Is the rating of top player overrated ?

[JJJ]

I put the question, because when they play in tournament with oppent having a much lower rating, they seems to almost always lost many of their elo point.

[Uri Blass]

I put the question, because when they play in tournament with oppent having a much lower rating, they seems to almost always lost many of their elo point.

I do not understand.

Can you give specific numbers?

for example:

1)Top player is (for example player with rating above 2700)
2)Definitions of the games that you consider(for example games against players that are more than 200 elo weaker)
3)Number of games you consider.
4)Average loss in elo that they lose per game.

[JJJ]

Top 26 players , most of them loosing elo in world cup knock out.

[Tony P.]

Specifically in a KO tournament, one of the main reasons why strong players lose rating points is that they play it safe. To advance to the next round, it's enough to score 1.5:0.5 and often even 1:1 in the classical part because their advantage in the tiebreaks is big enough. Thus a strong player often opts to draw a weaker opponent in a position where an attempt to win would increase the risk of loss, to save energy for later, tougher, rounds.

[JJJ]

Still many were defeated despite this strategy.

[Tony P.]

Yes, indeed. I guess that some of them weren't motivated enough, especially Carlsen, who didn't have to be a finalist to qualify for the next championship match and, to make things 'worse', has apparently recently got into a romantic relationship stealing some time and energy from his chess study. Some of them generally have rather unstable performance (very strong tournament runs alternating with dramatic failures), like Anand as of late. Some of them weren't playing at their full strength in the early rounds in an attempt to save energy for later stages. Some of them just didn't manage to properly navigate the complications that the weaker players were understandably creating.

Besides, the Elo rating model used by FIDE is rather inaccurate and underestimates weaker players' expected score in pairs with a large rating gap, thus deducting a bit too much from stronger players' ratings when they fail to win, as opposed to the Glicko system used by USCF (to some extent) and major online chess servers.