"If SIPN entries had been here" results:

CPOM (Schroeder, based on melt pond statistics) 4,2,2 = 8

UK met office (dynamic model, seasonal forecaster) -6,-6,none = -12

US Navy (dynamic model, often referred to on these boards as HYCOM) -4,-1,-4 = -9

NSIDC Slater model (concentration statistics) no entry,1,1 = 2 (spot on but no error margin given)

Nico Sun a.k.a. Tealight, (statistics on albedo calculations) 6,2,6 = 14

University of Washington (dynamic model, PIOMAS forced by a seasonal weather forecast) 2,6,6 = 14

Note how PIOMAS crushes the other dynamic models, not only because it was more accurate, but also because the other models have a very poor appreciation of just how inaccurate they are.

CPOM and Tealight are a lot closer than those numbers suggest, one being just the right side and the other just the wrong side of being put in higher value bins.

Statistical models did well, but it was a year that finished close to trend, and it needs a year with markedly good or bad melt to sort them out from one another.