Stars Are Where You Draw the Line: Reassessing Methods for Identifying Star Entrepreneurs

Entrepreneurship outcomes tend to be right-skewed and heavy-tailed, with a small fraction of “star” firms often accounting for a disproportionate share of value creation. Yet, how to define the “star” entrepreneurs driving these outcomes remains highly contested. In this paper, we argue that star identification is best understood as a modeling choice that depends on the research objective—no threshold rule can be evaluated as “optimal” or more “precise” independent of what the rule it is supposed to serve—and introduce a precision–recall framework that makes the trade-offs of threshold-based approaches more explicit. Using 13 years of Inc. 5000 data linked to subsequent public listings, we then compare nine threshold-based methods, including the recently proposed quantile absolute deviation procedure by Gala and Schwab (hereafter GS-QAD) which uses a bootstrap to set a cutoff tailored to the tail of the observed distribution. Three findings emerge. First, using simulations, we show that GS-QAD’s bootstrap procedure introduces systematic sample-size bias that can confound cross-industry comparisons. At the same time, as sample size grows, the method converges to a near-universal 17% to 20% rule. Second, firms classified as stars are 10 to 20 times more likely to become publicly traded and account for the majority of current market capitalization among publicly traded Inc. 5000 alumni. However, the choice of threshold determines a precision–recall trade-off in which no method dominates on all metrics. Finally, the probability of becoming a publicly traded company rises smoothly with revenue, with no detectable discontinuity at any threshold. We conclude that stars are where you draw the line—thresholds should be treated as modeling choices fit to the research or policy objective, validated externally where feasible, and reported with their explicit trade-offs.