<![CDATA[Competitive tendering requires contractors to select a bid markup that balances the probability of winning against the expected profit. Although the Gates and Friedman bidding models have been widely applied, empirical comparisons of their performance across alternative probability distributions remain limited, particularly in Indonesian construction tendering. This study compares the performance of the Gates and Friedman models using historical tender data associated with the Rehabilitation of the Krueng Aceh Weir Project in Indonesia. Both models were evaluated under three probability distributions: multiple discrete, multivariate normal, and univariate normal. For each model-distribution combination, the probability of winning, expected profit, and optimum markup were calculated. Differences in the resulting markup values between the two bidding models were subsequently examined using one-way analysis of variance. The results show that, within the analyzed tender dataset, the Gates model produced optimum markups ranging from 6% to 8%, whereas the Friedman model produced higher optimum markups ranging from 8% to 10%. The Gates model generally generated higher winning probabilities at lower markup levels, while the Friedman model generated higher expected profits at higher markup levels. These findings demonstrate that model selection and probability-distribution assumptions materially affect the recommended bidding strategy. The study provides empirical evidence to help contractors select bidding models consistent with their profit objectives and tolerance for a lower probability of winning. ]]>
