IPO pricing is challenging due to the limited operating history of new firms. We propose a simple text-based method to find peers for IPOs without relying on operating information. In particular, we create a text-based product similarity measure based on business descriptions of IPO prospectus and 10-K files of listed firms. We surprisingly find that text-based peers have higher aftermarket return correlations than traditional peers matched by industry and operating performance. Our results suggest that compared with text-based peers, the majority of IPOs are fairly priced except for a few years in the hot markets. IPOs prices are bid up in the aftermarket and revert to the fair value in the long run.