NQ Alpha Engine — Systematic MNQ Futures Strategy

Year	Mar	Apr	May	Jun	YTD
2026	+36.6%	+25.6%	+1.0%	+18.1%	+104.7%

Methodology

The Framework

NQ Alpha Engine is built around a core idea: certain price levels on the Nasdaq 100 have a measurable tendency to act as turning points — either support zones where price bounces, or resistance zones where price reverses. The system maps these structural levels daily, then uses an XGBoost machine learning model to score each one based on current market context: volatility regime, trend alignment, momentum, and session conditions. Only levels that clear a dynamic probability threshold generate a trade signal — long or short depending on where price is and what the model expects.

Before any signal reaches execution, it passes through a set of independent regime filters. If the market is in a strong directional trend, a high-volatility expansion, or an otherwise unfavorable environment for mean-reversion entries, the system stands aside entirely. This is intentional — the edge is specific to certain conditions, and forcing trades outside of them destroys it. The model is retrained on a rolling basis so it adapts to shifting market regimes rather than overfitting to historical data.

Risk management is hardcoded into the architecture. Every trade carries a fixed stop loss and a predefined profit target. There is no averaging into losing positions, no manual overrides, and no discretionary exceptions. Once a level has been traded in a session, it is locked out — preventing the system from repeatedly re-entering the same failed setup. Every decision is made by the model, consistently, on every trade.

Common Questions

FAQ

NQ Alpha Engine trades the Micro E-mini Nasdaq 100 futures contract (MNQ) on CME. Each MNQ contract controls $2 per index point. At current NQ levels (~29,000), one contract represents approximately $58,000 in notional value. The strategy trades one contract at a time.

Signals are generated by a machine learning model (XGBoost ensemble) that scores historical support levels on the MNQ H1 chart. The model evaluates 37 features including price structure, volatility regime, higher-timeframe confluence, RSI context, and session dynamics. Only levels scoring above a dynamic threshold trigger an order. No human makes any trade decisions.

Every trade carries a fixed 200-point stop loss, equal to $400 per MNQ contract. This is non-negotiable — the system places a hard stop on every order and it is never moved or overridden. There is no fixed take profit target. Exit timing is determined entirely by the model based on real-time market conditions — the system closes the position when its internal criteria are met. This means winning trades vary in size rather than being capped at a fixed R multiple, allowing the strategy to capture more on strong moves and exit efficiently when momentum fades.

Collective2 suggests a minimum of $25,000, and that is a reasonable target for comfortable operation. Realistically, $10,000 is workable — it covers margin requirements with buffer for normal drawdowns. Going below $5,000 becomes problematic: most brokers require higher overnight margin to hold MNQ positions through the close, and a thin account can get flagged or force-closed before the trade has time to work. Starting with less also amplifies psychological pressure during losing streaks, which is the most common reason traders abandon a system prematurely. The strategy is designed to be followed systematically — give it enough capital that a normal drawdown doesn't feel like a crisis.

Subscribe on Collective2, then connect your broker account via AutoTrade. Supported brokers include AMP Clearing (CQG), Interactive Brokers, Tradovate, Tradier, and others. Once connected, trades execute automatically in your account whenever the strategy places a signal — no action required from you. Full setup instructions are available at collective2.com/lets-get-started.

Losing streaks are a mathematical certainty in any probabilistic trading system with a 62% win rate — they are not a signal that the strategy is broken. The edge emerges over a large sample of trades, not individual outcomes. Stopping during a drawdown and restarting after a winning period is the most reliable way to underperform the strategy's published results. The correct response to a drawdown is to do nothing and let the system continue executing.

Yes. The strategy holds Trades-Own-Strategy (TOS) certification from Collective2, which independently verifies that the manager is trading this exact system in a real funded brokerage account at AMP Clearing (CQG). Since certification began on March 10, 2026, 99.3% of signals have been executed live. The manager's account runs at 100% scaling — identical to what subscribers receive.

MNQ futures are highly liquid, but large subscriber counts can create meaningful market impact at the specific price levels the strategy targets. Keeping subscriber count limited protects execution quality for existing subscribers. A strategy that degrades its own edge through overcrowding is not in anyone's interest.

The results are real. The strategy manager holds Trades-Own-Strategy (TOS) certification from Collective2, which independently verifies that real capital is being risked in a live funded brokerage account at AMP Clearing (CQG) — not a paper account, not a simulation. The manager trades this system at 100% scaling, meaning every signal that goes to subscribers is also executed with his own money on the line. The Collective2 track record reflects those live signals precisely. There may be minor fill differences between the published record and individual subscriber accounts due to broker and timing variation, but the signal record itself is a near-exact mirror of real trading. Since TOS certification on March 10, 2026, 99.3% of signals have been executed live.

Both. Depending on the model's signal and market conditions, a trade may close within the same session or remain open overnight — and in some cases across multiple days. The strategy does not impose an artificial intraday-only constraint; it exits when the model's criteria are met, regardless of the clock.

This makes it important to understand your broker's overnight margin requirements before subscribing. For MNQ, most brokers require a significantly higher balance to hold a position through the session close. As a practical guideline, maintaining at least $4,400 per MNQ contract is typically sufficient to satisfy overnight margin — falling below that threshold risks your broker automatically closing the position at end of day, which may not align with the strategy's intended exit. Confirm the exact requirements with your broker before going live.

This is a fair question, and the honest answer is: backtest data would tell you less than you think, and the live record tells you more.

Nasdaq 100 is highly regime-dependent. The structural dynamics of a trending bull market, a volatility spike, or a macro-driven selloff are fundamentally different — and a model optimized on one regime can look extraordinary in backtesting while failing in the next. Any backtest numbers I could share would be subject to overfitting, look-ahead bias, and the basic problem that the historical market conditions that generated them no longer exist in the same form.

This is precisely why the XGBoost model is retrained on a rolling basis — it is designed to adapt to current regime conditions rather than lock in on a static historical fit. A strategy that retrains regularly cannot produce a meaningful multi-year backtest anyway, because the model itself changes over time.

Trade the
Nasdaq 100
without discretion.

Performance

The Framework

FAQ

Trade theNasdaq 100without discretion.

Performance

The Framework

FAQ

Trade the
Nasdaq 100
without discretion.