Auto Correlation ( Trade Station )

[Igor 10]

Detecting Nonrandomness in the Markets

 

Economists have long debated whether stock market prices change randomly or are at least partially predictable. The random walk hypothesis says that stock market prices change in a random fashion, making prediction impossible. However, more recent research has uncovered a variety of inefficiencies in the markets that disprove the random walk hypothesis. Certainly, any trader who has followed the markets for any length of time, let alone successfully traded them, doesn't need an academic study to tell them the markets are not entirely random. Nonetheless, there's some benefit to being able to demonstrate it mathematically.

AutocorrelationInds.ELD

auto correlation.txt

[BlueHorseshoe 10]

Auto correlation can be especially useful when applied to things which do move in obvious cycles (this is not a comment on the random or non-randmness of price) such as volatility.

 

Consider, if volatility is rising, and volatility is highly auto-correlated, then a stop-loss based on current volatility is not likely to be sufficiently large. When volatility is falling there would be a good argument for tightening a stop-loss where the auto-correlation was high.

 

Also, consider the applications of high (or low) coefficients of auto-correlation for binary strategies.

 

BlueHorseshoe

[AndrewJackson 10]

Thanks blue horseshoe! Is there any special format styles and inputs you think are best?

Edited March 2, 2013 by AndrewJackson