Advisor consensus fails as weekly contrarian timing

The series that was tested

A weekly advisor-consensus series was examined as a short-horizon contrarian signal from 18 September 1981 through 9 February 1991. Advisor-consensus is the weekly share of investment-advisor letters classified as bullish versus bearish, used as a crowding reading.

One testable contrarian-procedure

A contrarian-procedure is a fully specified rule that treats crowded bullishness as a reason to exit or stand aside and crowded bearishness as a reason to enter, including the holding period and the cases where no trade is taken. The crowding-threshold is the consensus percentage at which a reading is treated as extreme enough to flip positioning. Common cutoffs in the record are 55 percent, 60 percent, and 65 percent.

The editorial task is to keep that rule as one procedure for entry, exit, and abstention, then ask whether the next week is the right horizon.

One-week hit rates in the record

A one-week-hit-rate is a count of whether the broad equity market rose or fell in the week after a crowded reading, used to test the rule as a short-horizon signal rather than as a regime-overlay.

When bearish consensus reached 55 percent, the following week advanced in six cases and declined in eight. Across 51 weeks with a 55 percent bullish consensus, the following week advanced 30 times and declined 21 times. A 65 percent bullish extreme appeared seven times, and the following week declined in only two of those cases.

The editorial reading of those counts is that a crowded print did not reverse the next week in a stable way.

Crisis timing and delayed entries

A 26 March 1982 reading of 60.9 percent bears and 25 percent bulls was treated as a clear contrary entry, but the advance that reading was meant to capture did not arrive until August 1982.

On 23 January 1987 bullish consensus was 63.9 percent and bearish consensus 12 percent, and a crowded-bullish mark was still 60.8 percent on 8 August 1987. Standing aside from January would have been out before the October 1987 break and would also have missed an intervening 500-point rise through August.

At the 11 December 1987 low of 1766.74, advisors were 29.5 percent bullish and 39.3 percent bearish, and a roughly 55 percent bearish reading did not appear until 27 May 1988, after a 200-point rebound.

Before the 190-point drop on 13 October 1989, bullish consensus sat between 50 and 54 percent in three episodes beginning 15 September 1989, so whether the rule stood aside depended on how the crowding-threshold was read.

On 14 September 1990 advisors were 56 percent bearish versus 29.6 percent bullish and the following week fell 60 points. By the mid-October low just below 2400, bearish consensus was 55.8 percent, and a lopsided bearish tilt persisted into the 15 January 1991 deadline.

The editorial point is that crisis timing broke down even when a crowding-threshold was met: entries arrived months early, stand-asides missed large advances, and some breaks sat just below the cutoff.

Crowding as a slower regime overlay

A regime-overlay is a weeks-to-months backdrop from crowding or positioning that frames whether a trade belongs in the book, instead of forecasting the next weekly print. Commitment-of-traders is a positioning report used to place one idea in a diversified or regime-aware context, on a horizon of weeks to months rather than as a next-week forecast.

The editorial placement keeps advisor-consensus crowding in that same slower slot. It does not turn the weekly letter share into a one-week flip, and it does not treat commitment-of-traders as a next-week forecast.