Skip to main content

AAII & Investor Surveys

Updated Jun 24, 2026 at 2:35pm

Research Draft High 1,178 words

Investor sentiment surveys poll a defined population — individual investors, newsletter writers, or fund managers — on whether they expect stocks to rise or fall, then publish the share of bulls, bears, and neutrals. The flagship for retail sentiment is the AAII Investor Sentiment Survey, run weekly by the American Association of Individual Investors since 1987. The central tension is that these surveys are almost always read contrarily: the crowd's collective conviction is treated not as a forecast to follow but as a warning that positioning has become one-sided. The catch is that "the crowd is wrong at extremes" is a far weaker, noisier rule than its popularity suggests.

How it's measured

AAII asks members one question each week: "Do you feel the direction of the stock market over the next six months will be up (bullish), no change (neutral), or down (bearish)?" Originally a mailed survey to a random member sample, it has run online since 2000, with results published each Thursday. Output is three percentages summing to 100% plus the derived bull–bear spread (bullish % minus bearish %). Over the survey's life AAII reports long-run averages of roughly 38% bullish, ~31.5% neutral, and ~30.5% bearish (AAII; the exact figures drift as the running mean updates, and earlier published cuts ran nearer 38–39 / 31 / 30). AAII publishes one- and two-standard-deviation bands (the bullish series has historically carried a standard deviation of roughly 10–11 percentage points per AAII) to flag "unusual" readings.

Related surveys differ mainly by who they poll:

  • Investors Intelligence (II) Advisors' Sentiment — devised by Abe Cohen of Chartcraft in 1963, it classifies the stance of 100+ independent market-newsletter writers as bullish, bearish, or "correction" (bearish-but-buy-lower). It surveys advisors, not the public, and was the first of these polls to be reinterpreted contrarily by its own creator.
  • NAAIM Exposure Index — reports the average actual equity exposure of active money managers (not an opinion poll).
  • Sentiment composites like CNN's Fear & Greed Index or the University of Michigan Consumer Sentiment survey overlap conceptually but mix in market-derived inputs (put/call, breadth, volatility) rather than pure polling.

A key distinction: AAII and II are survey ("soft") sentiment — what people say. They sit alongside market-derived sentiment (put/call ratios, the VIX, fund flows) that captures what people do. The two often disagree.

How it's used in practice

The standard reading is contrarian and applied at extremes, not week to week:

  • Extreme bullishness (high bull %, wide positive spread) is read as a caution flag — optimism is largely "priced in," leaving few marginal buyers.
  • Extreme bearishness (high bear %, deeply negative spread) is read as a setup for a bounce — pessimism implies sidelined cash and capitulated sellers.

Most practitioners smooth the noise: a 4-, 8-, or 10-week moving average of the bull–bear spread, or readings beyond AAII's ±1σ / ±2σ bands, rather than a single weekly print. Surveys are almost universally treated as secondary, confirming indicators — a sentiment overlay on a decision driven by price, trend, or fundamentals — not a standalone trigger. They are weekly and survey-based, so they are far too slow and coarse for intraday or short-swing timing.

Adoption, debate & evidence

AAII is among the most widely cited retail-sentiment gauges in financial media, and II is a fixture of the newsletter-writer "smart money / dumb money" framing. Adoption is broad; rigorous evidence of edge is thinner.

What holds up: AAII's own research notes that extraordinarily low optimism has historically preceded above-average six- and twelve-month S&P 500 returns, and bearish sentiment hit a survey-record 70.3% on March 5, 2009 — essentially the bear-market bottom. That fits the contrarian story at genuine panic extremes.

What is weaker / contested:

  • AAII explicitly cautions that readings beyond one standard deviation have a "mixed record" of signaling direction — the relationship is statistical and noisy, not reliable.
  • The well-known anecdote that II newsletter writers tend to be "almost always wrong at turning points" is part of why the survey flipped to contrarian use, but it is folklore-grade, not a controlled result.
  • Academic work on sentiment-and-returns centers on the Baker–Wurgler (2006) market-based sentiment index (closed-end fund discount, IPO volume/first-day returns, turnover, dividend premium, equity-issuance share) — not the AAII poll. Baker–Wurgler found sentiment predicts the cross-section (speculative/young/high-volatility stocks underperform after high sentiment), and chiefly during high-sentiment periods. Even there, later work argues much of the predictive content tracks business-cycle variables, with the residual "pure sentiment" component weak. A 2026 replication in Economic Inquiry (Leong et al.) reproduces the cross-sectional effect in both the original 1963–2002 sample and a fresh 2002–2023 sample — confirming the core result — while finding the predictive strength can shift or even invert over time and across markets.
  • Do not let the Baker–Wurgler academic result lend credibility to the AAII poll — they are different instruments. Standalone backtests of mechanical AAII bull–bear thresholds show modest, inconsistent, regime-dependent results, not a robust tradable edge.

Honest summary: surveys reliably describe crowd positioning and occasionally flag true capitulation/euphoria, but as mechanical timing signals their measured edge is small and unstable.

Strengths & limitations

Strengths: cheap, transparent, long history (AAII since 1987, II since 1963), and a direct read on positioning that price alone can miss. Most useful at genuine emotional extremes (2009 lows, 2000 peak-type euphoria), where they confirm capitulation or complacency.

Limitations: small, self-selecting samples (AAII members are engaged hobbyist investors, not a representative public); high week-to-week noise; no fixed threshold that consistently marks tops or bottoms; and "extreme" can persist for months in a strong trend — sentiment can stay extreme while price keeps going. The #1 misuse is treating a single weekly print as a timing trigger ("bulls are high, sell") rather than as a slow, smoothed, confirming overlay. Surveys also lag — by the time pessimism is extreme, much of the decline may be done.

Sources

Disputes flagged: (a) the II "advisors are wrong at turns" claim is folklore, not a controlled study; (b) academic predictability evidence applies to the market-based Baker–Wurgler index, not the AAII poll — kept separate above; (c) mechanical AAII threshold edge is regime-dependent and weak.