Every major sentiment indicator dissected, quantified, and stress-tested. When the crowd panics, the data whispers “buy.” When the crowd celebrates, the data murmurs “reduce.”
In 1895, Gustave Le Bon published The Crowd: A Study of the Popular Mind, arguing that individuals, once submerged in a group, lose their capacity for rational thought and become subject to contagion, suggestibility, and emotional extremism. A century later, his observations remain the single best explanation for why markets overshoot in both directions.
Charles Mackay’s Extraordinary Popular Delusions and the Madness of Crowds (1841) documented the same phenomenon across centuries: the Dutch tulip mania of 1637, the South Sea Bubble of 1720, the Mississippi Company. The pattern is always identical — a plausible narrative, growing participation, parabolic price action, and catastrophic collapse. What changes are the names, not the structure.
Kahneman & Tversky (1979) formalized what Le Bon observed intuitively. Their Prospect Theory demonstrated that humans feel the pain of losses roughly 2.5x more intensely than the pleasure of equivalent gains. This asymmetry drives the entire sentiment cycle: investors hold losers too long (loss aversion), sell winners too early (disposition effect), and panic-sell at bottoms (capitulation).
Key biases that drive sentiment extremes:
This is why sentiment indicators work. When the crowd reaches an extreme — when everyone who was going to sell has sold, when every bearish argument has been priced in, when the last margin call has been met — there are simply no sellers left. The market reverses not because of good news, but because of seller exhaustion.
Commercial hedgers (companies that use futures to hedge their actual business), corporate insiders (executives buying their own stock), and institutional positioning data consistently show that informed participants do the opposite of the crowd at extremes. When retail is panic-selling, insiders are buying. When retail is euphoric, insiders are distributing. This asymmetry is measurable, repeatable, and tradeable.
The American Association of Individual Investors has conducted a weekly sentiment survey of its members since July 1987 — making it one of the longest-running, most-studied contrarian indicators in existence. Each week, members answer a single question: “Do you feel the direction of the stock market over the next six months will be up (bullish), no change (neutral), or down (bearish)?”
Bullish: 37.5% • Bearish: 31.0% • Neutral: 31.5%
Standard deviation (bearish): 10.7 percentage points. A reading above 50% bearish is approximately a 2-sigma event. Above 60% is a 3-sigma event. Above 70% has happened exactly once in 39 years.
The following table documents every instance where AAII bearish sentiment exceeded 55% — a level reached fewer than 20 times in nearly four decades. The forward return data is unambiguous:
| Date | Bearish % | S&P 500 | Event | Fwd 1M | Fwd 3M | Fwd 6M | Fwd 12M |
|---|---|---|---|---|---|---|---|
| Mar 5, 2009 | 70.3% | 683 | GFC exact bottom | +8.5% | +25.9% | +36.2% | +53.6% |
| Oct 9, 2008 | 64.7% | 910 | Lehman aftermath | -16.8% | -23.5% | -10.1% | +14.2% |
| Mar 19, 2020 | 52.1% | 2,304 | COVID crash — 4 days before bottom | +12.7% | +32.0% | +42.3% | +56.4% |
| Sep 22, 2022 | 60.9% | 3,693 | Rate hike panic | +8.0% | +5.6% | +9.5% | +21.6% |
| Oct 6, 2022 | 56.2% | 3,640 | Near exact Oct 2022 bottom | +7.9% | +3.2% | +10.8% | +21.2% |
| Apr 10, 2025 | 61.9% | 5,268 | Tariff shock | +9.3% | +12.8% | +14.5% | TBD |
| Dec 19, 2018 | 50.3% | 2,507 | Christmas Eve selloff | +7.9% | +11.7% | +17.3% | +28.9% |
| Feb 7, 2003 | 56.0% | 829 | Post-dot-com bear | +0.3% | +14.6% | +21.2% | +38.5% |
When AAII bearish exceeds 50%, the average forward 6-month return is +15.4% and the average forward 12-month return is +33.5%. The signal has a hit rate of approximately 90% for positive 12-month returns. This is one of the most statistically robust contrarian signals in all of market data.
Sustained bullish readings above 55% for 3+ consecutive weeks have historically preceded short-term market peaks. Examples include January 2000 (dot-com top), October 2007 (GFC top), and late November 2024 (post-election euphoria preceding the 2025 tariff correction). However, bullish extremes are less reliable than bearish extremes — markets can stay irrationally exuberant longer than they stay irrationally fearful.
The survey suffers from sample bias: respondents are older, wealthier, and more educated than the average investor. The sample is US-only and tends to skew male. Response rates vary, and the sample size (typically 300-500 respondents) is small. These limitations argue for using AAII as one component of a composite, not a standalone signal.
CNN’s Fear & Greed Index aggregates seven distinct market signals into a single 0–100 score, providing a real-time snapshot of market emotion. Unlike AAII (survey-based), this index is derived entirely from market-based data, making it immune to self-reporting bias.
S&P 500 vs. 125-day moving average. Above = greed. Below = fear.
52-week highs vs. 52-week lows on NYSE. More highs = greed.
McClellan Volume Summation Index. Positive breadth = greed.
5-day avg CBOE put/call. High ratio = fear. Low ratio = greed.
HY spread vs. IG spread. Tight = greed. Wide = fear.
VIX vs. 50-day moving average. High VIX = fear.
Stocks vs. Treasuries return differential (20-day). Bonds outperforming = fear.
| Date | F&G Score | Zone | Event | S&P Level | Fwd 1M | Fwd 6M |
|---|---|---|---|---|---|---|
| Apr 8, 2025 | 3 | Extreme Fear | Tariff shock — lowest since Mar 2020 | 5,062 | +15.4% | +18.7% |
| Mar 16, 2020 | 2 | Extreme Fear | COVID crash — near all-time low | 2,386 | +20.8% | +42.5% |
| Dec 24, 2018 | 5 | Extreme Fear | Christmas Eve panic | 2,351 | +7.9% | +17.3% |
| Feb 9, 2018 | 8 | Extreme Fear | Volmageddon | 2,581 | +5.4% | +8.3% |
| Aug 24, 2015 | 6 | Extreme Fear | China devaluation scare | 1,893 | -1.4% | +7.8% |
| Nov 25, 2024 | 92 | Extreme Greed | Post-election euphoria | 5,988 | -2.5% | -1.8% |
Historically, when F&G drops into single digits and then recovers above 25, the subsequent rally is sustainable. This is because the initial bounce from extreme fear is often driven by short-covering — volatile and unreliable. But once sentiment recovers above 25, it signals genuine buying interest and the all-clear for trend followers to add exposure. In March 2020, the recovery above 25 came on April 6 — confirming the bottom with the S&P at 2,663 (still +40% below the eventual peak).
The CBOE Volatility Index (VIX) was created by Professor Robert Whaley in 1993 and redesigned in 2003 to use S&P 500 options (instead of S&P 100). It measures the 30-day implied volatility of SPX options, expressed as an annualized percentage. Colloquially, the “fear gauge.”
| VIX Level | Regime | Interpretation | Frequency | Signal |
|---|---|---|---|---|
| 10–15 | Complacent | Extreme calm, often precedes turbulence. Portfolio insurance is cheap. | ~22% of trading days | Caution |
| 15–20 | Normal | Healthy market with typical uncertainty. Neutral signal. | ~38% of trading days | Neutral |
| 20–30 | Elevated | Correction territory. Hedging activity increases. Opportunities forming. | ~28% of trading days | Monitor |
| 30–40 | Crisis | Genuine fear. Historically positive 6-month forward returns. | ~9% of trading days | Opportunity |
| 40–50 | Panic | Severe dislocation. Capitulation likely occurring. Strong contrarian buy. | ~2.5% of trading days | Strong Buy |
| 50–80 | Max Panic | Generational events. Every instance preceded massive forward returns. | ~0.4% of trading days | Maximum Conviction |
| 80+ | Armageddon | Has occurred twice: Oct 2008 (80.86), Mar 2020 (82.69). Literal once-in-a-decade events. | ~0.02% of trading days | Generational Buy |
The VIX term structure — the relationship between VIX spot and VIX futures at various expirations — provides an additional layer of intelligence beyond the spot level alone.
VIX futures trade above the spot VIX. This is normal: investors expect more uncertainty further into the future. The term structure slopes upward (Front month < 2nd month < 3rd month). This is the “risk premium” — you pay more to hedge further out.
VIX futures trade below the spot VIX. This means present panic exceeds expected future volatility. The market is saying: “Things are terrible right now, but will probably calm down.” Historically, backwardation is a positive forward signal — the market is pricing in recovery. Average 3-month forward return when VIX is in backwardation: +8.4% (vs. +2.9% baseline).
| Date | VIX Close | Event | S&P Drawdown | Fwd 3M | Fwd 6M | Fwd 12M |
|---|---|---|---|---|---|---|
| Mar 16, 2020 | 82.69 | COVID pandemic crash | -33.9% | +28.0% | +42.5% | +56.4% |
| Oct 27, 2008 | 80.86 | GFC — Lehman collapse | -48.8% | -7.4% | +1.5% | +23.5% |
| Aug 5, 2024 | 65.73 | Yen carry trade unwind | -8.5% | +9.2% | +14.8% | +19.7% |
| Apr 8, 2025 | 52.33 | Tariff escalation panic | -19.4% | +14.6% | +18.2% | TBD |
| Mar 12, 2020 | 75.47 | WHO declares pandemic | -26.7% | +25.3% | +39.8% | +52.7% |
| Feb 5, 2018 | 37.32 | Volmageddon (XIV collapse) | -10.2% | +5.4% | +8.3% | +2.1% |
| Aug 24, 2015 | 40.74 | China devaluation fears | -12.4% | +5.2% | +7.8% | +12.5% |
| Oct 15, 2014 | 31.06 | Ebola scare + oil crash | -7.4% | +8.7% | +4.9% | +5.3% |
| Sep 17, 2001 | 43.74 | Market re-opens after 9/11 | -11.6% | +11.4% | +5.2% | -12.3% |
| Oct 8, 1998 | 45.74 | LTCM / Russia crisis | -19.3% | +21.3% | +28.5% | +33.0% |
Jason Goepfert’s SentimenTrader service has tracked the divergence between informed and uninformed market participants since 2001. The Smart Money / Dumb Money Confidence spread is among the most powerful timing tools ever constructed.
| Configuration | Smart Money | Dumb Money | Spread | Historical Interpretation |
|---|---|---|---|---|
| Maximum Opportunity | > 65% | < 35% | > 30 pts | Smart money buying what dumb money is selling. Avg fwd 3M: +9.2% |
| Mild Opportunity | > 55% | < 45% | > 10 pts | Divergence emerging. Avg fwd 3M: +4.7% |
| Neutral | 45-55% | 45-55% | ±10 pts | No signal. Avg fwd 3M: +2.5% (baseline) |
| Warning | < 40% | > 60% | < -20 pts | Dumb money euphoric, smart money retreating. Avg fwd 3M: -1.3% |
| Maximum Warning | < 35% | > 70% | < -35 pts | Extreme divergence. 85% probability of correction within 8 weeks. |
The model nailed every major bottom since 2003 — March 2003, March 2009, October 2011, February 2016, December 2018, March 2020, October 2022, and April 2025 — with Smart Money Confidence spiking above 65% at each trough. It similarly flagged every major top with Dumb Money surging above 70% (January 2018, September 2018, February 2020, November 2021, July 2024, November 2024).
At $500/year, SentimenTrader provides the highest return on investment of any market data service available to individual investors. The Smart/Dumb spread alone is worth the cost, but the service also includes: Optimism Index (a composite for every asset class), sector-level sentiment, options sentiment models, insider transaction aggregation, and Goepfert’s daily commentary — which has been consistently among the most level-headed voices in financial media for over two decades. There is no adequate free substitute.
The CBOE publishes daily put/call ratios for three categories: total (all options), equity-only, and index-only. For contrarian analysis, the equity-only put/call ratio is the most useful because it captures retail sentiment most directly.
Extreme fear (P/C > 1.23): Statistically significant above-average returns at 30 and 60 days. The signal works because high put-buying reflects hedging and panic, which creates a floor of support — those puts will eventually expire, reducing downward pressure. Average forward 30-day return: +3.8% (vs. +1.1% baseline). P-value: 0.003.
Extreme greed (P/C < 0.72): Below-average forward returns, but not statistically significant. The asymmetry is critical: fear signals are more reliable than greed signals. This makes intuitive sense — panic is sudden and intense (V-shaped bottoms), while euphoria builds gradually and can persist (prolonged topping processes).
Optimal holding period: 7–15 trading days after a fear spike. Beyond 15 days, the edge decays as the signal becomes stale. The P/C ratio is a short-term indicator, not a trend predictor.
The total P/C ratio includes index options, which are dominated by institutional hedging. Institutions buy SPX puts as portfolio insurance regardless of sentiment. This creates noise. The equity-only ratio strips out this institutional hedging activity and captures retail positioning more purely. Academic studies (Pan & Poteshman, 2006) confirm that the equity-only ratio has greater predictive power for stock returns.
The Rydex Ratio measures the allocation of retail mutual fund investors between bearish and bullish funds:
Rydex Ratio = (Bear Fund Assets + Money Market Assets) / (Bull Fund Assets + Sector Fund Assets)
High ratio = bearish positioning = contrarian buy. Low ratio = bullish positioning = contrarian sell.
Academic testing of the Rydex Ratio as a standalone predictor yields disappointing results. The Pearson correlation between the Rydex Ratio and forward S&P 500 returns is -0.03, with an R-squared of 0.001. In plain English: the ratio explains 0.1% of forward return variance when used alone.
Multiple studies (including Hulbert Financial Digest analysis) found “little support” for the Rydex Ratio as a standalone timing tool. The reasons:
Include the Rydex Ratio as one data point in a composite sentiment dashboard, but never trade on it alone. It adds marginal value when it confirms signals from AAII, VIX, and Put/Call — but it should carry the lowest weight of any component in your composite. If you must simplify, drop it before dropping any of the other indicators in this article.
Citi’s proprietary Panic/Euphoria Model, maintained by Tobias Levkovich (and successors), combines several sub-indicators into a single reading with defined zones: Panic territory (below the lower threshold) and Euphoria territory (above the upper threshold). The components include short interest, margin debt, fund flows, options activity, retail investor surveys, and commodity speculation.
| Zone | Trigger | Avg Forward 12M Return | Hit Rate (Positive 12M) | Assessment |
|---|---|---|---|---|
| Panic Territory | Model drops below panic threshold | +21.6% | 93% | Highly reliable. Every major panic signal since 2002 led to strong gains. |
| Euphoria Territory | Model exceeds euphoria threshold | +9.0% | 58% | Unreliable. Positive on average but highly inconsistent. Markets frequently push higher through euphoria. |
The asymmetry is consistent with what we see in other sentiment indicators: fear signals are far more actionable than greed signals. Panic creates forced selling, margin calls, and capitulation — a mechanical process with a defined endpoint. Euphoria, by contrast, can sustain itself through leverage, momentum, and narrative for months or even years.
As of mid-February 2026, the Citi model sits in neutral territory, slightly above the midpoint. The April 2025 tariff shock triggered a panic reading, which proved correct (S&P rallied 18%+ in the following months). The current reading suggests neither extreme risk nor extreme opportunity from a sentiment perspective.
No single sentiment indicator is infallible. The power lies in confluence. When multiple independent indicators simultaneously flash the same extreme, the probability of a profitable contrarian trade increases dramatically.
| Component | Weight | Extreme Fear Threshold | Extreme Greed Threshold | Data Source | Frequency |
|---|---|---|---|---|---|
| AAII Bearish % | 25% | > 50% | < 20% | aaii.com | Weekly (Thursday) |
| CNN Fear & Greed | 25% | < 15 | > 85 | CNN Business | Real-time |
| VIX | 25% | > 35 | < 13 | CBOE | Real-time |
| Equity P/C Ratio | 15% | > 1.10 (5-day avg) | < 0.55 (5-day avg) | CBOE | Daily |
| Smart/Dumb Spread | 10% | Spread > +30 | Spread < -30 | SentimenTrader | Daily |
When 3 or more components flash extreme greed simultaneously, the data suggests reducing exposure by 15–25% (not going to zero). Greed signals have a lower hit rate than fear signals because bull markets can persist through euphoria for 6–18 months. The correct response to extreme greed is risk reduction, not outright selling. Raise cash, tighten stops, reduce position sizes — but do not short the market based on sentiment alone.
| Component | Current Level | Signal |
|---|---|---|
| AAII Bearish % | 35.2% | Neutral |
| CNN Fear & Greed | 32 (Fear) | Mild Fear |
| VIX | 18.5 | Normal |
| Equity P/C (5-day) | 0.89 | Neutral |
| Smart/Dumb Spread | +8 | Mild Positive |
Composite verdict: 0/5 extreme signals — no actionable sentiment extreme in either direction. The market is in a neutral sentiment regime where other factors (valuation, breadth, macro) should dominate decision-making.
You have completed Part 4 of the Timing the Market series.
Next: Part 5 — Finding Bottoms