"It is not so easy. I am into this research monitoring soil radon since 2006 in Kolkata and parts of north Bengal namely Matigara and Jalpaiguri that come under the active fault zone. What I gathered from the data is that there is a direct correlation between the soil radon anomaly within 1,000 kilometres from the measuring site, and for intensity above 4 in the Richter scale. They occur 7-15 days before an earthquake with few exceptions," said Ghosh, director of the Biren Roy Research Laboratory for Radioactivity and Earthquake Studies, Jadavpur University.
This is a very rational guy. Good thing he doesn't live in Italy! As we know from slow earthquakes, there are probably lots of radon anomalies that don't announce big shakers. Still, I've always found this fascinating. As we know from the Fish Theory of Earthquakes, there must be a 'pre-slip' to reach the critical displacement, where dynamic friction takes over. In a fractured and unstable regime, the effects of this slip can disturb the rock for a huge distance, as shown by discrete element modeling. So naturally, there should be a radon anomaly.
As the man says, this is probably useless for earthquake prediction. The nasty Italians want an exact date, size, and 100% certain 24 hour window, with no false alarms. Fat chance!