This thesis investigates the predictability of daily returns on the iShares Global Clean Energy ETF (ICLN) using Markov Regime Switching (MRS) models. The main hypothesis is that clean energy markets do not move in a single, stable pattern: periods of high volatility and drawdowns alternate with calmer, growth-oriented phases. A secondary objective was to investigate if exogenous predictors, and more specifically, clean-energy related uncertainty indices would improve models performance. Results show that allowing for multiple regimes improves in- sample results relative to single-regime benchmarks. Additionally, the ARX and MS-ARX, based on graphical representation appear to have the best out of sample performance.