Informational vs Analytical Edge

Ask a researcher why their signal makes money and you will usually get one of two answers: "we see something others don't," or "we do something with the data that others don't bother to do." Both are legitimate, but they are different bets, and mixing them up leads to spending money on the wrong side of the problem — buying an exotic dataset when the real edge was always in the modelling, or building an elaborate model on a dataset everyone already has.

What each one is

An informational edge comes from having data, or having it earlier, that most of the market lacks. Satellite counts of retailer parking lots before earnings, credit card panel data, a proprietary survey of shipping schedules — these work because the information genuinely is not in the price yet, for the simple reason that whoever sets the price hasn't seen it.

An analytical edge comes from doing something more careful with data everyone already has. Two managers can both use the same quarterly earnings releases; one builds a naive average of estimate revisions, the other properly weights revisions by analyst historical accuracy and controls for the sector's typical revision pattern. The data is identical. The edge is entirely in the processing.

Different economics, different risks

An informational edge has a real cost of goods sold: the dataset has a price, it may have exclusivity terms, and its useful life is set by when competitors discover and license it too — informational edges tend to have a visible expiration date once a data vendor starts selling more broadly. An analytical edge costs research time rather than data budget, and it decays more quietly: someone else eventually publishes the same technique, or a machine-learning model discovers the same relationship on its own, without anyone needing to buy anything new.

	Informational edge	Analytical edge
Source	Data others don't have or see late	Better processing of shared data
Primary cost	Data acquisition, exclusivity	Research and engineering time
Typical decay trigger	Dataset becomes widely available	Technique gets published or replicated
Defensibility check	Is this data genuinely scarce, or just obscure?	Is the edge in the model, or would a simpler model find it too?

A worked judgement call

A team backtests a signal built from a commercially available options-flow dataset and shows a strong Sharpe ratio. Before celebrating, the honest question is which kind of edge this is. If dozens of other funds subscribe to the same vendor, there is no informational edge left — the data is not scarce, it is merely inconvenient to process. The Sharpe ratio, if real, has to be coming from what the team does with the flow data: a smarter way to filter noise trades from informed ones, say. Framing it that way changes the research plan entirely — instead of negotiating an exclusivity clause with the vendor, the team should be trying to break its own filtering logic, because that logic, not the subscription, is the asset.