In many ways, investing is an exercise in decision making. In order to ensure consistent decisions, firms develop investment models with which to evaluate opportunities.
The models (both qualitative and quantitative) provide a framework for evaluating risk and return. The model defines the strike zone and helps shift investment decisions away from debating the merits of a class of investing, ie. do we want to deals like this, to the merits of a specific investment. The class of investment, what type of deals do we do, is defined by the model, the object, ie discrete investment decision, is the output of the model in action.
Another way to think of a model is as an articulation of the risks the firm is willing to take. For example, Hummer Winblad's model is to invest in Series A, enterprise software start-ups, and to back capital efficient plans at reasonable post-money valuations. Explicitly in the model is a checklist defining our focus, and inherently in the model is a level of risk acceptance, i.e. we are wiling to accept certain risks subject to appropriate pricing, capital investment, and overlap with our expertise.
If a firm's risk asset acceptance terms are well defined, ie what a strike looks like, then the firm can focus on sourcing and executing to the model. Also, to the extent an investment is outside of the model, the exceptions (ie areas of misfit with model) can easily be identified and analyzed. The investor can say, "I recognize that this deal is off our curve in the following dimension, however, the exception and added risk is worth taking for the following reasons."
Defining acceptable risks and a target profile creates consistent decision making and increased productivity. It also helps quickly flag execptions and puts the onus on the deal champion to explain the logic in swinging out of the strike zone.
Models are tools for decision making, but they also provide a logical framework with which a team of individuals can use to maintain a common cadence and definition of risk. The moral is not to never to go off the model, but to always know when you are off the model and to ask if the potential return is worth taking the extra risk.