Microstructure Signals on Major News Days

How Level II order-book models respond to event-driven market repricing—without ingesting news

Most market participants agree on one thing: large price moves are usually driven by information. The usual workflow is to incorporate that information explicitly—news feeds, calendars, analyst notes, and increasingly, automated summarization.

At Quantum Signals, our prediction models are trained on Level II limit order book (LOB) data—market microstructure. They are not connected to news sources, news analytics platforms, or LLM-driven “insights.” Even so, on many major event days, the signal reacts quickly enough to be useful as an intraday input.

The mechanism is straightforward: news doesn’t move markets directly—participants do. When an event changes expectations, traders adjust quotes, cancel and replace orders, shift liquidity, and change aggressiveness. That behavior is visible in the order book, often immediately. Microstructure-aware models can learn patterns in that behavior and produce short-horizon predictions.

This doesn’t mean the signal “captures every move.” On some days—especially sudden regime changes—it can miss or be on the wrong side. The goal is more modest: be directionally helpful more often than not, in a way that is measurable and systematic.

A quick framing note. We are not claiming that microstructure “beats” news or replaces news-aware models. The event days in this analysis are a stress test: does a microstructure-driven signal remain usable when the market is repricing around headlines? Since the signal is designed to run systematically every day, these sessions help answer whether it remains stable under event-driven conditions.

What the infographic shows

The infographic summarizes a set of major news days across 2023–2026. For each day we include:

• Total move (%): reported index move across pre-market + regular session
• RTH move (%): index move during our strategy’s core window (9:45 → 16:00 ET)
• Win Rate (%): signal accuracy during that window
• Return (%): return of a systematic intraday strategy using the signal

Results are shown for S&P 500 and Nasdaq-100 and the signal strategies are executed on ES and NQ index futures..

Examples from the event set

1) Fed day repricing (midday): Dec 18, 2024

Headline: Fed day: hawkish dots disappoint (fewer cuts signaled)
When news hit: Midday / early afternoon (FOMC timing)

This is a useful microstructure example because the information arrives mid-session and the market reprices quickly.

• S&P 500: Total -2.900% | RTH -2.885%
  
  • ES signal strategy: Win Rate 100.000% | Return +2.582%
• Nasdaq-100: Total -3.600% | RTH -3.336%
  
  • NQ signal strategy: Win Rate 100.000% | Return +2.864%

On days like this we typically see rapid changes in liquidity and imbalance after the release. The signal doesn’t “know” the Fed decision; it reacts to the market’s order-flow response.

2) Trade-war shock → continuation → reversal: Apr 3 / Apr 4 / Apr 9, 2025

This three-day sequence is helpful because the same macro theme produces distinct microstructure regimes: initial shock, follow-on escalation, and then a sharp reversal.

Apr 3, 2025 — Tariff shock (pre-market/open)

• S&P 500 Total -4.800% | RTH -1.583%
  
  • ES Win Rate 0.000% | Return -0.876%
• Nasdaq-100 Total -6.000% | RTH -1.565%
  
  • NQ Win Rate 0.000% | Return -0.854%

This is an example where the move is abrupt and the signal does not help. It’s representative of the kinds of days where the order book regime changes quickly and models can struggle.

Apr 4, 2025 — Retaliation escalation (pre-market/open)

• S&P 500 Total -6.000% | RTH -2.857%
  
  • ES Win Rate 52.540% | Return +0.023%
• Nasdaq-100 Total -5.800% | RTH -2.793%
  
  • NQ Win Rate 94.570% | Return +1.256%

Despite a large negative session, there was enough intraday structure for the signal to be directionally useful in NQ.

Apr 9, 2025 — Tariff pause → relief rally (midday/early afternoon)

• S&P 500 Total +9.500% | RTH +9.268%
  
  • ES Win Rate 90.670% | Return +5.993%
• Nasdaq-100 Total +12.200% | RTH +11.048%
  
  • NQ Win Rate 87.690% | Return +6.368%

This is a reversal day where the order book tends to show clear shifts in liquidity and aggressiveness as conditions normalize and positioning flips.

3) Election repricing: Nov 6, 2024 (pre-market/open)

Headline: U.S. election repricing: Trump victory sparks rally
When news hit: Pre-market / open

• S&P 500 Total +2.500% | RTH +0.544%
  
  • ES Win Rate 100.000% | Return +0.329%
• Nasdaq-100 Total +3.000% | RTH +0.714%
  
  • NQ Win Rate 3.990% | Return -0.399%

This is a good reminder that “headline direction” does not guarantee clean intraday structure across instruments. Even when the index finishes up, the intraday path can be rotational or choppy, and the signal can be wrong.

4) AI spending / ROI concerns: Feb 5, 2026 (pre-market/open)

Headline: AI capex/ROI fears (Alphabet 2026 spending plan)
When news hit: Pre-market / open

• S&P 500 Total -1.200% | RTH -0.747%
  
  • ES Win Rate 82.970% | Return +0.245%
• Nasdaq-100 Total -1.600% | RTH -1.031%
  
  • NQ Win Rate 80.800% | Return +0.341%

Industry-driven repricings like this often show up as sector concentration and liquidity shifts that are visible in the book without needing to interpret the narrative.

5) Global risk-off overnight, rebound during RTH: Aug 5, 2024

Headline: Global risk-off: Nikkei shock + U.S. slowdown fears
When news hit: Overnight / pre-market

This is a useful example because the headline narrative was clearly risk-off and the reported index move reflected that, but the market stabilized and rebounded during the regular session. Since the strategy is evaluated in our 9:45 → 16:00 ET window, it primarily captured the intraday recovery rather than the overnight gap.

• S&P 500: Total -3.000% | RTH +0.321%
  
  • ES signal strategy: Win Rate 94.570% | Return +0.416%
• Nasdaq-100: Total -3.400% | RTH +0.964%
  
  • NQ signal strategy: Win Rate 81.160% | Return +0.339%

This highlights a practical point: on some event days, a large portion of the move occurs outside the trading window (overnight gaps), while the regular session is dominated by mean reversion, liquidity return, and repositioning.

Takeaway

Microstructure-only models do not replace fundamental information. They also do not capture every event correctly. What they can do is provide a systematic way to respond to how market participants incorporate information—as expressed through quotes, depth, cancellations, and aggressive flow.

The event set in the infographic includes both examples where the signal was helpful and examples where it wasn’t. That mixture is important. The practical question for systematic users isn’t whether a model is “news-aware,” but whether it is consistently responsive to the market’s reaction in a way that adds measurable value within the trading window.

Disclaimer: For informational purposes only. Not investment advice. Past performance is not indicative of future results.