Premise

Traditional project management treats work as binary: do it or don't. But real projects have different efficiency curves: some return 80% of their value at 10% effort, others need 60% effort to get to 80% value. The framework asks a more precise question: for each project in your portfolio, where is the effort-to-value ratio maximized, and when does additional work cost more than it returns?

How it evolved

Started as a personal time allocation problem. Grew into a parametric framework when it became clear that single-point heuristics (the Pareto Index alone) couldn't capture the shape differences between project types. Added activation energy as a dead zone before returns begin, separate ramp and decline rates for the rise and fall of efficiency, and peak value as a multiplier. Four named archetypes (Quick Win, Marathon, High Risk, Plateau) make the parameterization accessible without requiring the user to think in six dimensions. A greedy portfolio optimizer with diversity constraints converts the curves into concrete weekly time allocation recommendations.

Technical crux

The key metric is Early Stop Value: what percentage of total project value do you capture at 90% effort? For Quick Win archetypes this is 97%+; for Marathons it's much lower because the curve stays elevated late. This single number reframes the decision from 'is this done?' to 'should I stop now?' The optimizer penalizes late-stage allocation using the same instantaneous efficiency logic: it will redirect hours to a project in its activation zone before it will extend a project past its peak. All computation runs client-side; the entire tool is a single HTML file with Chart.js.

Findings

Six-parameter curve model (activation energy, Pareto index, ramp rate, decline rate, peak value, max hours) generating real-time instantaneous efficiency and cumulative value charts. Four preset archetypes for rapid parameterization. Portfolio optimizer that takes total available hours per week and outputs a prioritized allocation queue. Runs fully offline with no backend and no build step. Live at rivirside.github.io/P-index.