The data at the start might be boring, but the discussion gets better towards the bottom (at least I think so).
Data
First is the simple table of results. Obviously, my win rate is much higher when I have the start.
The next two graphs show score and rank changes over the month. Pre-expansion I reached rank 2. After the expansion, I ranged 8-693. I finished at rank 80 (I play to last day).
The next graph shows my win and loss streaks.
Discussion
First, let's consider multiple games against competitors (one month only). The table below shows my matches. Numbers in red are where I won more games in the series, orange is a tie, and green is where I won more games. I played 45 people once and two people ten times each.
Next is my points per game, given the number of wins per series. You will note that with 5 games played and only 2 wins, I gained 50.2 points per game, and when I played 10 games and won 6, I lost 15.8 points per game. You also note that I lost more points per loss than I gained per win in single-game matches. Even worse, look at the 2 game series: if I lost both games, I lost 105 points per game...but if I won both games, I only gained 58.6 points per game. It was equally bad in my 4 game matches.
A series SHOULD see the orange (ties) being zero and the further from the orange (towards red of green) getting progressively better or worse. There is WAY too much randomness in the scores. I should never gain points if I lose more games to a player or lose points if I win more games against a player.
The ranking system is broken because of the peakiness of the data and because it does not recognize series. In a game of RNG and luck, skill/strategy only shows up in multiple games against the same person. The current system only looks at one-off games and uses a mechanism based on chess where there is no RNG nor luck.
We need a new ranking system!!!
I will carry this into a new discussion because I am going to draw on three months data.