Forecasts

Experimental forecast of next year's financial stress index, based on both indices in the current year. All models are shown with their actual past errors.

Naive model (previous year's value) · lowest error 50.0 2026 forecast · [32.5 ; 66.5] RMSE 13.2 · MAE 10.7

Support vector regression (SVR) 67.0 2026 forecast · [52.3 ; 99.1] RMSE 19.8 · MAE 16.5 Diebold-Mariano test vs naive model: p = 0.121

Linear regression 99.4 2026 forecast · [67.6 ; 134.0] RMSE 24.5 · MAE 20.5 Diebold-Mariano test vs naive model: p = 0.020

Out-of-sample evaluation over 17 years: each year is forecast using only earlier data. The interval in brackets matches the 10th and 90th percentiles of the model's past errors. When the naive model wins, the available information does not improve the forecast on such a short annual sample.

Forecast history and observed values

Year	Observed	Naive	Linear	SVR
2026	—	50.0	99.4	67.0
2025	50.0	50.5	83.7	67.9
2024	50.5	76.1	82.2	70.3
2023	76.1	82.5	90.7	69.7
2022	82.5	62.4	67.2	68.7
2021	62.4	82.2	82.8	70.8
2020	82.2	87.8	84.0	70.6
2019	87.8	65.7	59.0	66.8
2018	65.7	56.3	97.7	68.8
2017	56.3	59.3	74.3	69.0
2016	59.3	64.6	59.1	68.8
2015	64.6	69.6	68.3	70.0
2014	69.6	84.0	56.1	71.5
2013	84.0	100.0	71.4	70.2
2012	100.0	89.3	55.2	69.4
2011	89.3	75.2	84.1	57.6
2010	75.2	78.7	45.8	42.9
2009	78.7	77.8	36.3	38.4

Methodology and data

Target: financial stress index in year t. Predictors: financial stress and climate risk indices in year t−1. SVR hyperparameters are fixed in advance (RBF kernel, C = 1, ε = 0.1) and never tuned after seeing results. Everything is recomputed automatically when a new year is added.

This is a statistical exercise on about thirty annual observations. It is not an official forecast, and no result establishes a causal relationship.