Conformal Prediction Intervals for Daily Equity Returns Under Volatility Regimes

Risk reports rely on interval forecasts, yet parametric volatility models lose their nominal coverage when markets turn turbulent. We ask whether distribution-free conformal prediction can give dependable one-step-ahead intervals for daily equity returns across volatility regimes. Under one online protocol we compare an unconditional Gaussian benchmark, Gaussian and Student-t GARCH(1,1) intervals, a volatility-scaled split conformal predictor, and adaptive conformal inference, alongside five machine learning interval methods, on four European indices and the S&P 500 from 1990 to 2026. The unconditional benchmark overcovers yet covers only about 85 percent of high-volatility days. The GARCH intervals are efficient and well calibrated; the conformal methods give the most accurate marginal coverage and the best coverage on turbulent days. Significance tests and a sensitivity analysis confirm that volatility-scaled conformal correction is most reliable in turbulence, at a small cost in width. Laporan risiko bergantung pada prakiraan interval, tetapi model volatilitas parametrik kehilangan cakupan nominalnya ketika pasar bergejolak. Kami menguji apakah prediksi konformal yang bebas distribusi dapat memberikan interval satu langkah ke depan yang andal untuk imbal hasil saham harian di berbagai rezim volatilitas. Di bawah satu protokol daring, kami membandingkan tolok ukur Gaussian tak bersyarat, interval GARCH(1,1) Gaussian dan Student-t, prediktor konformal terbelah berskala volatilitas, dan inferensi konformal adaptif, bersama lima metode interval berbasis pembelajaran mesin, pada empat indeks Eropa dan indeks S&P 500 dari 1990 hingga 2026. Tolok ukur tak bersyarat memiliki cakupan berlebih secara keseluruhan, namun hanya mencakup sekitar 85 persen hari bervolatilitas tinggi. Interval GARCH efisien dan terkalibrasi baik, sedangkan metode konformal memberikan kalibrasi marginal paling akurat dan cakupan terbaik pada hari bergejolak. Uji signifikansi dan analisis sensitivitas menegaskan bahwa koreksi konformal berskala volatilitas paling andal saat bergejolak, dengan biaya lebar yang kecil.

Authors

Publication Details

Journal
Jurnal Buana Informatika
Published
2026-10-30
DOI
https://doi.org/10.24002/jbi.v17i2.15190
Primary Topic
Financial Risk and Volatility Modeling
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Conformal Prediction Intervals for Daily Equity Returns Under Volatility Regimes

Yasi Dani, Fathimah Al-Ma’shumah
1 links
Jurnal Buana Informatika
Financial Risk and Volatility Modeling
article

Conformal Prediction Intervals for Daily Equity Returns Under Volatility Regimes

Yasi Dani, Fathimah Al-Ma’shumah
article en

Abstract

Risk reports rely on interval forecasts, yet parametric volatility models lose their nominal coverage when markets turn turbulent. We ask whether distribution-free conformal prediction can give dependable one-step-ahead intervals for daily equity returns across volatility regimes. Under one online protocol we compare an unconditional Gaussian benchmark, Gaussian and Student-t GARCH(1,1) intervals, a volatility-scaled split conformal predictor, and adaptive conformal inference, alongside five machine learning interval methods, on four European indices and the S&P 500 from 1990 to 2026. The unconditional benchmark overcovers yet covers only about 85 percent of high-volatility days. The GARCH intervals are efficient and well calibrated; the conformal methods give the most accurate marginal coverage and the best coverage on turbulent days. Significance tests and a sensitivity analysis confirm that volatility-scaled conformal correction is most reliable in turbulence, at a small cost in width. Laporan risiko bergantung pada prakiraan interval, tetapi model volatilitas parametrik kehilangan cakupan nominalnya ketika pasar bergejolak. Kami menguji apakah prediksi konformal yang bebas distribusi dapat memberikan interval satu langkah ke depan yang andal untuk imbal hasil saham harian di berbagai rezim volatilitas. Di bawah satu protokol daring, kami membandingkan tolok ukur Gaussian tak bersyarat, interval GARCH(1,1) Gaussian dan Student-t, prediktor konformal terbelah berskala volatilitas, dan inferensi konformal adaptif, bersama lima metode interval berbasis pembelajaran mesin, pada empat indeks Eropa dan indeks S&P 500 dari 1990 hingga 2026. Tolok ukur tak bersyarat memiliki cakupan berlebih secara keseluruhan, namun hanya mencakup sekitar 85 persen hari bervolatilitas tinggi. Interval GARCH efisien dan terkalibrasi baik, sedangkan metode konformal memberikan kalibrasi marginal paling akurat dan cakupan terbaik pada hari bergejolak. Uji signifikansi dan analisis sensitivitas menegaskan bahwa koreksi konformal berskala volatilitas paling andal saat bergejolak, dengan biaya lebar yang kecil.

Jurnal Buana InformatikaVol. 17(2)
Openalex Percentile: Top 9%
Financial Risk and Volatility Modeling
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon