Player vs player stats ---------------------- player 1 Derek player 2 3 Norbert player 3 6 2 R2 player 4 2 4 2 Arne player 5 3 4 4 2 Jonas player 6 4 2 4 5 4 Palle --------------------------- player 1 2 3 4 5 D N R A J --------------------------- The table above shows how often each player played against each other. Those that clearly felt most pleasure by each others company were R2 and Derek who played 6 duels. Players who seemed to 'avoid' each other (only 2 duels) were Derek/Arne, R2/Norbert, Palle/Norbert, R2/Arne and Jonas/Arne. Deck vs deck stats ------------------ deck 1 Black shiny - Arnes sad story of Johanna deck 2 2 Black dull - Dereks All Soul's Night deck 3 4 2 White - (Derek) The Serra Inquisition deck 4 2 4 3 Gold shiny - Mattimmas the Sorcerer(ess) deck 5 5 4 4 2 Murky-gold - Palles Vault evaluation deck 6 2 4 6 3 4 Blue - (Jonas) Never taunt a bat ------------------------- deck 1 2 3 4 5 A Db Dw M P ------------------------- Again here 5 deck vs deck matchnings were only tested two times while the deck of the batty elephant did battle agaist the Serra Inquisition 6 times. Who played most with which deck? duels total -------------------------------- player 1 1 2 5 1 5 4 18 Derek player 2 3 3 5 1 2 1 15 Norbert player 3 2 3 3 1 5 4 18 R2 player 4 4 2 1 2 1 5 15 Arne player 5 4 3 3 3 1 3 17 Jonas player 6 1 3 2 6 5 2 19 Palle -------------------------------- deck 1 2 3 4 5 6 A Db Dw M P J -------------------------------- ---------------------------------------------------------------------- Player wins - losses ---------------------------------------------------------------------- This is the simple version, just counting wins/losses ignoring opponent stats and deck stats. I shall try to run the fancy code later - but with so few results it is not likely to work well. ------------------------------------------------ player wins losses total win-fraction ------------------------------------------------ Palle 14 5 19 74 % Arne 10 5 15 67 % Derek 11 7 18 61 % Jonas 6 11 17 35 % Norbert 5 10 15 33 % R2 5 13 18 28 % ------------------------------------------------ I am particular delighted to see that nobody fell completely off the chart - and even our "trainee" won more than 1/4 of his duels. I'm quite impressed by this, well done R2. ---------------------------------------------------------------------- Deck wins - losses ---------------------------------------------------------------------- This is the simple version, just counting wins/losses ignoring opponent stats and player stats. I shall try to run the fancy code later - but with so few results it is not likely to work well. ------------------------------------------------ deck wins losses total win-fraction ------------------------------------------------ Palle 13 6 19 68 % Matthias 9 5 14 64 % Jonas 10 9 19 53 % Arne 7 8 15 47 % Derek W 8 11 19 42 % Derek B 4 12 16 25 % ------------------------------------------------ ---------------------------------------------------------------------- Results from total multi-parameter fit ---------------------------------------------------------------------- Those are the results from looking for an optimized set of parameters to describe the relative log(strength) of both players and decks. I am not sure what the errors are for a dataset of only 51 duels, but I guess they are larger than a factor 2 (i.e. larger than +/-0.3 in log). Jonas and myself may run some simulations to determine the actual errorbars later. The data set is nicely distributed (fairly evenly with no zeros in any of the three diagrams above) which helped a lot for the stability of the solution. player log(strength) deck log(strength) 6 2.00 5 2.00 4 1.84 4 1.95 1 1.57 6 1.66 5 1.19 1 1.62 2 0.94 3 1.42 3 0.91 2 1.01 ----------------------------------------------------------------------