Momentum Top 15: the October 2026 book, a rough start, and what the backtest says to expect

Correction, Oct 19, 2026: the backtest figures in this note were updated to a stricter test that only counts stocks after they joined the S&P 500. The original version used today's members for the whole history and quoted 45.1% a year over the full period (62.9% out-of-sample, max drawdown -27.7%). The stricter test gives 17.0% a year over the full period (26.6% out-of-sample, max drawdown -25.3%). Holdings and month-to-date figures are unchanged.

Momentum Top 15 rebalanced at the 2026-09-30 close into 15 stocks at 6.7% each. Month to date through 2026-10-16, the book is -16.7% against +0.1% for SPY. That gap is large but not unusual for a concentrated momentum portfolio: in the backtest the worst peak-to-trough fall was -25.3%, and the strategy compounded at 17.0% a year over 2016-10 to 2026-09 (SPY 15.2%).

Thesis

• Momentum (owning recent winners, skipping the latest month) is one of the most-studied return patterns in markets; this model applies it mechanically to the S&P 500.
• The current book is concentrated: Information Technology accounts for 13 of 15 names, so its month-to-month path will follow that group.
• The rules do not change because of a bad month. The next rebalance re-ranks everything on the last trading day of the month.

What the data says

Exhibit 1: Holdings, signal and month-to-date move (signal = 12-1 momentum as of 2026-09-30; MTD from the 2026-09-30 close to 2026-10-16)

Exhibit 2: Sector mix

Changes at this rebalance vs 2026-08-31: bought MRNA, TWLO; sold AMAT, LRCX.

Exhibit 3: Backtest, after 10 bps trading costs per side (backtest only counts stocks after they joined the S&P 500)

The worst single month in the backtest was 2026-07 at -24.4% (SPY +0.0%). Turnover averaged 34% of the book a month. In-sample (2016-2021) the strict version trailed SPY; nearly all of its edge came after 2022.

Risks and what would change our mind

• Survivorship bias. The backtest only counts stocks after they joined the S&P 500, but companies later acquired or delisted are still missing, which still flatters past returns somewhat.
• Momentum crashes. Sharp market reversals hit recent winners hardest; a concentrated sector bet amplifies that.
• What would change our mind: sustained underperformance well beyond the backtest's worst stretch, or evidence that the signal has been arbitraged away. A single bad month is neither.

Sources and data as of

Prices: Yahoo Finance closes through 2026-10-16. Holdings and signals: Coteau model run of 2026-09-30. Backtest: Coteau research, 2016-10 to 2026-09, 10 bps per side. Hypothetical results; not advice. Written from stored data without an AI model.