Points deducted in the event of a tie

[Sonnenköpfchen]

Can someone please explain to me why points are deducted for a draw. I can't understand it and it makes no sense to me. No one has won or lost

Kann mir bitte jemand erklären, warum bei einem Unentschieden Punkte abgezogen werden? Ich verstehe das nicht und es ergibt für mich keinen Sinn. Niemand hat gewonnen oder verloren

[Bray]

Elo is a predictive tool, not a trophy system. When the higher rated player doesn't win, the rating is adjusted down. It's the same reason that your lower rated opponent gained points from the tie. The amount of change is proportional to the rating difference prior to the game.

You can dig through the details here if you're interested.
https://en.wikipedia.org/wiki/Elo_rating_system

[Meeplelowda]

Can someone please explain to me why points are deducted for a draw. I can't understand it and it makes no sense to me. No one has won or lost

Kann mir bitte jemand erklären, warum bei einem Unentschieden Punkte abgezogen werden? Ich verstehe das nicht und es ergibt für mich keinen Sinn. Niemand hat gewonnen oder verloren

The calculation of ELO is explained in the FAQ: https://boardgamearena.com/faq?anchor=faq_account_elo

The basic principle is that ELO is a measure of your skill in a game. We will assume this is true, because the whole concept of ELO depends on this being true. If the outcome of a game depends on your skill (we'll dig into this a little deeper later), then a higher skilled person would be expected to beat a lower skilled person more often than not. The bigger the rating difference, the more success the higher skilled person should have on average against the lower skilled person.

So, given those assumptions, if the higher skilled person either ties with or loses to a lower skilled person, then the ELO framework adjusts the higher skilled person's rating downward because the tie/loss indicates that they are slightly overrated, and adjusts the lower skilled person's rating upward because the tie/loss indicates that they are slightly underrated. In theory if these two players play each other repeatedly their ratings will trend toward an equilibrium value that reflects their true rating as long as neither gets better (or worse, which I guess could happen).

Now we assumed above that ELO reflects skill. But in a game that involves no skill, such as flipping a coin to determine the winner, ELO is effectively meaningless. Very few games involve no skill, but the less skill involved, the less it is certain exactly what ELO is measuring. I guess you could say it measures your ability to make the best out of the random situation you've been handed, but sometimes making the best out of the situation merely means losing by less, not winning.

You will also note that the more a high amount of luck determines the outcome of a game due to randomness, the fewer very high ELOs you will see among active players. The less randomness in the game, the more populated the peak ELOs are. For example, the current peak rating among active Can't Stop players is 451. The current peak rating among active Hive players is 1071, and there are eight players rated over 800. There are also 12 players over 800 ELO in Azul, indicating that skill clearly dominates the outcome despite the randomness.

tl;dr: If you tie someone you "should have" beaten, it makes sense for your rating to decrease.