larger sampled input values tend to accompany later completion. Correlation does not by itself prove that changing that activity will produce an equal and opposite completion improvement, especially when common drivers or logic interactions are present.
Risk-driver contribution
Risk-driver analysis ties completion variation to named risks. It is often more actionable than an activity-only ranking because a single driver may affect many activities. Teams should compare driver rankings before and after proposed treatments.
Sensitivity and validation
Run deliberate challenges before presenting a date:
Remove or vary a dominant risk driver and observe the change.
Test plausible correlation assumptions.
Compare results under reasonable distribution choices.
Inspect iterations that produce extreme outcomes.
Confirm that critical paths remain physically plausible.
Check whether uncertainty or contingency has been counted twice.
The goal is not to find one “correct” percentile to a false level of precision. It is to understand which assumptions drive the decision and how robust the conclusion is.
Communicating to management
A useful summary states the deterministic date, selected confidence dates, top risk drivers, assumptions, and treatment decisions. For example: “The deterministic plan finishes September 30. The current model produces P50 October 20 and P80 November 14. Management selected P80 for the external commitment, subject to closing the permit and vendor-data risks. Re-running the model after treatment is required.”
That wording tells readers what the dates mean and what could invalidate them. It is more defensible than calling P80 an AACE-mandated baseline or describing a probabilistic result as guaranteed.
Applied review: interpret the distribution, not one percentile
Simulation output is conditional on the model, input ranges, correlations, and risk events. Review the full distribution and the assumptions before selecting a confidence date. A P80 completion date means that approximately eighty percent of simulated outcomes finish on or before that date under the modeled conditions; it does not guarantee delivery and it is not universally mandatory. The chosen confidence level should reflect governance, risk tolerance, contractual context, and the decision being made.
Separate schedule contingency from management reserve and from hidden padding. Contingency should be visible, governed, and linked to the uncertainty it addresses. Correlation matters because common drivers can move several durations together; ignoring it can understate or distort overall spread. Sensitivity and criticality results help identify influential drivers, but they do not assign contractual responsibility.
Communicate the deterministic baseline, current forecast, modeled range, selected planning value, and key assumptions as distinct items. Explain what was included or excluded and which responses could change the result. Re-run the analysis when material scope, logic, progress, or risk information changes.