Tests of the efficient market hypothesis (EMH) suffer from a known flaw that has nevertheless received insufficient attention in the literature. Just as in Zeno’s famous paradox of Achilles and the tortoise, while we are testing market efficiency the measure of efficiency has already moved on. Prediction technology and data availability perpetually improve beyond the level that was available to market participants during the sample period. This introduces a bias into the tests. If one believes in the existence of deterministic patterns and of technological progress, one has to expect to discover inefficiencies in historical data. We can only expect prices to reflect all available in- formation to the degree that current technology enables. From this argument follows a necessity to adjust the EMH. This article discusses possible adjustments and develops the concept of technological efficiency. Based on a review of empirical evidence, this concept is shown to resolve many anomalies and to help bridge the gap between the financial economics literature and the rapidly advancing machine learning literature.