I have been looking for an indicator of market randomness for a long time and I believe I have now found a reliable one based on extensive backtesting results. The indicator shows when market activity is random and also when non-random activity gets over-extended and the market is due for a correction.
In academic papers there are often references to the Hurst exponent as a measure of (non) randomness but I have not been able to utilize this concept in developing a trading system. A trading indicator must be suitable for trading – not just for the analysis of time series – and in the case of the Hurst exponent that did not seem to be the case although I cannot discount the possibility that some others have found a way of using it in this context.
The Randomness Index (RI) indicator I have just developed is calculated based on a past period, just like in the case of a moving average, and it is expressed as a percentage. Usually it runs between +100% and -100% although values above and below those extremes are also possible. This is the important part: values between +10% and -10% denote a random trading range when the value of the indicator is on a downtrend or sideways motion only. On the SPY chart below, the bottom pane shows with red color the periods of random trading, again in the context of this indicator calculated for a period of 250 trading days:
According to this indicator, random trading activity is a small subset of overall trading activity as shown on the bottom pane and contrary to popular opinion. The most recent period of random trading activity in SPY was between August 2011 and July 2012, i.e. a full year of random action. This also offers another explanation as to why trading and in particular trend-following has been hard lately. However, in the past few weeks the index has jumped to above 30% and it is approaching levels that in the past have occurred just before major or minor corrections.
Below the same indicator is plotted for QQQ. It may be seen that random trading not only occurred during the same recent period as in SPY but also recently from the start of February to the end of April of this year:
Lastly, the GLD case. It is shown that the recent down-break turned non-random but as of last week trading has gotten random again:
Significance and curve-fitting issues
This indicator is useful because when used “as is” to develop a long/short symmetric trading system for SPY it produces a highly significant system at the 99.9% level when compared to random trading. In addition, the same system without modifications is profitable in many other indices, US and international, as well as in several other ETFs. This reduces the chance of curve-fitting substantially. Details will be offered in another post in the next couple of days.
I will keep the code of this indicator proprietary as it appears to provide some type of an edge. The reason I posted this is for stimulating thinking and showing that contrary to common opinion trading in SPY is mostly non-random according to some indicators. This makes more sense because traders, investors and robots usually flock as it is happening at this point in time. At the same time this frequently leads to corrections, minor and major.
Disclosure: no relevant position at the time of this post and no plans to initiate any positions within the next 72 hours..
Charts created with AmiBroker – advanced charting and technical analysis software. http://www.amibroker.com/