quantmodel

← Results index

The HiLo Breadth Indexes for SP-500 | RecessionAlert — exploratory proxy campaign

2026-09-03T16:10:00Z Research completed SHA-pinned RecessionAlert HiLo page discloses daily 13w/52w net-highs% and HiLo=diff smoothed plus HiLo Fast, 5-level warnings, Diffusion 0-5 and -5 reversal, but smoothing constants, denominator timing, confirmation/hold mapping and asset SHA256 are undisclosed. Thirty labeled NYSE HIGN/LOWN proxy variants were tested T+1 on 2001-2026 (22y) vs native 6y; disposition fail-closed NO-GO for native fidelity with interesting negative result — realistic variants underperform SPY buy-hold.
1586 recessionalert research_campaign hilo breadth proxy fidelity_mismatch data_limited exploratory_tested_proxy no_go_native interesting_negative_result

Source report: research/experiments/1586-synthesis.json

Objective

Explore the distinct RecessionAlert HiLo Breadth inventory row from checked-in offline sources without live requests, credentials, private data, raw source republication, or SQLite mutation; publish a fail-closed specification with page+asset SHA256 provenance, disclosed semantics, native/effective spans, timing/cadence, revision limits and explicit no-go where insufficient, while testing best-effort inferred proxies on longest defensible history.

Source and provenance

Page https://recessionalert.com/the-hilo-breadth-indexes-for-sp-500/ (Research Papers, April 9, 2014) is bound to SHA256 sha256:571bfeaee9483e45f40b75e8d6c0da7c099bef9dd9e154b79ec36d0ac4856ad1 (fetched 2026-08-22T23:44:39Z, 91387 bytes, digest verified from data/recessionalert.sqlite:pages.body and data/cache/recessionalert/raw/571bfea...). Five GIF embeds SPHiLo_1.gif, SPHiLo_2.gif, HiLo_STL.gif, SPHiLo_3.gif, SPHiLo_43.gif are referenced via https://recessionalert.com/wp-content/uploads/2014/04/ but unavailable offline (linked_assets=0 in intake, confirmed 0 GIFs in data/recessionalert.sqlite:assets (6 assets total) and no cache files) — recorded as unavailable with provenance; no thresholds invented. Dashboard pointer CHARTS > BREADTH > HILO with hover tooltips and 6+ year slider starting 2008 per native span.

Approach and results

Disclosed construction: Net13w%=(New13wHighs-New13wLows)/issues*100 and Net52w% similarly; proprietary smoothing then HiLo=smooth(Net13w%)-smooth(Net52w%) (green/red area; Net52w black, Net13w orange), HiLo Fast shorter smoothing (purple dotted, more aggressive). 5-level warnings least→most bearish: Fast<HiLo, Fast<0, HiLo<0, Net13w<0, Net52w<0; Diffusion 5…0 (5 none, 0 all; 0 rare) buys on 0→1 (Great Trough) and 1→2, sell on fall to 3. Entry HiLo<-5 and starts reversing gives dip-buy. Spread Fast-HiLo is short-term liquidity forewarning.

Best-effort reconstruction (ENFORCED): Missing smoothing/denominator/lag do not stop. Proxies: NYSE INDEX:HIGN/INDEX:LOWN (2001-06-05–2026-08-11, 6331 days) as broad-market S&P500 breadth proxy with smoothing differentiation (EMA/SMA 10/20/50) — we trade precision for length. SPY price overlay 1993-01-29–2026-07-24 (8428 days) for benchmarking. Synthetic horizon differentiation via smoothing windows (both horizons from same HIGN/LOWN counts) disclosed as fidelity mismatch. Constituent rebuild from market.sqlite not feasible (only ETF-level ohlcv), fallback to broad proxy. FRED/OECD searched none; RecessionAlert workbook checked: 1 xlsx CMHI_DATA.xlsx not HiLo.

State space 30 trials: F1 best-guess EMA20/10, F2 SMA20/10, F3 SMA50/20, F0 raw, F5 diffusion, F6 -5 reversal; lags 0/1/2, thresholds 0/-4/-5/-6, holdings 20/60/120, T+1 (lag0 lookahead control). Common window 2001-2026 (25y). Realistic F1 lag1 CAGR 3.1% Sharpe 0.32 MaxDD -43% vs SPY 9.2% Sharpe 0.56 MaxDD -55% — all realistic variants underperform buy-hold (negative result retained). Lag0 lookahead 10.3% CAGR vs lag1 3.1% confirms causality. Regime: 2008-2009 F1 -7.7% vs SPY -10.6% (cushion), 2010-2019 F1 3.2% vs SPY 13.4% (underperform), 2020-2026 F1 8.1% vs SPY 15.1%. Diffusion 0 8.9% days (562) not extremely rare vs claim. Threshold -5: 0 days under NYSE 3000 denominator vs 224 days/140 reversals under SP500 500 denominator — denominator decisive, disclosed.

Disposition

exploratory_tested_proxy — Other/Research publication only. NO-GO for native HiLo fidelity or promotion to a playable Indicator because the source omits smoothing method/constants (SPHiLo_1.gif unavailable), precise issues-traded denominator/timing, confirmation/hold mapping, and asset SHA256. Best-effort inferred/proxy variants are labeled and tested on longest defensible history; negative results count. Charts and tables below are exploratory labeled inferred/proxy.

Artifacts

Source artifact: research/results/1586.md