Hunt high and low for good data

We checked two on-line sources: comdirect.com and ft.com, as well as a newspaper source, the Financial Times, and found that prices differed markedly in some cases. We were so surprised by our findings that we undertook an exclusive share-price survey and have come up with some startling results.

Ten FTSE-100 companies were randomly chosen from ten different sectors as the basis for the survey, and 15 information sources were selected to gather the high and low share prices. These sources were chosen from a range of media: traditional print (Financial Times and Wall Street Journal Europe); a free London tabloid (Metro); electronic investment information sources (hemscott.net, reuters.com, bloomberg.co.uk, market eye.com and stockpoint.com); on-line offerings from established print publications (ft.com, ftmarketwatch.com, investors week.co.uk and thisislondon.co.uk); and on-line brokers (barclays-stockbrokers.com, comdirect.com and etrade.co.uk).

A sixteenth source, Datastream, was selected as a benchmark. Datastream has been the established provider of financial information for years, and provides a Data Guarantee ensuring the accuracy of all information on its service. All Datastream figures are true 52-week high/low closing prices.All data was gathered on 25 February and all on-line stock prices were searched for by ticker symbol to ensure the same stock was analysed in each case. Prices are in pence.

The table is colour coded to show the highest price (in blue) and the lowest price (in red) published for each company’s 12-month high and 12-month low prices. The range between the highest and lowest of these published prices has been calculated at the bottom of the table. The information providers are ranked top-to-bottom by the average percentage variance from Datastream of their published high/low prices across the ten companies surveyed.

As we analysed the figures it was easy to explain away some of the minor price differences. For example, half-penny differences such as the 502.5p and 503p seen in LloydsTSB’s 12-month low column can, most probably, be attributed to rounding-up. Indeed, it is in the interests of the casual user of an on-line broker, or Metro reader, to present easily digestible figures.

Unfortunately, these small variances are overshadowed by an extraordinary number of irregular prices. Of the 300 prices tabulated, a staggering 81 (27%) disagree with Datastream by more than 5%. Moreover, 51 of these (17%) vary by more that 10%.

The greatest variance recorded was 357.6%: comdirect.com, investorsweek.com and stockpoint.com’s remarkable 8,100p high for 3i Group – a whopping 6,330p over and above Datastream’s 1,770p. This, coupled with a Financial Times’s reported low of 445p (Datastream: 998p) makes 3i one hell of a volatile stock – if we believe what we read.

Second place in the “Trust Me, I’m an Analyst” stakes goes again to the comdirect.com, investorsweek.com and stockpoint.com trio. The blind seem to be leading the blind as they all report Cadbury Schweppes’ 12-month low as 62.5p, 80.6% lower than Datastream’s 322.5p.

Although these three information sources perform worst in our survey, with an average 28.8% variance, they are not the only culprits. Barclays, for example is also 56.6% out on Logica’s 12-month low (583p), and Reuters records an ICI high of 740p, 22.7% above the benchmark – contributing to an overall average variance of 10.10% and 4.8% for Barclays and Reuters respectively.

People making investment decisions on this sort of information should be forgiven their trust in the providers. After all, who would think that Investors Week could get it wrong? And surely the hallowed pages of the FT cannot contain erroneous information? Unfortunately they do and they can.Initially, we thought that the majority of variation could be attributed to differing reporting periods – year-to-date high/lows will differ from true 52 week high/lows. Similarly, intra-day, closing, bid and offer high/low prices will all exhibit a degree of variance. However, when we contacted the organisations, over half claimed to be running true 12-month high and low closing prices. Yet, there is not enough correlation in the results to corroborate this. Hemscott.net, ftmarketwatch.com and investorsweek.co.uk all claim they run true 12-month closing prices but they occupy the first, eighth and 13th positions in our survey respectively.

Most notable sources that report other than 12-month closing prices include the Financial Times which operates a very peculiar system. The small print at the bottom of the London Share Service pages reads: “the highs and lows are calculated from 01/01/00 and relate to closing middle prices”. All prices that claimed to be “52-week” in the main body of the FT were therefore 55-week at the point our survey was undertaken. Nevertheless, this still does not account for the published 445p 3i low, because the stock has not dipped below 600p since early 1999. The FT was still calculating prices from 01/01/00 at time of going to press.

Marketeye.com also has a unique reporting style, despite it being a part of Thomson Financial (which coincidentally owns Datastream). Marketeye claims it publishes 52-week high/low prices in the “fundamental data” section of its website, but it transpires that the figures are actually 1999 January to December closing prices.

When we confronted Marketeye with this fact the spokesperson was a master of understatement. “Our site is not as dynamic as we would like,” he said. “As a company we have a retail outlook – hard data is a valuable commodity.” The inference is that, if you want accurate figures, you have to pay for them. But, marketeye.com still performs better than comdirect.com, investorsweek.co.uk and stockpoint.com.

Reuters also claims that, as investors are getting something for nothing, accuracy cannot be guaranteed. “We only publish soft figures on our website,” said a Reuters press officer. “This stuff is free. We do have paid services where the information is more detailed.” Perhaps the Wall Street Journal Europe should contact Reuters. It cites Reuters as the source of its European market data and runs the “soft figures” every weekday in its “Money & Investing” section.

Metro also claims to derive prices from another surveyed source – Barclays Stockbrokers. But Metro occupies second place in the survey with a mere 0.4% variance – and barclays-stockbrokers.com is in 11th place with 10.10% variance.

Likewise, one would assume that the Financial Times, ft.com and ftmarketwatch.com all carry the same data. Ft.com comes fourth in our survey, ftmarketwatch.com comes eighth and the FT itself occupies a lowly tenth position. In addition, while most prices correlate with at least one other source, every ft.com price is unique in this survey, most notably in the zero values for Powergen’s 12-month high and low.

Whether it is human error, software limitations or a combination of both, something is very wrong with these figures. The London Stock Exchange sends the same raw data to all suppliers, such as Datastream, Reuters and Bloomberg. Pricing information is then manipulated, packaged and sold on to the smaller players, such as the on-line brokers.

Are the majority of on-line brokers at the bottom of our rankings because of a Chinese whispers effect? Perhaps not. Comdirect.com sources its information from Standard & Poor’s Comstock, but so does ft.com and a number of others. They should tally.

Is the Financial Times worth its 85p cover price if more accurate investment information is available free? Are Reuters and Bloomberg sacrificing accuracy where there is no money to be made? And, most importantly, can investors trust on-line brokers to publish accurate as well as instant information? Just now, it’s anyone’s guess.

View price comparison table

This article first appeared in Financial Director magazine.