The article characterizes high and low value patents based on the non‐linearity of the age‐value relationship, as an attempt to provide some statistical understanding on the difference in the value of patent characteristics over time. A set of 138 US singleton patents, mainly from the computers and communication field, successfully auctioned by an US auction firm called OceanTomo during 2006–2008 forms the data. Analysis shows evidence of non‐linearity in the patent age‐value relationship and the sensitivity of patent characteristics to temporal dimension in explaining value of patented knowledge. The U‐shaped temporal value of knowledge, identifying older patented knowledge as more valuable, thus, finds support. Furthermore, patents sold in their first half exhibit more lag, less patent scope and less forward citations. Patents sold by firms dominate the younger patent, while the older patent cohort finds more patents sold by individuals. These and other results are discussed for their significance for patent sale. We acknowledge the limitations of a small sample size. Nevertheless, the article provides statistical understanding on the potential characteristics of high and low value patents, explored through the non‐linear age value dynamics.