The strength of the power (for the purposes of calculating the strength of the game, which is the denominator in the equation to determine the expected scores (X) for all the players) is the pro-rated average of all the players that were that power during the game. So, given this example:
Germany 1: Rated 890, played from S1901M to F1903R
Germany 2: Rated 1140, played from F1903B to F1909B and eliminated.
The strength of Germany is 14/45 times the power of Germany 1, plus 31/45 times the power of Germany 2 -- that's pro-rated by phases played.
Using e^(R/500) to determine the strength of a player from his rating, we see it's 0.31 * 5.93 + 0.69 * 9.78, or 8.59. That's the strength of a player who's rated 1075 (since 500 * ln(8.59) = 1075), which seems about right.
Strengths for the other powers are generated the same way, and expected values are seven times the strength of a power, divided by the sum of the strengths of the seven powers. This number will average 1.0.
For Germany, though, the first power is assigned a expected value as though he had played the whole game, whereas all replacements have an expected value (X) of 0. So if this game had a total strength of 51.22, then X for Germany 1 would be 7 * 5.93 / 51.22 = 0.81, and X for Germany 2 would be 0.0. S for both would be 0.0, since Germany was eliminated. So Germany 1 would suffer a full-sized hit to his rating, whereas Germany 2 wouldn't change at all. Germany 2 doesn't even have this game added to his total of games played.
The strength of the power is calculated the same way as above -- a pro-rated average of the players who ran the power during the game.
The expected values are also pro-rated, though. So in the example above (we'll assume the game ended in F1909B in a AEGIR draw), X for Germany 1 is 0.31 * 5.93 * 7 / 51.22 = 0.25 and X for Germany 2 is 0.69 * 9.78 * 7 / 51.22 = 0.92.
The points won from the game are pro-rated as well. For Germany 1, that's 0.31 * 7 / 5 = 0.43; for Germany 2, it's 0.69 * 7 / 5 = 0.97. These values are used as usual to figure out the Delta by which each player's rating goes up or down, based on this game.
Note that for Germany 2, his score (0.97) barely exceeded his expectation (0.92), which means his rating will go up only a little bit. If it had been a six-way draw or if his opponents had lower ratings (which would have increased his expected score), he might have had a lower score than expectation. In the case of a replacement player, he would NOT lower his rating based on this game -- it would remain unchanged. Replacement players can't lower their ratings -- ever! However, if you thinking of puffing up your JDPR rating by taking on lots of replacement positions, here's a word of warning -- it's a slow, laborious way to do so, and what gain you make will only leave you *overrated* for the games you *do* start (which will cause your rating to drop back towards its "actual" value).
[ The Zine |
Online Resources |
Showcase |
Postal |
Email |
Face to Face
]
The Diplomatic Pouch is brought to you by
the DP Council.
The Email Diplomacy section is maintained by Ted Miller.