MNQ Intraday Noise Band Breakout — Backtest
Strategy 1103 · MNQ futures, long and short, 2x leverage · checks every 30 minutes, flat by 16:00 ET · real prices Nov 27, 2024 – Nov 20, 2026 · experimental
Experimental. This result is not statistically significant: the t-statistic of the daily results is 1.6, where 2 or more is the usual standard. Two years of prices is too short to tell this strategy from luck, and the same rules lost money on S&P 500 futures. It is in paper testing and has not traded live.
Growth of $10,000
Both lines start at $10,000. The strategy line is at 2x leverage after trading costs, with no interest on idle cash. Buy and hold is MNQ futures held with no leverage, measured from each day's 16:00 ET price to the next.
Rules
- Usual move: for each time of day, the average over the last 14 full sessions of how far MNQ was from that day's 09:30 ET open, as a percentage.
- Noise band: upper = the higher of today's open and yesterday's 16:00 close, plus the usual move. Lower = the lower of the two, minus the usual move.
- Checks: every 30 minutes from 10:00 to 15:30 ET, on the close of the 5-minute bar ending then.
- Enter (when flat): buy one contract above the upper band; sell one short below the lower band.
- Exit a long: the price is below the higher of the upper band and today's VWAP. Exit a short: the price is above the lower of the lower band and today's VWAP.
- Reversal: a position closed at a check can be opened the other way at the same check.
- Close: anything still open is closed at 16:00 ET. Nothing is held overnight.
- Days: full sessions only; nothing on holidays or early-close days.
- Instrument: the front Micro E-mini Nasdaq-100 futures contract (MNQ, $2 per index point). The front contract changes on the Thursday of the week before the quarterly expiry.
- Size: one contract (about $50,285 of exposure) at 2x leverage, so about $25,142 of capital.
- Source: the rules are those of Zarattini, Aziz and Barbon (2024), “Beat the Market: An Effective Intraday Momentum Strategy for S&P500 ETF (SPY)”, applied to MNQ without tuning.
Real MNQ prices, Nov 27, 2024 – Nov 20, 2026
| Return per year | Sharpe | Max drawdown | Worst day | Total return | |
|---|---|---|---|---|---|
| 1x | 11.6% | 1.15 | −8.3% | −1.5% | +24% |
| 2x (as traded) | 23.4% | 1.15 | −16.3% | −3.1% | +51% |
| Buy and hold MNQ (1x) | 0.3% | 0.13 | −25.9% | −6.5% | +1% |
These two years are all the expired-contract price history Interactive Brokers provides: 492 full sessions. The strategy traded on 280 of them (57%), about 1.4 times on those days.
Per contract, in dollars
| Trades | Win rate | Average trade | Net P&L | |
|---|---|---|---|---|
| All trades | 396 | 40% | +$25 | +$9,920 |
| Long trades | 197 | 44% | +$17 | +$3,436 |
| Short trades | 199 | 36% | +$33 | +$6,484 |
| All trades, before costs | 396 | 40% | +$27 | +$10,807 |
| Per contract | |
|---|---|
| Best trade | +$3,626 |
| Worst trade | −$647 |
| Best day | +$3,021 |
| Worst day | −$872 |
| Largest drawdown | −$4,028 |
| Buy and hold one contract | +$1,535 |
Results by year
| Year | Trades | Win rate | Net P&L per contract | Strategy (2x) | Buy and hold MNQ |
|---|---|---|---|---|---|
| 2024 (from Nov 27) | 20 | 45% | +$1,183 | +5.3% | −2.3% |
| 2025 | 188 | 40% | +$3,833 | +21.8% | +10.7% |
| 2026 (to Nov 20) | 188 | 39% | +$4,903 | +17.6% | −7.0% |
How concentrated the profit is
- The five best days made $9,502 of the $9,920 total. Without the ten best days the strategy's result is −$3,965.
- 40% of trades win. The strategy makes many small losses and depends on a few large trend days, which is normal for a breakout strategy but means a short test can mislead.
- 18 of 25 calendar months were profitable.
Checks
| Net P&L per contract | Trades | Win rate | Sharpe | t-statistic | |
|---|---|---|---|---|---|
| MNQ (this strategy), Nov 27, 2024 – Nov 20, 2026 | +$9,920 | 396 | 40% | 1.15 | 1.60 |
| Same rules on MES (S&P 500), Nov 27, 2024 – Nov 12, 2026 | −$1,148 | 428 | 34% | −0.17 | −0.23 |
- Other settings: of 48 combinations of lookback (14, 30 days), band width (0.75x to 1.5x), check interval (15, 30, 60 minutes) and stop (VWAP, band only), 46 were profitable on MNQ (Sharpe ratios from −0.35 to 1.07, median 0.52) and 10 on MES. The settings traded are the paper's, not the best of these.
- A different market from the paper: the published results are for SPY from 2007 to early 2024. On S&P 500 futures over the last two years the rules did not make money; only the Nasdaq version did.
- An earlier idea that failed: a simpler intraday momentum rule (trade the last half hour in the direction of the day's move) was tested first on the same prices and lost money on both MES and MNQ.
Assumptions
- Costs are $0.62 in commission and exchange fees plus one tick ($0.50) of slippage per contract per side: $2.24 a round trip.
- Orders fill at the open of the 5-minute bar after each check. Live, a market order goes in a few seconds after the check and can fill at a worse price in a fast market.
- VWAP is estimated from the 5-minute bars (each bar's average of high, low and close, weighted by its volume).
- Yesterday's close and today's prices always come from the same contract, so a contract roll never looks like a gap.
- The percentage returns model the position as exactly 2x of capital each day. Live, it is one whole contract, so the leverage drifts with the index level and the account's value. Idle cash earns nothing in this test.
- Prices are Interactive Brokers' 5-minute trade bars for each quarterly contract while it was the front one.
Risks
- Not statistically proven. See the note at the top. The live result could easily be zero or negative.
- A few days make the result. Missing or mishandling a handful of large trend days would remove the profit.
- Leverage doubles losses. The worst day cost 3.1% of capital at 2x and the largest drawdown was 16.3%.
- Fast markets. The strategy trades when the market is moving most, which is when market orders slip most.
- Process risk. The position is closed by a program at 16:00 ET. If it fails, the position stays open until it is closed by hand.
- Short history. Only two years of actual MNQ prices were available.
Generated Nov 21, 2026 by ibk/backtests/noise_band_backtest_report.py. Backtests use past prices and simple cost assumptions; past results don't guarantee future returns.