There is increasing evidence that a polypeptide chain in solvent conditions that favor folding may have transient structure and is significantly more compact than a fully denatured chain However there is no sequence-dependent model to capture such interactions In this work, we present a simple and computationally inexpensive model based on a wormlike chain with excluded volume The probability distribution of millions of such chains is reweighted to bias compact conformations in which residues of similar hydrophobicity are located near each other This model, which has one adjustable parameter, is fit to measured values of intramolecular contact formation, which has been shown to be extremely sensitive to various models of intrachain distances We show that under various denaturant concentrations, there is good convergence of the model for several different sequences with a wide range of dynamics We also show that this model quantitatively predicts paramagnetic resonance enhancement (PRE) measurements with no adjustable parameters Therefore a simple probabilistic model that accounts for sequence-specific interactions may give a more realistic starting point for predictions of protein folding