Claim check · NQ futures
Right-direction bias test on NQ futures: 9,048 trades per arm, 2025–2026
Summary
If the direction of every two-hour block were known in advance, 63.22% of 9,048 NQ futures trades won, against 49.38% for a coin-flip direction, from 2025-08-01 to 2026-07-31, before costs. Each trade used a 1:1 stop and target at 0.75 times the 30-minute ATR, entered every 10 minutes through the block. Tested by CE Market Solutions (cemarketsolutions.com), a backtesting service for CME futures strategies.
The claim tested
No single creator. This tests a common idea in futures trading content: that a correct directional bias is where the odds come from.
Once the direction is right, the entry hardly matters, and a trader’s bias is what tilts the odds.
Rules exactly as tested
- Blocks: three two-hour blocks per regular session on NQ, 09:30–11:30, 11:30–13:30 and 13:30–15:30 ET (120 one-minute bars each).
- Bias: the direction of the block’s own net move, D = close of the last bar minus open of the first bar. This uses hindsight on purpose: it is the most favourable case for a bias call, a ceiling.
- Entries: 12 per block, one every 10 minutes (minutes 0, 10, … 110), at the close of the 1-minute bar at that offset. One unit each.
- Barrier S: set once per block, from bars strictly before the block starts, as k times the Wilder ATR (14 periods) of 30-minute bars at the last completed 30-minute bar. Rounded to 0.25 points.
- Stop and target: both S points from the entry, so 1:1. 1R = S.
- Arms: right bias (trade in the direction of D), coin flip (random direction per trade, seed 42), and wrong bias (the opposite of D).
- Scoring: first barrier touched on 1-minute bars, scanning up to 360 bars past the entry. A bar that touches both counts as a loss (−1R). Trades touching neither are timeouts.
- The bias threshold T is the minimum size of D in points. T = 0 means every block has a bias from the sign of D alone. T of 25, 50, 100 and 200 points were also run.
Assumptions where the rules were ambiguous:
- The barrier uses ATR as a size input only. A second table uses a non-ATR measure, the standard deviation of 1-minute changes over the 120 bars before the block, scaled to 30 minutes, as a control.
- The test checks that the barrier cannot see the block: 756 of 756 ATR sources end at or before the block start, and 0 of 24 truncated rebuilds differed.
- Results with k of 1.5 and above have timeout rates over 10%; the test’s own rule is not to quote those as results, so they are listed for completeness.
Data
- Instrument
- NQ (E-mini Nasdaq-100) futures, 1-minute continuous data
- Date range
- 2025-08-01 to 2026-07-31, the most recent 12 full months in the file
- Sample size
- 754 blocks with a direction (756 eligible, 2 with D exactly 0 dropped; 9 incomplete half-day blocks excluded), 9,048 trades per arm
- Costs
- Gross, in R. No commission or slippage.
- In-sample or out-of-sample
- Not applicable to a ceiling test: the bias is hindsight by construction.
Results
| k | Trades per arm | Win rate, right bias | Win rate, coin flip | Average R, right bias (95% interval) | Average R, coin flip (95% interval) | Timeouts |
|---|---|---|---|---|---|---|
| 0.25 | 9,048 | 55.35% | 49.17% | +0.1070 (+0.0853 to +0.1284) | −0.0166 (−0.0378 to +0.0046) | 0.00% |
| 0.50 | 9,048 | 61.37% | 49.34% | +0.2275 (+0.2009 to +0.2541) | −0.0132 (−0.0337 to +0.0074) | 0.01% |
| 0.75 | 9,048 | 63.22% | 49.38% | +0.2734 (+0.2404 to +0.3057) | −0.0010 (−0.0217 to +0.0199) | 1.04% |
| 1.00 | 9,048 | 62.89% | 47.93% | +0.3016 (+0.2625 to +0.3399) | +0.0040 (−0.0159 to +0.0242) | 4.50% |
| 1.50 | 9,048 | 56.86% | 43.23% | +0.2951 (+0.2519 to +0.3386) | +0.0245 (+0.0048 to +0.0436) | 15.86% |
| 2.00 | 9,048 | 48.23% | 34.74% | +0.2837 (+0.2410 to +0.3267) | −0.0016 (−0.0186 to +0.0160) | 30.33% |
| 3.00 | 9,048 | 32.01% | 22.30% | +0.2386 (+0.1984 to +0.2784) | +0.0033 (−0.0120 to +0.0183) | 55.63% |
| k | Trades per arm | Win rate, right bias | Win rate, coin flip | Average R, right bias (95% interval) | Average R, coin flip (95% interval) | Timeouts |
|---|---|---|---|---|---|---|
| 0.25 | 9,048 | 53.02% | 46.85% | +0.0603 (+0.0387 to +0.0822) | −0.0630 (−0.0851 to −0.0409) | 0.00% |
| 0.50 | 9,048 | 59.58% | 50.25% | +0.1921 (+0.1673 to +0.2175) | +0.0056 (−0.0161 to +0.0270) | 0.07% |
| 0.75 | 9,048 | 62.60% | 49.41% | +0.2603 (+0.2292 to +0.2908) | −0.0025 (−0.0217 to +0.0170) | 0.93% |
| 1.00 | 9,048 | 61.23% | 48.13% | +0.2651 (+0.2289 to +0.3007) | +0.0032 (−0.0169 to +0.0236) | 4.07% |
| 1.50 | 9,048 | 57.73% | 43.55% | +0.2950 (+0.2527 to +0.3364) | +0.0067 (−0.0138 to +0.0273) | 13.76% |
| 2.00 | 9,048 | 51.32% | 37.10% | +0.2901 (+0.2476 to +0.3325) | −0.0034 (−0.0223 to +0.0156) | 25.38% |
| 3.00 | 9,048 | 38.35% | 27.35% | +0.2524 (+0.2126 to +0.2925) | +0.0031 (−0.0119 to +0.0183) | 45.60% |
- Total R and maximum drawdown are not reported in the source file for this test.
- The ATR and dispersion measures agreed from k of 0.75 upward and differed at k of 0.25, where ties occur.
Video transcript
Easiest way to a sixty percent win rate? The right bias. I tested twelve trades every two hours for a year. The only input is the correct direction. Sixty three point two percent won, versus no bias at forty nine point four percent. That’s just probability. So zoom out. Instead of a hundred and twenty one minute candles, you only call two hourly ones. Comment test, and I’ll backtest your bias free and show you what’s holding it back.
Questions and answers
What win rate does a perfect directional bias give on NQ?
With the correct direction of each two-hour block known, 63.22% of 9,048 trades won at a 1:1 barrier of 0.75 times the 30-minute ATR, against 49.38% with a coin-flip direction, from 2025-08-01 to 2026-07-31, before costs.
How was the right-bias test run?
Three two-hour blocks a day, 12 entries per block every 10 minutes, each with a stop and target the same distance away. The direction was the block’s own net move, so it is a ceiling, not a forecast.
What sample size was used?
754 blocks and 9,048 trades per arm.
Does the barrier size change the right-bias win rate?
Across barriers from 0.25 to 3.0 times ATR the right-bias win rate ran from 32.01% to 63.22% and the coin-flip rate from 22.30% to 49.38%; wider barriers also leave more trades unresolved.
Related reports
- Asia range sweep reversal backtest on NQ futures: 3,622 trades, 2016–2026
- ICT Silver Bullet backtest on NQ futures: 426 trades, 2016–2026
- ICT Power of 3 test on NQ futures daily candles: 2,419 sessions, 2016–2026
Read more: R multiples and expectancy, explained · How many trades does a backtest need?
Want your own strategy tested the same way? Every version is $19 at cemarketsolutions.com.
More tests like this on Instagram and TikTok: @cemarketsolutions.