The Biggest Objection I Get to Monte Carlo Is the One the Data Says Is Wrong: Why Complexity Was Never the Variable Your Forecast Was Missing

The prevailing industry belief that project risk scales linearly with size is fundamentally flawed and unsupported by empirical evidence. While conventional wisdom suggests that large, complex projects are inherently more prone to failure than smaller ones, data analysis reveals that project size is not a reliable predictor of risk. Instead, the most significant threats to project timelines and budgets often reside in smaller, overlooked initiatives that lack the rigorous oversight applied to flagship projects.

Research involving 5,094 IT projects demonstrates that cost overruns do not follow a standard bell curve but rather a power law with a "fat tail." This distribution indicates that while most projects perform near their estimates, a disproportionate few experience catastrophic failure. Notably, the smallest projects in the sample exhibited the highest mean cost overruns, largely because they are frequently exempted from formal forecasting and lack the statistical buffer provided by a high volume of work items.

The methodology behind these findings relies on comparing actual project outcomes against initial estimates, rather than relying on subjective, self-reported survey data. This distinction is critical, as survey-based metrics often conflate project visibility with actual risk. By shifting focus from project size to flow stability—governed by principles such as Little’s Law—producers can more accurately assess risk. Monte Carlo forecasting serves as a robust tool in this context, as it utilizes historical throughput and cycle time data to model potential outcomes. By sampling from a team’s actual performance history, these simulations naturally incorporate the "fat tail" of variance, providing a more honest and accurate forecast than traditional, size-based estimation methods.