Racing the News: Market Efficiency and First-Mover Advantage After U.S. Macroeconomic Releases at Eurex and CME

GitHub Repository: https://github.com/QuantLet/HFT_Deutsche_Boerse

Peer Review Status

Attribute	Detail
Target Journal	Journal of Financial Economics (JFE)
Submission Date	November 2025
Outcome	Reject
Referee Concerns	4 fatal, 8 additional major, 19 minor

The paper was submitted to JFE and received a detailed referee report recommending rejection. The 4 fatal concerns address: (1) theoretical framework disconnected from empirics, (2) no identification strategy, (3) pre-physical reaction times likely timestamp artifacts, and (4) massive multiple testing problem. See Referee Report Analysis for full documentation of all concerns.

Authors

Author	Affiliation	Contact
Rahul Tak	Bucharest University of Economic Studies; IDA Institute Digital Assets	takrahul24@stud.ase.ro
Joerg Osterrieder	University of Twente, Faculty of Behavioral Management and Social Sciences	joerg.osterrieder@utwente.nl
Stefan Schlamp	Quantitative Analytics, Deutsche Borse, Eschborn	stefan.schlamp@deutsche-boerse.com
Daniel Traian Pele	Bucharest University of Economic Studies; IDA; Institute for Economic Forecasting, Romanian Academy	danpele@ase.ro

Executive Summary

This research investigates the limits of market efficiency by examining ultra-high-frequency reactions to U.S. macroeconomic announcements at Eurex (European derivatives exchange, Deutsche Borse) and the Chicago Mercantile Exchange (CME). Using nanosecond-precision timestamp data from 2019 to 2025, the study documents:

Market reactions occurring at 4 nanoseconds (Eurex) and 2 microseconds (CME), approaching physical transmission limits
A profound first-mover advantage: the vast majority of price discovery and trading profits concentrate within the first 200 microseconds on CME and 30 milliseconds on Eurex
Monotonic PnL decay on CME where trades beyond the initial 200 us window result in average losses
Significant asset class heterogeneity: fixed-income futures dominate aggregate volume and profit, while equity index futures exhibit higher immediate price sensitivity
Evidence that modern markets are efficient at the microsecond scale, driven by a socially costly speed race

Project Overview

flowchart TB
    subgraph DATA["Raw Data Sources"]
        EUREX_RAW["Eurex Exchange\n(nanosecond timestamps)"]
        CME_RAW["CME Exchange\n(microsecond timestamps)"]
        DBAPI["Deutsche Borse\nMarketplace API"]
    end

    subgraph PROCESSING["Data Processing"]
        PICKLE["Event Pickle Files\n(trade reactions around\nmacro announcements)"]
        CSV["Processed CSVs\n(CME + Eurex individual data)"]
        PKL["Aggregated PKL Cache\n(Eurex ISM volume)"]
        XETRA["XETRA CSVs\n(2025_Info_to_Alpha/Data/)"]
    end

    subgraph NOTEBOOKS["Analysis Notebooks"]
        NB1["Distribution Difference CME\n(Statistical Tests)"]
        NB2["Distribution Difference Eurex\n(Statistical Tests)"]
        NB3["Pickle Analysis\n(Event Visualization)"]
        NB4["Final Plot Creator\n(Publication Plots)"]
        NB5["Eurex Volume Analytics\n(Latency-Notional)"]
        NB6["PnL Analytics\n(Correlations + ML)"]
    end

    subgraph OUTPUT["Research Output"]
        PAPER2024["2024 Paper\n(Eurex-only)"]
        PAPER2025["2025 Paper\n(CME + Eurex)"]
        PLOTS["Publication-Ready\nFigures & Tables"]
    end

    subgraph REVIEW["Peer Review"]
        REFEREE["JFE Referee Report\n(Reject)"]
        ANNOTATED["Annotated Paper\n(with reviewer notes)"]
    end

    EUREX_RAW --> PICKLE
    CME_RAW --> PICKLE
    DBAPI --> XETRA
    PICKLE --> CSV
    PICKLE --> PKL

    PICKLE --> NB3
    PICKLE --> NB4
    PICKLE --> NB1
    CSV --> NB1
    CSV --> NB6
    PICKLE --> NB2
    PKL --> NB5

    NB1 --> PLOTS
    NB2 --> PLOTS
    NB3 --> PLOTS
    NB4 --> PLOTS
    NB5 --> PLOTS
    NB6 --> PLOTS

    PLOTS --> PAPER2024
    PLOTS --> PAPER2025

    PAPER2025 -.->|"submitted to JFE"| REFEREE
    PAPER2025 -->|"copy + annotations"| ANNOTATED

Quick Start Guide

Reader Type	Start Here	Then Explore
Researcher	Research Context	Key Findings > Statistical Methodology
Quant Developer	Data Structures	Data Pipeline > Notebook Deep Dives (see sidebar)
Paper Reviewer	Referee Report Analysis	LaTeX Paper Structure > Figures & Tables Catalog
New Contributor	Repository Architecture	Complete File Inventory > Environment Setup
Bibliographer	Bibliography & References	Paper Evolution > Plot Gallery

Macroeconomic Events Studied

Event	Abbreviation	Source	Typical Release	Frequency
ISM Manufacturing PMI	ISM_MANUFACTURING	Institute for Supply Management	10:00 AM ET	Monthly
ISM Services PMI	ISM_SERVICES	Institute for Supply Management	10:00 AM ET	Monthly
Non-Farm Payrolls	NFP	Bureau of Labor Statistics	8:30 AM ET	Monthly
FOMC Statement	FOMC	Federal Reserve	2:00 PM ET	8x per year

Exchanges Under Study

Exchange	Location	Timestamp Precision	Asset Focus	Operator
Eurex	Frankfurt, Germany	Nanosecond (1 ns)	European derivatives (all futures)	Deutsche Borse
CME	Chicago, USA	Microsecond (1 us)	US derivatives (equity + fixed income futures)	CME Group

Key Time Windows

Interval	From	To	Significance
Ultra-fast	0	200 us	First-mover advantage window; highest PnL concentration
Fast	200 us	30 ms	Extended profitability on Eurex; losses begin on CME
Medium	30 ms	100 ms	Cross-Atlantic transmission window
Extended	100 ms	10 s	Full markout evaluation period

Glossary of Key Terms

Term	Definition
Markout PnL	Profit/loss measured at a fixed time horizon after trade execution
Aggressor side	Whether the initiating party was buying or selling
Basis points (BP)	1/100th of a percentage point; standard unit for price changes
Co-location	Placing trading servers in the same data center as the exchange matching engine
t3a timestamp	Eurex network switch arrival timestamp (outermost measurable point)
Offset	Temporal window relative to announcement (e.g., 100ms, next_1day)
Pre-action	Trade flagged as occurring before the event timestamp
Reaction	Trade occurring after the event timestamp
KS statistic	Maximum distance between two empirical CDFs
Rank-biserial r	Effect size for Mann-Whitney U test; proportion of concordant pairs
Fisher z	Transform to compare independent correlation coefficients
OOB score	Out-of-bag error estimate from Random Forest bagging
HHI	Herfindahl-Hirschman Index measuring market concentration

Documentation Structure

mindmap
  root((Documentation))
    Overview
      Research Context
      Key Findings
    Data
      Structures
      Pipeline
      Exploration
      XETRA
    Notebooks
      6 Deep Dives
    Methodology
      Statistical Tests
      ML Pipeline
    Results
      Executive Summary
      Cross-Exchange
      Latency
      Statistical Testing
      Temporal Offsets
      Event Time-Series
    Paper
      Referee Report
      LaTeX Structure
      Bibliography
    Evolution
      Paper Timeline
      XETRA Infrastructure

Key Visualizations

US Event Reactions at Eurex

CME Cumulative PnL

Price Formation Forecast

Home