Market sentiment

Sentiment is not observable. What is observable is a set of proxies, each measuring something narrower than the word suggests: surveys of investors or fund managers, reported positioning in futures markets, the ratio of long to short accounts published by some brokers, option based volatility measures such as a volatility index, flows into and out of funds, and the behaviour of instruments that historically move together in risk seeking and risk avoiding phases. Each proxy covers a different population over a different period, which is why two sentiment readings can point in opposite directions on the same day without either being faulty.

Two conventions attach to the readings and both are disputed. The first treats sentiment as confirmation, on the reasoning that a prevailing disposition is what sustains a move. The second treats an extreme reading as contrarian, on the reasoning that once nearly everybody is positioned one way there is little unexpressed demand left to push price further. The two conventions contradict each other at exactly the readings that get quoted, and neither has a settled empirical record: the results in published tests vary by market, by proxy and by the period tested.

Three limitations are worth holding. Positioning data is usually published with a lag, so a reported figure describes a market that has since moved. Retail positioning covers one segment of participants and not the institutional flow that sizes most markets. And sentiment carries no timing at all: a reading can persist at an extreme for a long period, which is the reason it is treated as context by the practitioners who use it rather than as a trigger.