Scrabble Odds
Joey's Index of Scrabble Mistakes
I have handicapping systems for everything of relevance, and Scrabble is no exception.
Using a grading scale where large mistakes count as 1.0, medium mistakes counts as 0.5, and small mistakes count as 0.2, we can reasonably assess how well a person is playing in very plain, relative terms. Using JISM, we can comfortably assess the odds someone will win a tournament, or singular game.
The overarching thing that defines a player's strength is "What percentage of the time is this person making the correct play?" While assessing positions in Scrabble is often too complex to "know for sure what the right play is", that's beyond the point here, as we just need to get into a narrow range to approximate strengths. Over a reasonably sized sample of annotated game analysis, we can figure out what percentage of the time the player is making the correct play, and then after that we consider some raw average scores and performance rating data to further approximate the player's strength.
There are other factors that potentially go in to a player's relative strength being calculated. For example, someone like Nigel Richards has an additional edge on human opponents in tournaments because he executes everything very QUICKLY, which subsequently forces more opponent ERRORS, because they will be in time pressure that much more often.
A central system that tracks player ELOs must be maintained in order to accurately predict results. With now very many organizations and lexica, it is one large clusterfuck, so any given rating a player has on paper is liable to be wholly inaccurate.
Here is the template that shows the different strengths of relevant players:
2300+ - Perfect Player (does not technically exist, upper limit unknown)
2270 - Nigel Richards (the greatest Scrabble human of all time)
2240 - BestBot (the best Scrabble computer engine, at woogles.io)
Here are the mistakes/game estimates for each corresponding level of play.
2300 - 0.0
2250 - 0.2
2200 - 0.5
2150 - 0.8
2100 - 1.2
2050 - 1.5
2000 - 1.7
1950 - 1.9
1900 - 2.3
1850 - 2.6
1800 - 2.9
1750 - 3.3
1700 - 3.8
1650 - 4.2
Individual Game Odds Template
Here is a sample of player matchups that show the single game odds for the different skill gaps.....there is a point spread followed by the win percentage, and then the money line (odds to win the game, including rake, using betting notation).
2150 v :
2140 (10 point gap)
-2.5 51%
ML: -112 / -108
2100 (50 point gap)
-8.5 53%
ML: -115 / -105
2050 (100 point gap)
-16.5 56.5%
ML: -130 / +110
2000 (150 point gap)
-26.5 60%
ML: -155 / +130
1950 (200 point gap)
-35.5 66%
ML: -210 / +185
1900 (250 point gap)
-42.5 70%
ML: -255 / +220
1850 (300 point gap)
-51.5 74%
ML: -310 / +250
1750 (400 point gap)
-68.5 80%
ML: -450 / +350
1650 (500 point gap)
-80.5 85%
ML: -650 / +500
1550 (600 point gap)
-93.5 89%
ML: -1050 / +750
1400 (750 point gap)
-107.5 93%
ML: -1400 / +1000
1300 (850 point gap)
-121.5 95%
ML: -2000 / +1300
1200 (950 point gap)
-132.5 97%
ML: -4000 / +2500
1100 (1050 point gap)
-143.5 98%
ML: -6000 / +3000
Individual Game Point Totals (NWL)
Here is the template of player matchups that show the corresponding single game over/unders for total points scored for the game, and also by each player:
2150 v :
2140 (10 point gap)
862.5
432.5 / 429.5
2100 (50 point gap)
858.5
433.5 / 424.5
2050 (100 point gap)
854.5
435.5 / 419.5
2000 (150 point gap)
850.5
438.5 / 412.5
1950 (200 point gap)
847.5
441.5 / 405.5
1900 (250 point gap)
844.5
443.5 / 400.5
1850 (300 point gap)
839.5
445.5 / 393.5
1750 (400 point gap)
833.5
451.5 / 381.5
1650 (500 point gap)
829.5
454.5 / 374.5
1550 (600 point gap)
822.5
457.5 / 364.5
1400 (750 point gap)
814.5
460.5 / 353.5