The link: So Many Jumps, So Few News (from NBER).

From the abstract: We find that most relevant news, both idiosyncratic and systematic, lead quickly to price jumps, as market efficiency suggests they should. However, in the reverse direction, the vast majority of price jumps do not have identifiable public news that can explain them, in a departure from the ideal of a fair, orderly and efficient market.

Addendum from the paper: It is possible that some of the jumps we observe are due to trading linked to news that are not publicly available, i.e., insider or other form of private information trading. (As we detail below, we design time windows to control for the possible advance leakage of soon-to-released public news.) We view this explanation as unlikely to rationalize the bulk of jumps without news: a firm is unlikely to generate significant information, private or public, at a pace consistent with the high rate of incidence of jumps we find in the data. Furthermore, as predicted by sequential trade models, an insider endowed with private information is expected to trade slowly to exploit and avoid revealing the content of his or her information, making a rapid price adjustment, i.e., a jump, fairly unlikely, compared to the situation where fast traders acquire public information sooner than their slower counterparts and know that their window to act is short.

So what? It's interesting, isn't it? If most price jumps don't come from public news, and they are too frequent to plausibly come from fresh public or private information, then what drives those jumps? One possibility is that we underestimate the range of types of information that count as relevant to prices; as an example, I wrote last year informal paper about company technology portfolios with a valuation model more sensitive to the sort of exogenous variables that aren't usually reflected in news.

More generally, there might be more cognitive heterogeneity across market actors than we assume in terms of their models of the world and their lags in processing information; if contextless memes on a Twitter account can move prices significantly, what other let's say unusual information signals are out there outside systematic analysis?

If you're a fund, this is all great news. Other people's epsilon is where you get your alpha from. It looks like there might be a lot lying around.