December 2022-July 2023: https://boardgamearena.com/forum/viewtopic.php?t=31498
To make a longer story short, I now have official approval to analyze Agricola stats and share the results with the community. I'm very thankful to BGA and Lookout for their time and effort to make this process compliant with modern data laws and for letting me proceed with analysis that will benefit the community.
Unless otherwise stated, all data is for games ended between Apr 2 and Sep 29, where all players end with 200+ Elo, no additional spaces, and with both CD and "New" cards enabled. (The New cards came in late spring/early summer, so the start date is effectively later.)
First is analysis of Seating Order (starting position):
2-player (n=952)
Code: Select all
order elo_start score end_1st end_2nd avg_rank
1 390.73 42.26 51.5% 48.5% 1.49
2 381.63 42.02 48.9% 51.1% 1.51
Code: Select all
order elo_start score end_1st end_2nd avg_rank
1 390.24 47.72 48.0% 52.0% 1.52
2 384.56 48.12 52.3% 47.7% 1.48
Code: Select all
order elo_start score end_1st end_2nd end_3rd avg_rank
1 311.12 44.03 38.3% 31.7% 30.0% 1.92
2 332.28 44.29 37.2% 32.8% 30.0% 1.93
3 328.63 42.62 25.0% 35.6% 39.4% 2.14
Code: Select all
order elo_start score end_1st end_2nd end_3rd end_4th avg_rank
1 420.27 44.15 25.1% 25.5% 25.2% 24.1% 2.48
2 422.27 44.50 26.9% 26.5% 23.3% 23.3% 2.43
3 418.19 44.12 25.8% 24.9% 24.5% 24.8% 2.48
4 418.52 43.64 23.1% 23.3% 26.7% 26.9% 2.57
Code: Select all
order elo_start score end_1st end_2nd end_3rd end_4th avg_rank
1 392.40 46.68 26.2% 25.8% 25.7% 22.3% 2.44
2 394.12 46.73 26.7% 23.7% 26.5% 23.1% 2.46
3 394.55 46.45 25.5% 26.1% 22.3% 26.1% 2.49
4 399.45 45.73 22.7% 24.0% 25.2% 28.1% 2.59
4-player with banlist, Expert players (n=3114)
Code: Select all
order elo_start score end_1st end_2nd end_3rd end_4th avg_rank
1 570.63 48.51 40.7% 28.4% 18.3% 12.6% 2.03
2 567.13 48.68 42.1% 29.2% 17.9% 10.9% 1.98
3 564.93 48.29 40.5% 26.4% 19.4% 13.7% 2.06
4 572.90 47.84 36.3% 28.5% 20.7% 14.5% 2.13
Code: Select all
order elo_start score end_1st end_2nd end_3rd end_4th avg_rank
1 571.87 50.64 44.7% 31.2% 17.7% 6.4% 1.86
2 581.26 51.77 48.9% 24.1% 19.5% 7.5% 1.86
3 575.41 51.34 47.9% 25.7% 17.1% 9.3% 1.88
4 579.61 50.42 40.1% 34.3% 19.0% 6.6% 1.92
Across all formats (with the exception of 2-player AAS!) there is a clear disadvantage to being seated last, in terms of both score and winning chances. Furthermore, in 3-player and 4-player games, being seated 2nd is associated with better scores and winning chances than being seated 1st. Expert players appear to be affected by seating order in the same way the typical player is at a 200+ Elo, 4-player table. In the aggregate, these effects do not appear to be determined by differences in starting Elo, just seating order.
Key terms for the card statistics in following posts:
PWR is a measure of how often a card is on a 1st place tableau. ADP is the average draft position. APR is the average played round. WtdPWR is an experimental variant of PWR where instead of counting 1st places as 1 credit, I gave 0.6 credits to 1st place, 0.3 credits to 2nd place, and 0.1 credits to 3rd place. Elo/Play is a new calculation, the average Elo won or lost across all times the card was played.
The latter 2 stats are attempts to evaluate which cards are possibly strong because of how often they help a player get 2nd or 3rd place, or more generally win Elo - as opposed to regular PWR which only cares about getting 1st place or not.