NAAIM–AAII relative-sentiment benchmark
A frozen, point-in-time test of whether professional equity exposure relative to individual-investor stock allocation can scale exposure between SPY and BIL. Of four pre-specified variants, only the continuous V1 rule passed the publication gate.
Full-window result
| Portfolio | CAGR | Volatility | Sharpe | Max drawdown | Exposure |
|---|---|---|---|---|---|
| V1 relative sentiment | 8.92% | 10.70% | 0.72 | -23.00% | 63.4% |
| 10-month SMA | 7.71% | 12.95% | 0.53 | -24.72% | 84.2% |
| SPY buy and hold | 13.82% | 17.15% | 0.76 | -33.72% | 100.0% |
Rounded derived statistics. V1 sacrificed return versus SPY while reducing volatility, exposure, and maximum drawdown.
Frozen V1 rule
Smooth NAAIM exposure over 26 weekly observations and AAII stock allocation over two
monthly observations, then subtract AAII from NAAIM. After a 36-month minimum history,
convert the expanding z-score to a risky weight of clip(0.5 + 0.25z, 0, 1).
The remainder holds BIL. Survey observations are usable only after their conservative
publication date; trades occur at the next close.
Pre-specified robustness gate
| Evaluation | V1 CAGR | SMA10 CAGR | V1 max drawdown | SPY max drawdown |
|---|---|---|---|---|
| Through 2016 | 7.16% | 6.32% | -13.37% | -18.61% |
| After 2016 | 10.18% | 8.71% | -23.00% | -33.72% |
| Extra-day lag, through 2016 | 7.12% | 6.27% | -13.43% | -18.61% |
| Extra-day lag, after 2016 | 10.14% | 8.71% | -23.13% | -33.72% |
V1 met the locked requirement in both subperiods and after delaying both survey sources one additional trading session. V0, V2, and V3 were rejected rather than tuned after inspection.
Important limitations
- This is a research benchmark, not investment advice or a live allocation signal.
- Results are hypothetical, before taxes, and depend on adjusted SPY/BIL data and assumed transaction costs.
- Historical survey publication timestamps are unavailable; the test uses conservative release floors plus an extra-day sensitivity.
- The AAII workbook required 22 narrowly specified date-label repairs. Raw AAII, NAAIM, and price observations are not redistributed.
- Only derived results are published; independent regeneration downloads and checksum-validates the source files at runtime.
Reproducibility manifest
This snapshot was retrieved at 2026-07-30T11:02:29.433020+00:00.
The base derived-return SHA-256 is
12aadb16852db076e572383da3e307ee3f5acab9b9b7ab06abb249810128fe8e;
the extra-day-lag derived-return SHA-256 is
273f12c67114f967aebec0b9f55e3c4e9ef483f8b5212241a6c1bbf0fef5f756.
The checked-in frozen benchmark and robustness JSON files have SHA-256
2b6d6905dc527ef42ce4a902461269d8a75a72569f6c90a4d1cfeb72e7cfbb55
and cd7a1bc09616cc2a5753450dc5517f26a90b381d1e082f5f7500eab8e7798dd4, respectively.
- AAII: July 2006–June 2026 after 22 constrained date-label repairs; source SHA-256
3a32f8f52ae3d644820c5bd73437812c79c510d05d5d6df8da1e6706ed432eee. - NAAIM: 2006-07-05–2026-07-22; source SHA-256
3256761fb758851e1a809eb2a8649b73a2c38d020258d21cfc3e0abd65f0930a. - Adjusted SPY coverage is 2006-01-03–2026-07-29 and adjusted BIL coverage is 2007-05-30–2026-07-29; the common evaluated base window is 2010-04-08–2026-07-29.
- AAII month M is available on the first U.S. trading day of M+1; NAAIM is available on the next U.S. trading day; observations must be known by the prior close and trades occur at the next close. The sensitivity delays both sources one further trading day and begins 2010-04-09.
- The manifest and download contain derived statistics and provenance only. Raw survey workbooks and adjusted price observations are used at runtime but are not committed or redistributed.
Download the derived robustness artifact (JSON) · Return to the catalog