> Historically they were used to show how the company was doing ex of one time events( almost exclusively negative events) such as layoffs, lawsuit settlements,

> Historically they were used to show how the company was doing ex of one time events( almost exclusively negative events) such as layoffs, lawsuit settlements, divesting/closing money loosing units, etc. Can these really be considered to be one-time events though? I would hypothesize that a company which has seen at least one lawsuit settlement is more likely to see another one down the road than a company which has never seen a lawsuit. Thus, the existence of a settlement should be a reasonable prediction of more losses to occur in the future. So why ignore them? It seems like it would be at least more reasonable to consider the past M GAAP results than the past M non-GAAP results when analyzing a trend. The latter will consistently overestimate growth, whereas the former will have more noise, but averages out to being more representative of the actual growth. And the analyst at least has the necessary data to reasonably calculate error-bounds on his/her estimate.