Patterns in stock market movements tested as random number generators

This paper shows that tests of Random Number Generators (RNGs) may be used to test the Efficient Market Hypothesis (EMH). It uses the Overlapping Serial Test (OST), a standard test in RNG research, to detect anomalous patterns in the distribution of sequences of stock market movements up and down. Our results show that most stock markets exhibit idiosyncratic recurrent patterns, contrary to the efficient market hypothesis; also that OST detects a different kind of non-randomness to standard econometric long- and short-memory tests. Exposure of these anomalies should contribute to making markets more efficient.

► An efficient stock market should resemble a random number generator (RNG).
► Therefore, RNG tests can also test the efficient market hypothesis (EMH).
► One, the Overlapping Serial Test (OST), reveals patterns in stock market movements.
► These patterns cannot be detected by standard long- and short-memory tests of EMH.
► Results are based on 76 different price indices, covering all major stock markets worldwide.