QQQ Opening-Hour Momentum

Conclusion: Rejected

Out-of-sample Sharpe was -2.38, versus 0.22 in-sample. A bootstrap test returned p = 0.996 (not significant).

Trade count is adequate (219 OOS trades), so this is a real rejection, but the short hourly history and fold-to-fold sign flips mean the honest read is 'no exploitable opening-range continuation on QQQ after costs.' Opening-range momentum, if it ever existed intraday, is not present in this instrument/timeframe once realistic costs are charged. No promotion, no paper trade.

Method

Rule / pseudocode

DATA ← adjusted observations for the stated instrument and horizon

RULE ← On QQQ hourly bars, the sign of the first regular-session hour's return predicts the sign of the return over the remainder of the session, giving a positive after-cost intraday continuation edge.

METHOD ← Instrument: QQQ

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

_Opening-range momentum: early-session direction reflects overnight information being priced in plus institutional order flow that persists through the day._

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

-2.38

Ann. return

-30%

Max drawdown

-28%

Win rate

43%

p-value

0.996

not significant

WF folds +

2/5

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

On QQQ hourly bars, the sign of the first regular-session hour's return predicts the sign of the return over the remainder of the session, giving a positive after-cost intraday continuation edge.

Method

Instrument: QQQ

Results

Out-of-sample Sharpe was -2.38, versus 0.22 in-sample. A bootstrap test returned p = 0.996 (not significant). Walk-forward Sharpe stayed positive in 2 of 5 folds. Worst out-of-sample drawdown was -27.69%.

Analysis

In-sample the rule looks like nothing: over 508 trading days the after-cost IS Sharpe is only 0.218 (total return +6.3%, win rate 51.2%, drawdown -13.6%). That is a coin-flip-plus-a-whisker, already below the 0.5 promotion threshold before any out-of-sample test — a weak prior that opening-hour direction carries through the session.

Out of sample the weak prior inverts hard: across 219 independent daily trades the after-cost OOS Sharpe is -2.38 with a -26.5% total return, a -27.7% drawdown and a 42.9% win rate. First-hour direction did not continue — the remainder of the session moved against it, so a continuation bet was consistently wrong over the most recent ~30% of the sample.

Walk-forward exposes extreme instability rather than a decayed-but-real edge: fold Sharpes are -0.84, +2.26, +0.80, -3.76, -0.85, an oos_consistency of 40% (2 of 5 folds positive) and a wildly negative worst fold. The single strong fold (+2.26) is swamped, and the sign of the edge flips fold to fold — the hallmark of noise, not structure.

The block-bootstrap significance test gives p = 0.996 on the OOS per-trade returns: the observed strongly-negative Sharpe sits deep in the left tail of the zero-edge null, i.e. there is no evidence of a positive continuation edge and the negative OOS result is itself within noise given the small (~2-year) hourly sample yfinance allows.

Verdict: REJECTED. Trade count is adequate (219 OOS trades), so this is a real rejection, but the short hourly history and fold-to-fold sign flips mean the honest read is 'no exploitable opening-range continuation on QQQ after costs.' Opening-range momentum, if it ever existed intraday, is not present in this instrument/timeframe once realistic costs are charged. No promotion, no paper trade.

Source

Originated from discussion on r/manual.

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._