Pullback Sniper on Micro Nasdaq-100 (MNQ)

What the numbers say

• It meets the brief. The target was a profit factor above 1.0 and at least 3 : 1 net profit to max drawdown. In NinjaTrader 8’s Strategy Analyzer, with the “NinjaTrader Brokerage Free” commission template, the recommended set delivers PF 1.21 and 6.9 : 1 — more than double the ratio asked for.
• Steady, not spiky. Profitable in each calendar year of the test, 22 of 26 months positive, and the equity curve climbs without a single outsized trade carrying it: the largest single winner is $78 and the largest loss is capped at −$122.
• Short exposure. Entries only between 09:30 and 12:00 ET, average 29 minutes in the market, every trade closed the same session — no overnight risk, nothing carried across a contract roll.
• Verified twice. Every figure above comes from NT8’s own engine; the 35,700-run parameter sweep behind the choice was made in a replica that matches NT8 to within one trade.
• Built to survive the search. Most parameter sets that look best in a sweep fall apart afterwards — the walk-forward check shows exactly that happening here. This set was chosen because it kept working after each of four earlier cut-off dates, with forward PF 1.18–1.25 every time. Plan on PF 1.1–1.2 and net/DD 3–4 live rather than the full-period headline.

Set it up in NinjaTrader

Add ToredaPullbackStrategy to an MNQ 3-minute chart (or the Strategy Analyzer) and change only these inputs — everything else stays at the strategy default:

1 contract vs 3 contracts

One position at a time, so size scales cleanly: the ratios stay the same and every dollar figure is three times larger.

Year by year

How robust is it?

At the default settings

A pullback with a 40/80-tick bracket fires on a large share of all bars, and at a 32% win rate the 1 : 2 bracket does not cover it. This is the baseline the sweep has to beat, and the reason the recommended set looks nothing like the defaults.

Know before you trade it

• Selection bias — measured, not assumed. 35,700 configurations were searched, and the walk-forward check shows the in-sample winner failing forward at every cut-off. The recommended set was chosen for its recurrence across cut-offs and its plateau, but the final choice between it and its neighbours was made with the whole period in view, so even its 2026 “out-of-sample” figures carry some selection. The forward segments in the walk-forward check (PF 1.18–1.25) are the numbers to plan on.
• Two years, one regime. The sample covers Aug 2024 → Sep 2026, a period in which NQ trended up; longs contributed three-quarters of the net. The filter is symmetric, so a sustained downtrend should flip the balance rather than break the strategy, but that is untested.
• Instrument-specific. The bracket is in ticks, so the set does not transfer: on the two months of MES history available the same inputs lose (PF 0.79, 53 trades) because 80 MES points is 1.2% of price versus 0.35% for MNQ.
• Fill assumptions. Both engines fill stops at the stop price without slippage and assume the target limit fills when touched; live fills will be a few ticks worse on average. The 2-tick slippage row is the realistic floor.
• Contract rolls. The replica rolls at the first bar of each new contract’s NT8 history with back-adjustment; NT8 uses its own roll dates. The two agreed to within one trade, so the effect is negligible, but roll-week trades are the least reliable in either engine.

Getting started

1. Deploy the recommended set on MNQ (3-minute) in simulation first, sized for a $120 stop per contract, and compare the first 60–80 live trades against the forward statistics (PF ≈ 1.2, ~65% winners, ~1.8 trades/day, ≈ $340/month per contract). A PF below 1.0 after 80 trades or a drawdown beyond $1,800 per contract is outside anything seen in the test and is the signal to stop.
2. Do not add breakeven, trailing or a per-day cap to this set; all three were tested and hurt.
3. Do not chase the leaderboard. Re-optimising to whatever tops the in-sample ranking is exactly the procedure the walk-forward check shows losing money; if the set is re-tuned, re-tune within the 3-minute plateau (stop 240–320, target 120–160, EMA 20–50) and only on evidence from a forward segment.