The perils of filtering stocks

This is the second article in the series “From ideas to investments” in which I look at the process of finding and evaluating investment ideas. Fishing for stocks using filtering, or stock screening, is a quick way to create a shortlist of shares that are likely to perform well – but only if we buy the whole shortlist or pick from it wisely.

Filtering 101

Filtering is a method of finding investments with certain characteristics. Once we’ve designed and set up our filter, we can, in an instant, narrow the field of potential investments down from the many thousands of instruments included in SharePad to a more manageable list. That is a lot easier than going through them one by one.

The output of the filter is a list of shares that meet the criteria we selected. SharePad allows us to filter tens of thousands of investments including funds, investment trusts, bonds, and shares.

When it comes to shares, there is a huge range of criteria. We can home in on big companies, small companies, companies listed in London or on a large number of foreign exchanges. We can deploy an array of financial ratios to find companies that are particularly profitable or cash rich and we can hunt for bargains trading on low valuations. We can look for companies whose directors own large shareholdings, or companies whose directors have been buying or selling shares recently. We can find out which companies have grown revenue or profit in recent years, and companies which brokers believe will grow in future years.

I will explain why we might filter for specific criteria in future articles but a substantial amount of testing on historical data by professional investors and professors, shows it is possible to improve our returns using just a few them. In other words, we can tip the playing field slightly in our favour.

The perils of systematic trading

One way to capture this value is to blindly follow a system. For example, buying the cheapest shares in the market (according to the statistics), holding them for a year then selling them and starting all over again.

There’s an important caveat though. The returns promised in tests of the longest established filters, for example by using common valuation ratios like price to earnings or price to book value, are not typically more than a few-percent higher than the returns from the indices they are benchmarked against. In some years they will be lower. It is difficult to stick with a strategy when it doesn’t always seem to be working even though the extra returns should compound into a substantial sum over many years.

Another problem for would-be systematic private investors is the data. SharePad contains an immense amount of data about the shares we can invest in today, but to test a strategy properly we also need the data for shares we could have bought many years and decades ago that are no longer listed. These companies may have gone bust, gone private, or they may have been taken over. Such data sets are incredibly expensive – beyond the means of most of us. If we cannot test a system, it is difficult to have faith in it.

Although I’ve been investing and writing about the stock market for over twenty years I’ve never met a private investor who invests entirely by the numbers, without any attempt to interpret them, or dig deeper into the business and its prospects. My own attempt to design a system that relies only on the data and test it against my performance as a stock picker has produced very poor results, comparatively, over the last eight years.

I remain open-minded, though. We hear that hedge funds and investment bankers with huge computing resources beat the market with quantitative strategies (though we hear less about the failures) and perhaps even quite basic techniques can beat the market if we persevere with them. If you are a systematic trader, also known as a mechanical investor, using affordable tools like SharePad, I’d love to hear from you.

The perils of cherry-picking

The alternative to systematic investing is to use the time-saving benefit of filters to screen out businesses that we are very unlikely to invest in and cherry-pick investments from the output. Unsurprisingly perhaps, given my experience with systematic trading, this is the approach I favour. But if we are going to get our hands dirty, it is important to know what we will be handling.

Financial statistics and ratios are very blunt instruments when it comes to prediction. Measures of profitability and value, for example, frequently used to discover ‘good’ businesses and ‘cheap’ businesses respectively, often depend on profit, but company profits can be extremely variable. Just because a company makes a lot of profit in a particular period, it doesn’t mean it will necessarily continue to profit, or grow profit, in future. If a company does worse than people expect, its share price will probably go down regardless of what the statistics said.

A collection of 50 or so statistically cheap shares, like the baskets of stocks tested by academics and professional quantitative analysts, isn’t a collection of shares growing in value at a uniform 2% or 3% more than the broad market year-on-year. It is a mad boiling cauldron of potent ingredients. Some of these shares, like Carillion earlier in the year, will probably go to zero, and others could double, treble, or more. The individual performances are nothing like the average.

While it is tempting to believe we can cherry-pick investments from a list we’ve filtered, we can’t just pick a share or a handful of shares on the assumption they will perform like the whole basket. They almost certainly won’t. Either we must buy the whole basket like a systematic investor, or we need be able to discriminate between the companies that will do well and the companies that will do badly. For the fundamental investor this means making judgements about the validity of the statistics we are relying on and the prospects of the businesses they are identifying.

Cherry-pickers are not just playing the numbers game like the systematic investors, but starting with similar output, a filtered list, we too are tilting the playing field slightly in our favour. Whether we improve our returns, or compromise them, depends on our judgements.

A good mindset for investing

I’ve focused on the perils of filtering today to get us in the right mindset to build our own filters and generate ideas for investment. If we choose filters that have stood the test of time, preferably many decades, and filters that fit in with what we know about the way stock markets work and the way businesses perform, we will get off to a sound start.

In future articles we will build basic filters using criteria that have helped earn investors outsized returns in the past. These will include well-known criteria like value and profitability but also some less well known criteria like ‘seasoning’ – how long a company has been listed on the stock market.