Multi Manager Hedge Funds
According to academic research, 40% of positive earnings results make the stock price go down. So, even if you correctly call if a stock will beat earnings than there is still a high chance that you lose money. With this level of uncertainty how exactly are the LS equity pods at multi manager hedge funds making money?
The topic question is sort of based on a misunderstanding.
1. "Beat on earnings," has two meanings. One is literal, that the stock's EPS was reported to be above Bloomberg consensus. The second is colloquial, but more accurate: the stock outperformed the market's weighted expectation of results on a broad number of KPIs. The academic study will look at reported consensus for EPS, but that's the wrong measure for a few reasons
a) Companies guide EPS conservatively, and sell-side models out company guidance. Investors know this, so the true market expectation is consistently higher than "consensus EPS" as reported by Bloomberg or some other source. Thus, "beating consensus EPS" is relatively meaningless.
b) Sometimes EPS just isn't that important a metric. Many companies have KPIs that are more relevant than EPS (bookings, billings, backlog, new users adds, etc.), and beats vs. consensus on those measures matter more.
2. Multi-manager analysts and PMs know this, and thus try to predict outperformance vs. market consensus (as opposed to reported sell-side consensus) on the KPIs that matter. In broad outlines, the way they try to figure out what they need to predict is:
a) Talk to sell-siders who are "in the information flow" and go to hedge fund idea dinners and such, asking what the "bar" is for various KPIs. In short, try to get a better sense of what people really expect.
b) Measure how the stock goes up and down in response to different KPI levels of outperformance vs. reported consensus or other measures of expectation. See which KPI correlates best with stock outperformance. Then try to predict that KPI.
That's the basics of it. Obviously in practice it's tricky and ugly. But it's not just "mechanically out-predict reported consensus EPS."