I don’t know what the definition is of insider-dealing in Sunderland, but in the City it means trading shares with privileged information. It doesn’t mean writing newspaper stories that move share prices.

Perhaps I am victim of a very late April fool’s joke but Sunderland university has apparently received £90,000 to study the correlation of press headlines with insider-dealing. It is calling the project Cassandra that raises my suspicions: it stands for Computerised Analysis of Stocks & Shares for Novelty Detection of Radical Activities.

The “novelty” is that it will use artificial intelligence to analyse headlines to detect insider trading and see how news stories impact on price movements.

But why the premise that newspapers must be involved? There is one form of market abuse in which false rumours are given to the press by people who have already bought or sold in the hope publication causes prices to move. There have even been instances of journalists dealing before printing.

Technically, inside false information may still be inside information, though there is no “inside” that it comes from. But most inside-dealing involves people who know of actual planned events – such as a profit warning or takeover – because of their privileged position and who trade accordingly.

Sometimes the insider passes it to a third party like a relative or golf partner who deals – with or without the source’s knowledge – but it remains insider-dealing.

However, newspapers are not involved until after the event, either reporting the price movement if it is material or the subsequent announcement of the warning or bid.

So press headlines do not precede insider-dealing. If there are headlines at all they come after the event. Even if a story is mere speculation, it precedes any price movement.

If the paper has a scoop and the headline causes shares to move, that may justify a leak inquiry but there is no insider-dealing unless an insider has dealt ahead of the story. Once the story is printed the news is in the public domain and not inside information – and share prices will then reflect the credibility of the story.

Sunderland’s Cassandra will be programmed to eliminate false positives – news that justifiably causes shares to move. These actually account for the vast vast majority of price fluctuations, not least because the news often comes from the company, via stock exchange announcements, before it makes the next day’s headlines.

But it would be a flaw to try to correlate news stories with price movements even if stories are sensational or wrong unless they are written to manipulate markets for profit. And the Sunderland students cannot know the motives of the writers.

The financial regulator’s sophisticated computers regularly identify suspicious price movements, but even with the added ability to identify the dealers the success in revealing insider-dealing is low. They probably discover a small fraction of actual market abuse. Perhaps Sunderland will help the City. Perhaps not.

Share this: