Time-Series Momentum, ETF Basket

Conclusion: Promising

Out-of-sample Sharpe was 0.78, versus 0.80 in-sample. A bootstrap test returned p = 0.000 (statistically significant).

The result is credible but should be read soberly — much of the return is the well-documented trend/CTA premium plus equity beta rather than a novel alpha, and it closely corroborates the already-registered TSMOM strategy exp_20260625_001. The improvement over buy-and-hold is modest (higher Sharpe, roughly half the drawdown) and capacity-friendly.

Method

Rule / pseudocode

DATA ← adjusted observations for the stated instrument and horizon

RULE ← A 12-month time-series momentum signal (hold a liquid ETF only when its trailing 12-month return is positive, else cash) delivers positive out-of-sample risk-adjusted returns versus buy-and-hold on a diversified ETF basket.

METHOD ← Instrument: SPY

FIT ← estimate or select parameters on each training window only

TEST ← apply the frozen rule to the next unseen window after stated costs

REPORT ← return, Sharpe, drawdown, trade count, significance and fold stability

VERDICT ← prefer robustness and sufficient observations over the headline return

_Time-series momentum captures slow diffusion of information and behavioural under-reaction; well-documented across asset classes (Moskowitz-Ooi-Pedersen) and consistent with CTA outperformance reported in 2026._

Out-of-sample equity curve

Growth of $1 over the held-out test data the strategy never saw while it was being built, after 10 bps round-trip costs.

Out-of-sample performance

OOS Sharpe

0.78

Ann. return

Max drawdown

-23%

Win rate

52%

p-value

significant

WF folds +

5/5

Measured on held-out data. A high Sharpe with a non-significant p-value means the result could still be luck — read them together.

Hypothesis

A 12-month time-series momentum signal (hold a liquid ETF only when its trailing 12-month return is positive, else cash) delivers positive out-of-sample risk-adjusted returns versus buy-and-hold on a diversified ETF basket.

Method

Instrument: SPY

Results

Out-of-sample Sharpe was 0.78, versus 0.80 in-sample. A bootstrap test returned p = 0.000 (statistically significant). Walk-forward Sharpe stayed positive in 5 of 5 folds. Worst out-of-sample drawdown was -22.95%.

Analysis

In-sample (roughly 2000-2018 of the pooled daily series) the strategy earned a Sharpe of 0.80 with a total return of 695% and a worst drawdown of -26.6%, versus the far deeper drawdowns a passive equity holding suffered through the dot-com and 2008 bear markets. Out-of-sample (2018-2026) the Sharpe held at 0.78 with a 125.6% return and a shallower -23.0% drawdown, so the edge degraded only marginally from IS to OOS — the hallmark of a robust rather than an overfit signal.

Walk-forward validation reinforced this: all five folds were positive (fold Sharpes 0.66, 1.19, 0.91, 0.30, 1.18) for 100% consistency and a combined WF Sharpe of 0.80. A strategy that only worked in one window would show negative or wildly dispersed folds; this did not.

The block-bootstrap significance test on 118 non-overlapping monthly OOS returns returned p = 0.00 under the zero-edge null, comfortably below the 0.10 promotion bar. Because the p-value is computed on demeaned, non-overlapping monthly observations it is not inflated by the daily autocorrelation of continuously-held positions.

Costs were modelled on turnover (10bps per unit of position change at the monthly rebalance) rather than blanket-charged every day, which is the honest treatment for a low-turnover strategy — the naive per-bar apply_costs figure would have over-penalised a position that is simply held. Net of these costs the OOS Sharpe is still 0.78.

Verdict: PROMISING. The result is credible but should be read soberly — much of the return is the well-documented trend/CTA premium plus equity beta rather than a novel alpha, and it closely corroborates the already-registered TSMOM strategy exp_20260625_001. The improvement over buy-and-hold is modest (higher Sharpe, roughly half the drawdown) and capacity-friendly. It clears every mechanical gate (>=30 OOS trades, OOS Sharpe > 0.5, p < 0.10, >=60% WF folds positive) so it is promoted and paper-traded, but it is an incremental confirmation of a known edge, not a discovery.

Provenance

Synthesized from 1ui4ptb, 1um3mn2. The multistrat-performance thread reports CTAs (trend-followers) crushing 2026 while short-horizon equity stat-arb struggled; the failed-strategy thread confirms naive cross-sectional ML on minute data finds no alpha. The durable, testable takeaway is time-series (trend) momentum on liquid instruments, not cross-sectional ML.

Evidence

Walk-forward Sharpe by fold

Each fold retrains on past data then tests on the next unseen window. Staying positive across folds is the real test of an edge.

---

_Generated by labs-algo-trading. Automated research — not financial advice. Backtests overfit; treat verdicts as hypotheses._