Our model · NQ futures
CE 50% Model backtest on NQ futures: 2,625 trades, 2016–2026
Summary
The CE 50% Model, a pre-open range model on NQ futures, took 2,625 trades from 2016-06-01 to 2026-08-27 with a 34.4% win rate and +0.141R average after $2.20 round-trip commission, with a net maximum drawdown of 34.2R. The rules were chosen on this same data; out-of-sample runs on ES, gold and earlier NQ gave +0.052R, −0.004R and +0.076R, each with an interval that included zero. Tested by CE Market Solutions (cemarketsolutions.com), a backtesting service for CME futures strategies.
The claim tested
CE Market Solutions. This is our own model, published with its rules, so there is no outside claim being checked.
The 8:30–9:30 ET range split at its midpoint, with the side picked by the first 3-minute close after the open, stopped at the other side of the range and trailed under 15-minute swings.
Source: https://cemarketsolutions.com/ce50
Rules exactly as tested
- Range: the high (RH) and low (RL) of the 08:30–09:30 ET bars (twelve 5-minute bars). The midpoint is MID.
- Signal: the close of the first 3-minute candle from 09:30 ET. A close above MID means long; a close below MID means short.
- Entry: the next 1-minute open, so 09:33 ET for the 3-minute candle.
- Initial stop: RL for a long, RH for a short.
- Exit: the stop trails to each confirmed 15-minute three-bar fractal swing, never backwards, until the stop is hit or 16:00 ET. There is no profit target.
- Signal timeframes tested: 1, 3, 5, 15 and 30 minutes. A second exit that holds to the stop or 16:00 ET with no trail was also run.
- Both directions are traded.
Assumptions where the rules were ambiguous:
- A trade stopped on its own entry bar is counted as a loss and kept: 49 of 2,625 (1.9%), all −1.00R.
- A gap that opens beyond the stop fills at the open.
- The 3-minute candle and the 15-minute trail were chosen from ten combinations (five timeframes, two exits) on the same data they are reported on.
Data
- Instrument
- NQ (E-mini Nasdaq-100) futures, 1-minute data
- Date range
- 2016-06-01 to 2026-08-27 (the 3-minute model); out-of-sample tests below use other dates and markets
- Sample size
- 2,625 trades for the 3-minute model
- Costs
- The signal timeframe table is gross: no cost is subtracted. The net figures subtract a $2.20 round-turn commission per NQ contract (0.11 point), with no slippage. The compounding run subtracts $2.20 round turn per NQ contract, or $0.60 per MNQ contract, times the contracts in each trade, fills at the recorded entry and exit prices, with no slippage or size impact.
- In-sample or out-of-sample
- In-sample for NQ 2016–2026: the rule, the 15-minute trail and the 3-minute candle were all chosen on this data. Out-of-sample runs on ES, gold and NQ before 2016, and a live week, are listed separately.
Results
| Signal candle | Trades | Win rate | Average R (95% interval) | Points per trade | Median stop (points) | Max drawdown |
|---|---|---|---|---|---|---|
| 1 minute (entry 09:31) | 2,630 | 31.6% | +0.152 (+0.073 to +0.236) | +4.57 | 31 | not reported |
| 3 minutes (entry 09:33) | 2,625 | 34.4% | +0.147 (+0.074 to +0.219) | +5.58 | 36 | 33.8R gross |
| 5 minutes (entry 09:35) | 2,628 | 35.5% | +0.119 (+0.053 to +0.185) | +4.02 | 39 | not reported |
| 15 minutes (entry 09:45) | 2,634 | not reported | +0.036 (−0.024 to +0.096) | not reported | not reported | not reported |
| 30 minutes (entry 10:00) | 2,634 | not reported | +0.051 (−0.002 to +0.102) | not reported | not reported | not reported |
| Cost model | Trades | Average R (95% interval) | Points per trade | Max drawdown | Longest losing streak |
|---|---|---|---|---|---|
| NQ, $2.20 round turn (0.11 point) | 2,625 | +0.141 (+0.068 to +0.214) | +5.47 | 34.2R | 13 |
| MNQ, $0.60 round turn | 2,625 | +0.132 (+0.065 to +0.209) | not reported | not reported | 13 |
| Year | Trades | Win rate | Average R (95% interval) | Total R | Points per trade | Median stop (points) |
|---|---|---|---|---|---|---|
| 2016 (June on) | 150 | 37% | +0.272 (−0.051 to +0.634) | +40.8 | +3.2 | 8 |
| 2017 | 255 | 34% | +0.223 (−0.017 to +0.484) | +56.8 | +1.8 | 8 |
| 2018 | 256 | 32% | +0.311 (+0.016 to +0.630) | +79.7 | +4.4 | 18 |
| 2019 | 254 | 34% | +0.097 (−0.110 to +0.335) | +24.6 | +3.4 | 18 |
| 2020 | 254 | 32% | +0.040 (−0.190 to +0.295) | +10.1 | +1.9 | 37 |
| 2021 | 258 | 37% | +0.155 (−0.045 to +0.352) | +40.0 | +8.8 | 40 |
| 2022 | 258 | 31% | +0.076 (−0.147 to +0.338) | +19.5 | −1.6 | 62 |
| 2023 | 257 | 35% | +0.067 (−0.117 to +0.265) | +17.2 | +4.6 | 46 |
| 2024 | 258 | 36% | +0.163 (−0.040 to +0.377) | +42.1 | +17.9 | 55 |
| 2025 | 256 | 37% | +0.158 (−0.044 to +0.363) | +40.4 | +10.2 | 64 |
| 2026 (to 08-27) | 169 | 33% | +0.084 (−0.190 to +0.400) | +14.1 | +5.8 | 92 |
| Test | Trades | Average R (95% interval) | First half and second half |
|---|---|---|---|
| ES, 2016-06 to 2026-08 | 2,594 | +0.052 (−0.020 to +0.120) | +0.124 and −0.006 |
| Gold (GC), 2016-09 to 2026-09 | 2,551 | −0.004 (−0.055 to +0.045) | −0.078 and +0.050 |
| NQ, 2010-06 to 2016-05 | 926 | +0.076 (−0.069 to +0.221) | −0.193 (221 trades) and +0.160 (−0.008 to +0.339; 705 trades) |
| Measure | Value |
|---|---|
| Ending equity | $5,902,867 (118.06 times the start) |
| Compound annual growth rate | 59.37% |
| Maximum drawdown | 45.03% (trough 2019-06-06) |
| Lowest equity | $45,386 on 2016-06-24 |
| Longest drawdown | 444 days, 2020-10-05 to 2021-12-23 (depth 43.5%) |
| Lowest and highest calendar year | 2020 (+3.2%) and 2018 (+187.4%) |
| Total commission paid | $159,838 |
| Day | Side | R gross | R net of $0.60 | Points gross |
|---|---|---|---|---|
| Mon 10-12 | Short | +0.2456 | +0.2433 | +31.50 |
| Tue 10-13 | Short | −1.0000 | −1.0033 | −92.00 |
| Wed 10-14 | Short | +0.7892 | +0.7882 | +244.25 |
| Thu 10-15 | Short | −1.0000 | −1.0024 | −126.00 |
| Fri 10-16 | Short | +5.5218 | +5.5192 | +632.25 |
| Week | 5 trades | +4.5566 | +4.5451 | +690.00 |
- Forward log of the frozen model, 2026-08-28 to 2026-10-07: 29 trades, 44.8% win rate, +0.548R average net (−0.238 to +1.585), +15.9R cumulative, 4.0R maximum drawdown. The 2026-09-21 long was +11.94R, 75% of the total.
- With one fixed NQ contract, 2022 was −446 points net. R and points disagree in a year when small-stop days win and large-stop days lose.
- In the live week, Friday carried the whole week: without it the four earlier trades were −0.9652R gross.
- Figures shown in the video version of this test: 34.4% win rate, 2,625 trades, +370.9R net cumulative, and 1,723 of 2,625 trades at or below 0R net (65.6%).
Video transcript
This is my own NQ model, and I’m giving it to you for free. Mark the high and low from eight thirty to nine thirty, and split it at fifty percent. If the first three minute candle after the nine thirty open closes above it, buy the next open. Below, sell it. Stop goes on the other side of the range, then trails down to every swing high made on the fifteen minute until four EST. This day it held to the close for fifteen R. Over ten years, most trades lost small, and the trend days paid for them. From twenty sixteen to twenty twenty six, with commissions included, it returned over fifty nine percent a year, compounded. Comment CE fifty and I’ll send it to you.
Questions and answers
How was the CE 50% Model backtested?
On NQ futures 1-minute data from 2016-06-01 to 2026-08-27. The 08:30–09:30 ET range is split at its midpoint, the first 3-minute close after 09:30 picks long or short, the stop is the other side of the range, and the stop trails under confirmed 15-minute swings until 16:00 ET.
What sample size was used for the CE 50% Model?
2,625 trades for the 3-minute model over 2016-06-01 to 2026-08-27, with 34.4% winning. Out-of-sample runs used 2,594 trades on ES, 2,551 on gold and 926 on NQ from 2010 to 2016.
Is the CE 50% Model result in-sample or out-of-sample?
The NQ 2016–2026 result is in-sample, because the rule, the trail and the 3-minute candle were chosen on that data. The out-of-sample averages were +0.052R on ES, −0.004R on gold and +0.076R on NQ 2010–2016, each with an interval that included zero.
What did the CE 50% Model return after costs?
After $2.20 round-turn NQ commission the 3-minute model averaged +0.141R per trade. Compounded at quarter Kelly with commissions, $50,000 became $5,902,867 with a 45.03% maximum drawdown, using sizing fitted on the same data.
Related reports
- Opening range breakout backtest on NQ futures: 324 variations, 1,050,582 trades, 2016–2026
- Asia range sweep reversal backtest on NQ futures: 3,622 trades, 2016–2026
- Kasen Kennedy 8:00 ET opening range retest backtest on NQ futures: 500 trades, 2020–2026
- ICT opening range gap 50% level test on Nasdaq futures: 732 sessions, 2022–2025
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.