I think that "top 10" is a bit misleading because for most of those games, hardly anyone on the site actually played them. Their style of voting was a bit out of the ordinary - instead of voting for the best game, what they did is have everyone rate every game that they played from 2015. The list is derived from the Bayesian average ratings of each game.
If you limit the list to games that were played by a reasonable number of voters, then their full GotY rankings would be:
1. Underrail (played by 58%) - rating 4.20
2. AoD (played by 63%) - rating 4.18
3. Witcher 3 (played by 51%) - rating 4.00
4. Serpent in the Staglands (played by 35%) - rating 3.56
5. Shadowrun: Hong Kong (played by 44%) - rating 3.52
6. Pillars of Eternity (played by 76%) - rating 3.30
7. Sunless Sea (played by 17%) - rating 2.91
8. Undertale (played by 29%) - rating 2.60
9. Fallout 4 (played by 44%) - rating 1.86
10. Sword Coast Legends (played by 11%) - rating 1.71
Every other game was only played by 8% or less of the voters. (The full spreadsheet is available at their site, my list is taken from the GotY tab of it.) #7 and #10 weren't played by many, so maybe they should be dropped too. Now the list begins to look pretty reasonable.
If they had actually voted like we do here, I'm assuming AoD would have won their GotY, due to its somewhat higher % of players than Underrail (and Witcher 3), and similar rating.