Retail positioning data: crowd sentiment is not a scalping signal

[Amanhasnoname]

Some brokers publish the percentage of clients long and short on each pair. The data looks tempting as a contrarian signal: when most traders are long, price should fall.

Over longer horizons, there's some evidence that extreme positioning precedes reversals. At scalping timescales, the signal is far too slow and noisy. Positioning figures update every few minutes, reflect one broker's clients, and change slowly relative to the moves a scalper trades.

The data is also biased by the type of client. A broker with many beginners shows a different crowd from one with professionals.

Where it can help is in context. Extreme readings suggest a crowded trade, and a sudden move against the crowd can become violent as stops trigger.

Using it as a mood gauge, not a trigger, keeps expectations realistic.

Checking whether a given source has any statistical relation to subsequent short-term price moves, using stored snapshots, can settle the question for good.

Do you use positioning data, and has it added anything to your decisions?