This is an idea developed by Dean Bedlington of Olympian Games in Australia about 4 years ago for a fun way of awarding a prize he sponsored. Dean has lost interest in it in recent times but it is still occasionally used. Dean used to have an explanation of it on the web but it seems to have disappeared now. I found an old copy I had and have placed it in the files in TopDog.zip.
As somebody pointed out, the Top Dog handicap doesn't use actual Glicko ratings but takes advantage of the data collected and published in the DBM Glicko websites. They use average score per game, usually calculated on 0 to 10 basis. After a while I added optional computer calculated columns to some of the Glicko pages showing player and army top dog handicap components. That was so that Dean wasn't trying to work them out manually mid-tournament and players could accurately see beforehand what their own and, particularly, their available army choices' handicaps were. It also made it easier for Top Dog to be used elsewhere (Kevin Donovan was interested from early on I think).
As has been pointed out, the total Top Dog handicap is 2/3 of the player handicap plus 1/3 of the army (Dean decided on these out after doing some analysis on the predictive ability of different proportions).
Generally the player handicap is equal to their historic average points per game but there is a discount for newer players to especially encourage them (bearing in mind that they are often realistically not in the running for other prizes), thus (where P is player average points per game under 0-10):
No. of rated games Handicap
0 1.00
1 to 7 max( 1.00, P * 0.3333)
8 to 15 max( 1.00, P * 0.6667)
16 P
There is more explanation in the uploaded file (including the minimum of 1)The army handicap is based on the army's average score per game but with discounts for rarely used armies and penalties for over-used ones to encourage diversity of army usage, thus (where A is army average points per game under 0-10)::
No. of games for army Handicap
0 0.00
1 to 7 A * 0.25
8 to 15 A * 0.50
16 to 23 A * 0.75
24 to 99 A
100 to 199 A + 1
200 to 299 A + 2
300 to 399 A + 3
etcSo if I have played 12 rated games and have an average score of 4.50 per game and choose to use a virgin army, my handicap would be:
Player : 4.50 * 0.66667 = 3.00 Army: 0.00 Total: 2/3 * 3.00 + 1/3 * 0.00 = 2.00 per game
Which is pretty low and sees me in with a good chance of claiming the prize if I play well.
If however I was an experienced player of 50 games with an average score of 6 and used a popular army like LIR that had 250- games up and an average score of 5.5, my handicap would be:
Total = 2/3 * 6 + 1/3 *( 5.5 + 2) = 6.5 per gameExperienced average and better players sometimes took virgin or almost never used armies to a comp in the hope of picking up a Top Dog prize. The trick was to find such an army that wasn't actually a dog. I tried it with Mamlukes at Cancon a couple of years ago and got close (a new player did reasonably well and got the prize). Before that with Khwarizmians at MOAB with much less success.
Note the Glicko website Top Dog columns include all of the adjustments for number of games for players and armies.
You can use Top Dog with scoring other than 0-10. When I have used it with 0-32 scores, I scaled total handicaps up using the formula (3H + 1) where H is the 0-10 handicap. I have never done it with GCS15 but for simplicity you could just use 1.5H (imperfect but consistent with overall scores and results) or, as I think Bruce suggested, use the 0-10 handicap and scores.