I know, it sounds unlikely. How can one possibly use Twitter to provide an accurate forecast of gold prices over the days ahead?
It may sound crazy, but researchers at Indiana University have demonstrated how they can use Twitter to predict stock prices. They haven’t tried it specifically on gold, but they have tried it with stocks. And they claim to be able to predict changes in stock prices up to six days in advance, with an accuracy of 87.6%.
That’s some pretty accurate forecasting. No charts, no secret-sauce algorithms – but just with what people are writing on Twitter.
They do it by gauging sentiment, and they use sentiment as a means to determine whether people will buy more stocks, or fewer.
They measure just six different mood sates: happiness, kindness, alertness, sureness, vitality and calmness. The higher the number of people on Twitter who express these mood states, the higher stock prices go.
Would their same methodology work with forecasting gold prices?
I think so, and I think it work even better with gold, because the buying or selling of gold is even more closely linked to sentiment.
It sounds odd, and counter-intuitive. But on another level it makes sense. We know that gold prices are linked to sentiment. And if you can scan millions of messages on Twitter and get an accurate sense of people’s moods, it is logical that this could give you an insight into buying patterns.
You can read the full report on these finding here.