Forum Archive :
Probability and Statistics
Distribution of points per game
|
I asked Jellyfish to play 1000 moneygames at level 5 (both sides,
using the 'Finnish game' command) and recorded the results.
This is the distribution of points per game:
1 point: 35.4%
2 points: 40.7%
4 points: 20.3%
6 points: 0.6%
8 points: 2.5%
16 points: 0.5%
Sum : 100.0% (Average points per game: 2.296)
(Sample Variance : 8.334)
The Jacoby rule was not in use, but the effect in ppp is small.
(It'll only affect those games where the equity swings from
less than .5 to greater than 1 in one sequence, making the
average ppp smaller).
An example of how this information can be used:
If you've played n games in a chouette with P others, then you
can be 96% sure that your winnings was due to skill, and not luck
if you won more than
5*sqr(nP) points.
(The Central Limit Theorem is used. n should be more than 12).
Thanks for reading.
Stig Eide, stig.eide@avh.unit.no
|
|
Kit Woolsey writes:
The most interesting result of this trial to me is that the % of 1 point
games was 35.4%. This means that Jellyfish thinks that with proper play,
64.6% of initial cubes should be taken. I have always believed this --
in fact, I think the proper figure might even be higher. However, in a
survey of over 1000 matches (most of which were played in the 1980's,
most involving at least one world-class player), the percent of initial
doubles taken was only 40%. I only looked at match scores where the
score figured to be irrelevant to initial cube action -- i.e. both
players had several points to go. I wonder who is doing the right thing
-- jellyfish, or these players.
Kit
|
|
|
|
Probability and Statistics
- Average game and match length (JP White, Dec 2000)
- Average luck of each roll (Timothy Chow+, Mar 2013)
- Average luck of each roll (Jørn Thyssen+, Feb 2004)
- Calculating winning chances (Douglas Zare, June 2000)
- Chance of rolling x doubles in y rolls (Raccoon+, July 2007)
- Chance of rolling x or more pips in y rolls (Tom Keith, Feb 2004)
- Clumping of random numbers (Gary Wong, Sept 1998)
- Counting shots (Koyunbaba+, June 2007)
- Counting shots (John Little+, Mar 2007)
- Distribution of points per game (Roland Sutter, June 1999)
- Distribution of points per game (Stig Eide+, Sept 1995)
- Expected variation in points after a series of games (Achim Müller+, Feb 1999)
- How many games to decide who's better? (Stephen Turner, Mar 1997)
- How often is too often? (Gary Wong, Oct 1998)
- Losing after bearing off 14 checkers (Daniel Murphy, July 1999)
- Number of games per match (Jason Lee+, Jan 2005)
- Number of rolls to enter x checkers from bar (Michael Depreli+, Mar 2011)
- Visualizing odds (Daithi, Mar 2011)
From GammOnLine
Long message
Recommended reading
Recent addition
|
| |
|