Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.16800

4Cast Lab: Open-Source App for Machine Learning & Statistical Analysis

Periklis Gogas
View original source ↗Read source document ↗Export metadata ↓

Abstract

4Cast Lab is a free, open-source application for machine learning, econometrics and statistical analysis that runs on the user's own computer, with no installation, no programming and no data leaving the machine. It was developed at the E|M|E|F Lab, Department of Economics, Democritus University of Thrace, for researchers, analysts, instructors and students in any field. Its Forecasting mode trains eight learning algorithms (with three kernels for SVM/SVR), implemented from scratch, under leakage-safe validation for time-ordered data; it selects models by cross-validation, compares the selected model with every alternative and with naive benchmarks through formal tests of predictive accuracy, and documents each run in a methods report and a Python script that the user can export. Its Estimation mode covers the core of applied econometrics, from OLS with robust inference to ARIMA, ARDL, VAR/VECM and GARCH. This paper describes the design, the interface and the workflow of version 4.6.2, reports a validation of 1,234 numerical results against statsmodels, linearmodels, arch, scikit-learn and NIST certified values, illustrates the workflow with three examples, and discusses current limitations.

Explore related subjects

q-fin.CP

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Periklis Gogas. 2026-10-12. 4Cast Lab: Open-Source App for Machine Learning & Statistical Analysis. https://arxiv.org/abs/2610.16800

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Real interest rates, exchange rates and growth in three emerging economies: an exploratory analysis of Brazil, India and Nigeria

This paper studies the dynamic relationships between real growth, the real lending interest rate and the exchange rate in Brazil, India and Nigeria over 2000-2022. The theoretical framework places these relationships within the literature on the procyclicality of macroeconomic policies, ``fear of floating'' and the global financial cycle, without treating the observed variables as monetary policy instruments directly controlled by central banks. Activity is measured by real GDP in constant local currency (World Bank, NY.GDP.MKTP.KN). Real growth is defined as 100 x $Δ$log(real GDP) and the change in the exchange rate as 100 x $Δ$log(exchange rate), with the exchange rate quoted in units of local currency per dollar. The interest rate variable is the World Bank real interest rate (FR.INR.RINR), that is, a lending rate adjusted for inflation as measured by the GDP deflator; it is not a policy rate. Parsimonious bivariate VARs are estimated after examining stationarity, with lag selection by BIC and orthogonalised impulse responses. In the baseline specification, Brazil shows two pointwise exclusions of zero following a real interest rate innovation: a contraction at one year and a rebound at three years. These two features are not, however, robust to the same specification choices: the rebound disappears in a VAR(1), while the contraction is no longer distinguishable from zero when the sample ends in 2019. The estimated responses for India and Nigeria remain imprecise, as do the growth responses to exchange rate innovations. Given the very small sample, the large number of horizons examined, the absence of global financial variables and the non-structural nature of the innovations, the results are interpreted as descriptive and exploratory rather than as causal effects of monetary policy.

q-fin.CP↗
arXiv2026

The Fourier estimator of spot volatility: Unbounded coefficients and jumps in the price process

We study a Fourier-based estimator of the spot volatility of a semimartingale, originally introduced by Malliavin and Mancino, and extend its asymptotic theory. First, we remove the uniformly bounded volatility assumption and establish almost sure convergence of the estimated Fourier coefficients under mild integrability conditions. We further derive explicit convergence rates that yield almost sure convergence of the reconstructed spot volatility via Fejér inversion. We then extend the analysis to processes with jumps. In this setting, we provide a precise decomposition of the Fourier coefficient estimation error into diffusion-diffusion, diffusion-jump, jump-diffusion, and jump-jump interaction terms, and obtain $L^p$-rates for each component. We show that the estimated coefficients converge almost surely to the sum of the true volatility coefficients and an explicit jump-induced term. Moreover, we prove that an appropriate rescaling of the trigonometric reconstruction yields almost sure pathwise recovery of the quadratic jump process.

q-fin.CP↗
arXiv2026

Tail Instability and Scale Control in Conditional Normalizing Flows for Treasury Yield-Curve Scenarios

A conditional normalizing flow can produce extreme scenarios even when its point forecasts appear reasonable. We compare two fitted models for daily Treasury yield changes at eleven maturities reconstructed from Federal Reserve curve parameters. Both use an LSTM, an auxiliary location head, and an affine residual flow. On a previously unused 60-date window, the frozen Student-t model generated an absolute standardized coordinate of 34.3 million. A Gaussian-base model with tighter coupling scale control reached 14.72. Each generated 600,000 vectors. The original model's scenario-mean MSE was 33,908.95, while its location-head MSE was 0.3376. The stabilized model's two estimates remained close. Across five development windows, the stabilized maximum was 17.33 among three million vectors. Stability did not imply better prediction: historical simulation had better distributional scores and coverage closer to nominal on the new window. The original flow also had a better log score despite its extreme samples. Frozen-model evaluation is reproducible, but the comparison does not isolate architectural effects or establish a general forecasting advantage.

q-fin.CP↗