meanydeany

MSc Economics thesis · Revised 14 October 2026

Volatility Regime Filtering in Futures Markets

How does daily volatility conditioning change a long-only intraday breakout framework? This thesis studies an EGARCH-based participation gate and exposure-scaling bundle in NQ, ES and Crude Oil (CL) futures.

Read revised thesisOriginal submissionRevision notes
Markets
NQ · ES · Crude Oil (CL)
Frequency
Daily model · 5-minute evaluation
Model role
Participation gate and exposure scaling
Evidence
Retrospective development comparisons

Research scope

Volatility context and a separate directional rule

The thesis asks whether an econometric volatility layer can improve the discipline of an otherwise comparable intraday research framework.

The research combines daily EGARCH volatility estimates with five-minute futures observations. A regime gate determines participation, while exposure scaling changes its size. Long-only breakout entries use separate RSI and EMA conditions.

The 14 October 2026 revision treats the reported backtests as retrospective development comparisons. The gate and scaling form a combined intervention; their individual effects and the value of EGARCH asymmetry are not separately identified. Information timing, source provenance and execution accounting limit the performance claims.

Research tags

Revised academic study
Financial econometricsEGARCHNQ futuresES futuresCrude Oil (CL) futuresIntraday dataRisk filtering

Research question

What changes when volatility conditions participation and size?

Model input
Daily futures returns
Research output
Conditional-volatility regime
Intraday context
Five-minute NQ, ES, and Crude Oil (CL) data
Comparison
Gate-and-scaling bundle versus no-filter framework

Methodology

Three layers with separate responsibilities

  1. Volatility layer01
    • EGARCH(1,1) with Student-t innovations
    • Daily conditional-volatility estimates
    • Percentile-based volatility-regime classification
    • Asymmetric response to signed return shocks
  2. Intraday layer02
    • Five-minute intraday NQ, ES, and Crude Oil (CL) observations
    • Daily regime labels mapped to intraday dates; availability timing requires audit
    • Long-only breakout entries with RSI and EMA conditions
    • Transaction-cost and risk-control assumptions
  3. Historical comparison03
    • EGARCH gate-and-scaling bundle versus no-filter ablation
    • Alternative volatility filters
    • Historical walk-forward comparisons
    • Reported bootstrap, subperiod, and market-level checks

Framework architecture

A traceable path from data to comparison

  1. 01

    Data

    Daily and five-minute NQ, ES, and Crude Oil (CL) futures observations

  2. 02

    Volatility

    EGARCH conditional volatility and regime labels

  3. 03

    Participation

    A volatility gate and exposure scaling operate as a bundle

  4. 04

    Direction

    Long-only breakout rules with RSI and EMA conditions

  5. 05

    Comparison

    Historical ablation and walk-forward results, with chronology and accounting limits

Results interpretation

What the historical comparison can tell us

Research contribution
  • The thesis develops a testable separation between daily volatility context and intraday directional rules.
  • Reported backtests compare historical configurations during research development; they are not an independently reproduced performance record.
  • The ablation changes a participation gate and exposure scaling together, so the comparison concerns that combined design.
Evidence limits
  • The comparison does not separately identify the effect of the gate, scaling, or EGARCH asymmetry.
  • Unresolved timing, provenance and execution accounting prevent a clean claim of independent out-of-sample profitability.
  • Academic backtests do not provide live-trading permission or investment advice.

Robustness

Limitations stay attached to the finding

Information timing
Full-sample volatility ranks incorporate future observations. Decision-time estimation and calibration require a separate evaluation.
Bundled intervention
Gate and exposure-scaling changes must be separated to attribute their individual contributions.
Execution accounting
Contract units, fills, costs and marked-to-market returns require reconciliation before performance can be accepted.
Development history
Historical comparisons and walk-forward labels do not alone establish an untouched evaluation sample.
Source provenance
Results need traceable data versions, executable specifications and a reproducible calculation chain.
Market scope
NQ, ES and Crude Oil (CL) results depend on their own samples and contract conventions; the revision is not a fresh empirical replication.

Capabilities demonstrated

  • 01Financial econometrics
  • 02EGARCH volatility modeling
  • 03Intraday futures data
  • 04Comparative research design
  • 05Chronology and accounting review

Research boundary

Academic evidence, not execution permission

This page describes an academic research project. It is not a live trading system, does not provide signals or execution, and is not investment advice.