14 point dogs IIRC. Based on what we know about Sagarin which isn’t much (ie what goes into his formula) it’s probably among the least least accurate “popular” football predictors because there simply aren’t enough data points. From what I’ve been able to glean he looks at an extended tree of who beat who by how much and probably massages the data in other ways. Does it look at broad based matchups - strengths and weaknesses of individual teams? Efficiencies and explosiveness data, defensive havoc factors? I doubt it.