Risk taking is described as an integral part of financial services. For micro-financing in particular, engaging in proactive risk taking is essential to their viability and long term sustainability. Maintaining a good strategy that ensures an optimal mix in risk-return trade-off is much more important for the microfinance banks (MFBs) that operate on a for-profit basis. Having faulted the value-at-risk technique which is common in the asset and liability literature, we introduce the multi-stage stochastic programming using econometric time series model. Specifically, for the scenario generation, we specify a VaR model with the inclusion of dichotomy regime which captures the multi-stage characteristics of assets. We use the liability derived investment (LDI) model to generate the liability series over the period of study. The optimization result showed that MFBs in Nigeria are by far more risk averse than they are profit seeking. This comes with the attendant effect of not being able to achieve the outreach and sustainability objectives to the fullest. MFBs in Nigeria need to look into their investment strategy with a view to structuring the mix and value of the balance sheet components at different periods to meet their stated objectives.