This is a question I have been pondering for the past few days and it has come about as a result of several conversations Scott L and I have had regrading some testing we are looking at. It seems that the extreme survivor bias of the index is causing all sorts of problems. When you do any form of systems testing you need to be aware of how an index is constructed. Each equity index is made up of stocks that are regularly rotated in and out – in the case of the S&P/ASX 200 this rotation is extreme. When looking at an index you are seeing a collection of winners ( at least temporary winners) this has the effect of skewing any test results you generate.
To get a sense of how bad this was I have just done a very quick count of the stocks that made up the index in February 2013 and those that currently make up the index. Only about half of those from 2013 are still in the index the rest have been binned – by any standards this is a fair amount of turnover. You could even say that it makes the index largely irrelevant for anything other than generating a handy list of stocks to scan – as a metric its utility might be non existent.