We carry out an in-depth analysis of the capability of the upcoming space-based gravitational wave mission eLISA in addressing the Hubble tension, with primary focus on observations at intermediate redshifts (3 < z < 8). We consider six different parametrisations representing different classes of cosmological models, which we constrain using the latest datasets of CMB + BAO + SNIa, to find out the up-to-date tensions with direct measurement data. Subsequently, these constraints are used to construct mock catalogues for eLISA. We then employ a three-pronged approach involving Fisher analysis, Markov Chain Monte Carlo, and Machine Learning using Gaussian Processes on the simulated catalogues to forecast on the future performance of each model. Based on our analysis, we present a thorough comparison among the three methods as forecasting tools, as well as among the different models predicted by each method. Our analysis confirms that eLISA would constrain H 0 at the sub-percent level. MCMC and GP results predict reduced tensions for models which are currently harder to reconcile with direct measurements of H 0 , whereas no significant change occurs for models at lesser tensions with the latter. This feature warrants further investigation in this direction.