Non-parametric density estimation is the correct method to solve the problem of estimating VaR between data points in the historical simulation approach. This method treats data as if they were drawings from some unspecified or unknown empirical distribution function. The existing data points can be used to 'smooth' the data points, paving the way for VaR calculation at all confidence levels. This means that non-parametric density estimation allows for the estimation of VaR at non-discrete confidence levels, such as 95.5%, which is not possible with the historical simulation approach. This flexibility makes non-parametric density estimation a powerful tool in risk management, particularly in situations where the data does not follow a known or specified distribution.

Choice A is incorrect. Applying Brute Force does not address the limitation of historical simulation in estimating VaR between data points. It refers to a method of solving problems through sheer computational power rather than through more efficient, strategic methods. In the context of VaR estimation, it would involve calculating all possible outcomes and their probabilities, which is not feasible or efficient.

Choice B is incorrect. Bootstrapping is a resampling technique used to estimate statistics on a population by sampling a dataset with replacement. It can be used to estimate the distribution of a statistic and its confidence intervals but it does not specifically address the issue of estimating VaR between discrete data points in historical simulation.

Choice D is incorrect. The use of a large number of re-samples may improve the accuracy and reliability of estimates in some statistical methods but it does not solve the problem inherent in historical simulation where we cannot estimate VaR between discrete data points due to its nature.