This is something that we are working hard on. Backtesting is a double edged sword. There are some complex issues to grapple with in it, not least survivorship bias and look-ahead bias, which can make it unreliable.
- Survivorship bias - occurs when performance results are based upon a data set that does not include companies that have delisted or gone bankrupt in the interim. (Our current data set has survivorship bias)
- Look Ahead bias - occurs when performance results use historical data that was not available at the time the results were announced (e.g. due to companies later restating their results). (Our current data set is restated historically to reflect the latest announcements).
Many of the systems commercially available to investors are based on databases that fail to accurately deal with these issues. As such, they are potentially more harmful than helpful. There are many, many examples of investors who have convinced themselves of the merits of a strategy though backtesting, only to come horribly unstuck when they try to apply that strategy in reality.
The availability of so called 'Point in Time' databases is prohibitively expensive and generally only available to hedge funds and professional investors at subscription levels of hundreds of thousands of dollars annually.
As a result, all the tracking we show has been actual / real-time based on our own portfolios held and rebalanced. It is not back-tested as we didn’t want to use results we didn’t believe in.
In the meantime, there is an excellent American web-site called AAII that is worth referring to– it has been tracking a range of Guru strategies live since about 2002 (the first few years of results were apparently back-tested). Their approach/rules tend to be different but it’s indicative, albeit for the US market, not UK stocks.
There’s also a very good academic paper which has back-tested (in a rigorous way) the NCAV strategy on the London Market.