Writing your own line-specific rating engine subclass

The AbststractRatingEngine class is the parent class for all built-in line-specific rating engine classes. It serves two purposes.

• Defines a basic structure for rating that works for most lines
• Defines helper functions that could be useful across lines of business

The most important top-level method on AbstractRatingEngine is the rate method. The default rating plugin calls this method. This method is responsible for rating and then adjusting the Cost rows appropriately. Do not modify this method, generally speaking. It converts cost data objects to cost entity instances, adds them to the policy line, and removes no-longer-used costs on that policy line.

The rate method calls out to the rateOnly method. It rates costs per slice, then merges and prorates, and then finally rates window mode costs. That algorithm suffices for most lines of business. However, other lines such as workers’ compensation and inland marine override the rateOnly method to work differently in built-in code. You could override rateOnly in a rating engine to perform your own custom behavior. The rateOnly method returns the list of CostData objects, which the rate method converts to Cost rows to persist in the database.

For typical rating lines, your actual formulas and table lookups would happen in the rateSlice and rateWindow methods. The rateSlice method must rate a PolicyLine viewed as a slice on a specified slice date. The rateWindow method must rate all the window-mode costs for the PolicyLine. If you write your own rating engine subclass, writing your actual rating formulas in these methods are your primary tasks.

For a typical rating integration project, implementing your actual rating algorithm, typically in rateSlice and rateWindow, is the majority of your engineering time.

In nearly all cases, it is best to use AbstractRatingEngine as the parent class for new rating engines. There is a parent class of AbstractRatingEngine, called AbstractRatingEngineBase. If the normal paradigm of rating one line of business at a time does not mesh well with the rating algorithm, the new base class allows this. For example, to integrate with an external rating engine that rates all lines of the package in a single call. Refer to AbstractRatingEngineBase in Studio for details.