New Analysis Shows There's A Signal And Noise In Asset Markets

A new analysis shows that there is a signal and noise in asset markets, Stuart Reid wrote on Turing Finance. http://www.turingfinance.com/lossless-compression-algorithms-and-market-efficiency/#conclusion Reid developed a "meta test" for market randomness tests based on greedy lossless compression algorithms which indicate that such tests are highly sensitive to any noise. He also applied greedy lossless compression algorithms to binarized and de-trended market index returns and spoiler alert. Following are the key conclusions from the analysis: "If the sensitivity of the test to noise is greater than our trading strategies robustness in the presence of noise, the negative result does not imply that markets are efficient in the sense that they cannot be beaten." "Our toy market shows that the sensitivity of the greedy lossless compression algorithms used to test market efficiency can be as much as 10x more sensitive to noise than a simple betting strategy." Reid said further research include a study of lossy compression algorithms most notably, a stacked auto-encoder neural network.