"Trades pass statistical significance test at specified confidence level." In this case, provided the trade distribution stays the same in the future, the probability that the system or trading method will be profitable in the future is at least as great as the specified confidence level.
Put simply, a winning strategy that passes a statistical significance test at a confidence level will remain a winner till the market's nonstationary ways (e.g. in volatilities, trendiness, oscillation etc.) eventually invalidate the strategy.
The best way to analyse the market's nonstationarity and better yet, the probability distribution of your Market Sytem is to find the z scores of the indicators employed in the M Systems and only use them when the indicator means are within a certain range.
The issue is fully described in the following link;
http://www.breakoutfutures.com/Newsletters/Newsletter1108.htm
An indicator making use of the method is found in the link in the Tradetsation Code form. It would be very kind of you to make this method available to NT6.5/7 users.
It is an indicator that calculates z scores ;
x = sample mean
m =population mean
s =standard deviation of the population
n =the sample size.
for a group of indicators and outputs the "% of recent samples with |z| <= 1.96*" in a print log.
To address/tackle the nonstationarity, Indicators that are relatively stable would have a % of recent samples with |z| <= 1.96 output of >45% - and could be safe to include in a Market System.
Indicators that are relatively stable would have a % of recent samples with |z| <= 1.96 output of <10%could be potentially abysmal to a M System in the long run.
It also seemed like something you would be interested in...
Faithfully,
Eddy2d
P.S
*The value 1.96 comes from the z or standard normal distribution tables. The area under the standard normal curve between the z values of -1.96 and +1.96 is 0.95 (95% of the total area)