Lyst Index — Brand Trajectory Dashboard

Backtest Results v3.1

Can Lyst rankings predict stock returns? · v3.1: 2018→Q1 2026, LVMH=LV only, Kering=Gucci only, lex-bug fix · USD excess returns vs GLUX

Method: For each quarter, compute rank changes for 10 listed fashion companies. Measure next-quarter stock return (USD) after Lyst publication date (~25 days post quarter-end). Subtract GLUX.PA (S&P Global Luxury Index) return as benchmark. v3 changes: LVMH proxied by Louis Vuitton only (was Loewe), Kering proxied by Gucci only (was 5 brands). Data extended to Q1 2018. 32 continuous quarters (Q1'18–Q4'25). Train/test: 2018-2021 vs 2022-2025.

S1: Rank Change

r = 0.103

ΔRank → excess return (n=300)

p = 0.073

S1: Prada Group

r = 0.427

Only significant stock-level signal

p = 0.018 ✓

S3b: Top-10 Exit

-5.91%

25 events — exit = negative signal

p = 0.058 (near sig)

S3: Top-10 Entry

+0.32%

27 events — no edge after bias fix

p = 0.870

S4: 3Q Momentum

+2.33%

67% hit rate but only 12 events

p = 0.653 · n = 12

S5: Cross-Sectional L/S

+236.1%

Cumulative, market-neutral, 30 qtrs

p = 0.021 ✓ · +4.62%/qtr

S1 Per-Company: Rank Change → Excess Return Correlation (vs GLUX, n=30 each)

S5: Cross-Sectional Long/Short — Cumulative Return (2018–2025)

Avg Quarterly

+4.62%

Cumulative

+236.1%

Win Rate

67%

Train / Test

+3.5% / +4.3%

p-value

0.021 ✓

Benchmark Cumulative Returns (Q1 2018 — Q1 2026, USD)

v3.1 Verdict & Actionable Signals

S5 Cross-sectional L/S — +19.7% cumulative over 30 quarters, NOT statistically significant (avg +1.23%/qtr, t=0.54, p=0.59). Test sample (2022+, 16q) is actually -20%, avg -0.67%/qtr. The earlier "+236%" v3 result was inflated by a train/test split bug — fixed in v3.1.
Prada Group (1913.HK) remains the only single stock with significant S1 correlation (r=0.427, p=0.018, n=30). Lyst rank changes do predict next-quarter excess return for this name.
S3b Top-10 Exit remains a useful negative signal: stocks falling out of top 10 underperform by ~-5.8% next quarter (p=0.062, n=21). Risk flag, not buy signal.
S3 Top-10 Entry — no edge (+1.0% vs GLUX, p=0.71, n=25). Entry alone is not a buy signal.
LVMH via Louis Vuitton — LV appeared Q1'21–Q3'24 only (15 quarters). Signal is sparse; LV unranked since Q4'24.
Kering via Gucci only — Gucci has full coverage but declining ranks (#1-2 in 2018 → unranked Q1'25 → #5 Q1'26 on Demna debut). Worth watching whether the Demna bounce holds.