Statistical arbitrage strategies attempt to benefit from empirical regularities without the need for a strong theoretical underpinning in economic theory. Consequentially, most techniques employed in this field work with daily (or higher frequency) price data and much less with other economic or financial data. The techniques involved employ sophisticated statistical algorithms.
The most well-known examples are pairs trading and volatility pumping. The first strategy attempts at identifying a pair of two securities that are glued together by a statistical relationship (cointegration) that results into a mean reverting spread between both securities.
The investor takes a long position in a high-frequency (intraday) rebalanced equal-weight portfolio and a short position in a low-frequency (dai ly) rebalanced equal-weight portfolio. While both portfolios should have the same expected average return, the difference in geometric return should grow over time as the continuously rebalanced portfolio remains more diversified and as such suffers from a lower variance drain.