It's a strange thought for some of us that so many people - and, especially, financial economists - believe wholeheartedly in Market Efficiency. The problem for critics of Efficiency is that most publicly-traded, finanical markets are indeed very efficient for most of the time. The evidence for that conclusion is strong. This means that super-normal trading profits, adjusted for risk, aren't conceivable. However, questions remain: are all markets efficient? And are any markets efficient all of the time?
Academics often have an unhealthy addiction to pristine abstractions. This results in a kind of religious fervour in their defence, and a fundamentalist's hatred of alternative, messier facts or theories. While the evidence for market anomalies is also formidable (trends, scams, speculative bubbles and crashes,..., are all not-uncommon.) my own work highlights an interesting inefficiency. Briefly put, if you extract sentiment (i.e. based on emotive word-use) from microblogs about stocks, the resulting times series of bullish and bearish signals can be used to forecast share prices for around 2-30 minutes. The sentiment signals offer the possibility of trading for super-normal profits: