Jump to content


Photo

Top 20 Common Game Setups at Gravon site


  • Please log in to reply
21 replies to this topic

#1 Dobby125

Dobby125

    Sergeant

  • Members
  • PipPipPipPipPipPip
  • 359 posts
  • Coat of arms
  • Silver Major

Posted 25 August 2016 - 02:30 AM

Hey everybody!

 

I did some more research on the Stratego game setups on the Gravon Stratego site.

 

Here's the link to the old thread: http://forum.strateg...s/?hl=analyzing

 

I found the most common game setups in their archives.  Here are the top 20.  The game setup listings below are how you would look at your own setup.  The number at the bottom of the gameboard grid is the number of times that game setup was played.  I used the European numbering format in the grid, except the Marshal is a 0 instead of a 10 and the Spy is a 1.  I might have to try some of these setups.  Eventually, I'll find and list the game setups that have the best winning records.  

 

2680229742
4B425618B4
3727B6325B
325BFB5363
 452
 
528972074B
2BB15226B4
B453426862
32B637F353
 356
 
242B427453
B2252B9802
43B3561837
B235B6667F
 295
 
2502257226
6834691842
4753373B26
BB3BFB5B42
 203
 
24B1766542
6253094B26
3583B38B43
227BFB7522
 197
 
2463629562
45B7251382
73B6B82022
5B4BFB3437
 187
 
22B6223722
4681765965
54053B284B
37B324B3BF
 169
 
4538629752
6707561862
BB522B22BB
4FB3342334
 158
 
62572264B2
80715986B5
232B325742
64BFB334B3
 158
 
2572622867
2B81923550
46B45B4327
3B46BFB323
 144
 
7527254625
6820682927
2234533136
B3BBF4BB4B
 134
 
4257422052
2235269374
67B383186B
24B5B63BBF
 134
 
4224B26058
6263B5933F
322B4B7681
5225B4B737
 132
 
457B42B362
2B23B63026
B358187629
45354BF722
 124
 
5459450624
2681672368
B23272372B
B3B3F2B45B
 117
 
6257526524
80212928B6
237B533742
64BBFB34B3
 112
 
522854B862
3696273225
245102743B
3BB736FBB4
 108
 
4222459375
2622B81460
B22357873B
435B636BBF
 107
 
24B1824B62
6B420275B5
B73823695B
532632F734
 104
 
2574761952
2623082245
5B4B33B728
23B46BFB36
 103
 
 

 


  • dviltje, faq and General J. Wolfe like this

My Youtube Stratego Channel: https://www.youtube....cGDvlZZkGbgq0LA


#2 GaryLShelton

GaryLShelton

    Flagbearer

  • Moderators
  • 6,494 posts
  • Coat of arms
  • Silver Colonel

Posted 25 August 2016 - 11:16 PM

Dobby125, it's nice to have you back in the forum reporting cool things. Where have you been?

i77rs4m.jpg

The complete GS&F Rules can be found here: http://forum.strateg...rum-rules-2016/

Draw Refusal Rules, specifically, can be read here: http://forum.strateg...931#entry468931


#3 Dobby125

Dobby125

    Sergeant

  • Members
  • PipPipPipPipPipPip
  • 359 posts
  • Coat of arms
  • Silver Major

Posted 27 August 2016 - 09:14 PM

Dobby125, it's nice to have you back in the forum reporting cool things. Where have you been?

 

I think I got burned out by playing so many stratego games last year.  I've been very busy at work.  In my spare time, I'm either watching youtube tutorials or MIT programming lectures.  I also like to play CIV5.  That's really a fun strategic game, but a big time killer.  When I have a stratego urge, I usually play the Probe PC game.  I like playing on the hardest setting and letting the computer play a designed setup, while I get a random setup.  My goal is to win 5 in a row.  It's hard to do.  You need lots of luck with a random setup draw.


Edited by Dobby125, 27 August 2016 - 09:15 PM.

My Youtube Stratego Channel: https://www.youtube....cGDvlZZkGbgq0LA


#4 Dobby125

Dobby125

    Sergeant

  • Members
  • PipPipPipPipPipPip
  • 359 posts
  • Coat of arms
  • Silver Major

Posted 27 August 2016 - 09:24 PM

I ran a python script on all the classic Stratego games played in the gravon archives.  I wanted to see if Player 1 wins more than Player 2.  Here were the results:

 

player 1 wins  19798
player 2 wins  21154
number of draws  1527
total games =  42479
 
player 1 wins  46.60%
player 2 wins  49.79%
draws occur 3.59%
 
It seems going 1st doesn't give you an advantage.

  • maxroelofs likes this

My Youtube Stratego Channel: https://www.youtube....cGDvlZZkGbgq0LA


#5 GaryLShelton

GaryLShelton

    Flagbearer

  • Moderators
  • 6,494 posts
  • Coat of arms
  • Silver Colonel

Posted 27 August 2016 - 09:28 PM

It's only available here, folks!


:)
  • Dobby125 likes this

i77rs4m.jpg

The complete GS&F Rules can be found here: http://forum.strateg...rum-rules-2016/

Draw Refusal Rules, specifically, can be read here: http://forum.strateg...931#entry468931


#6 Astros 17

Astros 17

    Major

  • Members
  • PipPipPipPipPipPipPipPipPip
  • 1,224 posts
  • Coat of arms
  • Platinum Scout

Posted 27 August 2016 - 09:31 PM

Yea, but did you control for player skill?
3 - 0

#7 Dobby125

Dobby125

    Sergeant

  • Members
  • PipPipPipPipPipPip
  • 359 posts
  • Coat of arms
  • Silver Major

Posted 27 August 2016 - 09:54 PM

Yea, but did you control for player skill?

 

Unfortunately, the players are anonymous.  No skill level is given.  The game files do tell you how a person wins or if it's a draw.

 

type0 = player can't move anymore (win)

type1 = player captures flag  (win)

type3 = player gives up (win)

type4 = players agree to a draw (draw)

type10= both players leave table (draw)

type11 = timeout of opponent (win)  probably means an internet disconnect.

type 12 = clock runs out draw (draw)

type 13 = clock of opponent runs out (win)  I hate losing this way!

type 14 = not sure what this type is.  they didn't give a message at the end of the games, but the winning player was usually dominating the game. (win).

 

I'll break down the games by type too.  It's only a few lines of code in Python.


My Youtube Stratego Channel: https://www.youtube....cGDvlZZkGbgq0LA


#8 Astros 17

Astros 17

    Major

  • Members
  • PipPipPipPipPipPipPipPipPip
  • 1,224 posts
  • Coat of arms
  • Platinum Scout

Posted 27 August 2016 - 09:58 PM

Unfortunately, the players are anonymous.  No skill level is given.  The game files do tell you how a person wins or if it's a draw.
 
type0 = player can't move anymore (win)
type1 = player captures flag  (win)
type3 = player gives up (win)
type4 = players agree to a draw (draw)
type10= both players leave table (draw)
type11 = timeout of opponent (win)  probably means an internet disconnect.
type 12 = clock runs out draw (draw)
type 13 = clock of opponent runs out (win)  I hate losing this way!
type 14 = not sure what this type is.  they didn't give a message at the end of the games, but the winning player was usually dominating the game. (win).
 
I'll break down the games by type too.  It's only a few lines of code in Python.


Unfortunately, because players are able to select whether they go first it is not randomly distributed. Skill level is distorting win probability of moving first.
3 - 0

#9 GaryLShelton

GaryLShelton

    Flagbearer

  • Moderators
  • 6,494 posts
  • Coat of arms
  • Silver Colonel

Posted 27 August 2016 - 10:32 PM

Is there reason to think that just because going first is an option chosen by the player that on this big site randomness would not still be achieved? Wouldn't just as many want to go second as go first? And across all skill levels? Or is it established that higher ranks prefer to go first, for example?

i77rs4m.jpg

The complete GS&F Rules can be found here: http://forum.strateg...rum-rules-2016/

Draw Refusal Rules, specifically, can be read here: http://forum.strateg...931#entry468931


#10 maxroelofs

maxroelofs

    Major

  • Members
  • PipPipPipPipPipPipPipPipPip
  • 1,160 posts
  • Coat of arms
  • Platinum Marshal

Posted 27 August 2016 - 10:36 PM

It is to say that agressive players want to go first. 

 

It is also to say that players with a surprising front row want to go first, in case you want to put a sergeant in front of your general for example.


To watch stratego videos: https://www.youtube....HOHXWONQMsVcOLA

#11 Dobby125

Dobby125

    Sergeant

  • Members
  • PipPipPipPipPipPip
  • 359 posts
  • Coat of arms
  • Silver Major

Posted 27 August 2016 - 10:57 PM

Here's the breakdown on the "types".

 

type0 - 3932    Opponent can't move

type1 - 12994  Player captures flag
type2 - 6          Some type of draw
type3 - 23412  Player gives up
type4 - 913      Both players agree to a draw
type10 -  573   Both players leave table  
type11 - 357     Timeout of opponent
type12 - 37       Clock runs out draw
type13 - 208     Clock of opponent runs out
type14 - 47       Win by player with dominant lead(i only viewed a few games with this type)
totals  42479
lines 42479
 
There was also a type 2.  It's a draw.  I'd have to watch one of the 6 games to see the message it gives at the end of the game.

My Youtube Stratego Channel: https://www.youtube....cGDvlZZkGbgq0LA


#12 Astros 17

Astros 17

    Major

  • Members
  • PipPipPipPipPipPipPipPipPip
  • 1,224 posts
  • Coat of arms
  • Platinum Scout

Posted 28 August 2016 - 12:59 AM

Is there reason to think that just because going first is an option chosen by the player that on this big site randomness would not still be achieved? Wouldn't just as many want to go second as go first? And across all skill levels? Or is it established that higher ranks prefer to go first, for example?

To correctly measure the effect of going first one would construct a probabilistic regression model. The dependent variable would be whether player 1 wins and the variable of interest is going first. If one were to perform their regression with going first as the only regressor their regression would likely be biased.

 

Omitted variable bias occurs when the omitted variable is a determinant of the dependent variable (winning) and is correlated with another regressor. The variable relative rating of player 1 to player 2 is correlated with winning; if player 1 has 600 more ELO they are more likely to win. Relative rating is plausibly correlated with going first; as Max stated, better players may prefer going first. Therefore, one needs to consider other factors.

 

Furthermore, on Gravon opponents are not randomly selected therefore it is difficult to determine to study whether going first makes a difference.


3 - 0

#13 The Prof

The Prof

    Colonel

  • Honorary members
  • PipPipPipPipPipPipPipPipPipPip
  • 1,518 posts
  • Coat of arms
  • Platinum Scout

Posted 28 August 2016 - 03:25 AM

I'm sure those who have commented are correct about it not being random who goes first at Gravon.  If it were, I highly doubt we would see such a statistically significant result that Dobby has observed.  Correct me if I'm wrong but isn't who goes first at Gravon decided by which player created the game (getting the 1st move) versus which player decided to join (getting the 2nd move)?

 

My theory then is that the advantage in win percentage for the 2nd player is due to more players choosing to join games against players ranked lower than themselves than higher, and/or that the difference in ratings is typically larger in the former case than in the latter. I saw this kind of thing at Metaforge, where there were a number of players who enjoyed winning more than being in competitive matches, so they would only join games of new or low ranked players and they would never create their own games.  If this kind of thing occurred at Gravon, even to a minor extent, it could easily produce the kind of data that Dobby presented.



#14 GaryLShelton

GaryLShelton

    Flagbearer

  • Moderators
  • 6,494 posts
  • Coat of arms
  • Silver Colonel

Posted 28 August 2016 - 03:56 AM

astros, that's mighty impressive. What kind of day job do you have? You must be a grad math or statistics student, I assume?

The site is well-populated with math-skilled people, I think. Good for us. :)

i77rs4m.jpg

The complete GS&F Rules can be found here: http://forum.strateg...rum-rules-2016/

Draw Refusal Rules, specifically, can be read here: http://forum.strateg...931#entry468931


#15 Astros 17

Astros 17

    Major

  • Members
  • PipPipPipPipPipPipPipPipPip
  • 1,224 posts
  • Coat of arms
  • Platinum Scout

Posted 28 August 2016 - 04:17 AM

I am an econ and math major hoping to be a grad student in the next year or two.


3 - 0

#16 Dobby125

Dobby125

    Sergeant

  • Members
  • PipPipPipPipPipPip
  • 359 posts
  • Coat of arms
  • Silver Major

Posted 28 August 2016 - 03:47 PM

I'm sure those who have commented are correct about it not being random who goes first at Gravon.  If it were, I highly doubt we would see such a statistically significant result that Dobby has observed.  Correct me if I'm wrong but isn't who goes first at Gravon decided by which player created the game (getting the 1st move) versus which player decided to join (getting the 2nd move)?

 

My theory then is that the advantage in win percentage for the 2nd player is due to more players choosing to join games against players ranked lower than themselves than higher, and/or that the difference in ratings is typically larger in the former case than in the latter. I saw this kind of thing at Metaforge, where there were a number of players who enjoyed winning more than being in competitive matches, so they would only join games of new or low ranked players and they would never create their own games.  If this kind of thing occurred at Gravon, even to a minor extent, it could easily produce the kind of data that Dobby presented.

 

This makes sense.  When I first ran the calculation and saw the numbers, I thought my program logic was wrong.  :)   The number of wins should have been much closer in a random pick start, especially since the data was fairly large.  I then checked the data file in Microsoft word.  You can do a search and replace for strings in word and it will give you the total number of finds.  So I just searched for winner 1, winner 2 and winner 3(draw).  The data was correct.  I never played Stratego on Gravon, so I just assumed the game starts were randomly chosen.  Oops.

 

The next 2 things I will try to find out is:  winning percentage of closed flag vs open flag games and the game setups with the best winning percentage.


My Youtube Stratego Channel: https://www.youtube....cGDvlZZkGbgq0LA


#17 TheOptician

TheOptician

    Marshal

  • Honorary members
  • PipPipPipPipPipPipPipPipPipPipPipPip
  • 3,498 posts
  • Coat of arms
  • Platinum Captain

Posted 28 August 2016 - 04:03 PM

I'd be interested in the distance in squares from the marshall to the flag.



#18 Dobby125

Dobby125

    Sergeant

  • Members
  • PipPipPipPipPipPip
  • 359 posts
  • Coat of arms
  • Silver Major

Posted 29 August 2016 - 12:58 AM

I'd be interested in the distance in squares from the marshall to the flag.

 

I found the distance the marshal has to travel to capture his own flag. So if the flag was in the bottom left corner and the marshal was in the top right corner, it would take the marshal 12 moves to get to his flag.  Just subtract 1 from the averages if you want to know how many moves to get next to a piece.

 

Average distance for Marshal to take Flag:

total game setups analyze  84956
sum of all distances  351350
average distance away  4.135670229295164
 
Average distance for General to take Flag: 4.952834408399642
 
Average distance for Spy to take Flag:  4.184201233579736
 
Average distance for General to take Spy: 2.4310937426432506
 
Average distance for Marshal to take Spy: 3.6485592542021754
 
Average distance for Marshal to take General: 3.954505861857903
 
Here's the python code:
 
def moves(x, y):
    x1 = x[0]
    x2 = x[1]
    y1 = y[0]
    y2 = y[1]
    answer = abs(x1 - y1) + abs(x2 - y2)
    return(answer)
 
 
gameboard = {0: (0, 0), 1: (0, 1), 2: (0, 2), 3: (0, 3), 4: (0, 4), 5: (0, 5), 6: (0, 6), 7: (0, 7), 8: (0, 8), 9: (0, 9), 10: (1, 0), 11: (1, 1), 12: (1, 2), 13: (1, 3), 14: (1, 4), 15: (1, 5), 16: (1, 6), 17: (1, 7), 18: (1, 8), 19: (
    1, 9), 20: (2, 0), 21: (2, 1), 22: (2, 2), 23: (2, 3), 24: (2, 4), 25: (2, 5), 26: (2, 6), 27: (2, 7), 28: (2, 8), 29: (2, 9), 30: (3, 0), 31: (3, 1), 32: (3, 2), 33: (3, 3), 34: (3, 4), 35: (3, 5), 36: (3, 6), 37: (3, 7), 38: (3, 8), 39: (3, 9)}
 
numgames = 0
sumdis = 0
with open('test.txt', 'r') as f:
    for line in f:
        numgames += 1
        flag = (line.index("M"))
        mar = (line.index("L"))
        x = gameboard[flag]
        y = gameboard[mar]
        print(x, y)
        distance = moves(x, y)
        sumdis = sumdis + distance
    print("total game setups analyze ", numgames)
    print("sum of all distances ", sumdis)
    print("average distance away ", sumdis / numgames)
 
 
Here's a link to the data file.  It takes awhile to load.  Then you can run the python script on the file.
 
M = Flag
L = Marshal
K = General
C = Spy

Edited by Dobby125, 29 August 2016 - 01:04 PM.

  • TheOptician likes this

My Youtube Stratego Channel: https://www.youtube....cGDvlZZkGbgq0LA


#19 Dobby125

Dobby125

    Sergeant

  • Members
  • PipPipPipPipPipPip
  • 359 posts
  • Coat of arms
  • Silver Major

Posted 29 August 2016 - 02:14 PM

Here's some stats on row placement of Marshal, General, Flag and Spy at the start of the game.

 

Marshal:

Placement on top row:          29696    0.349
Placement on 3rd row:          40173    0.472
Placement on 2nd row:         13389    0.157
Placement on bottom row:      1698    0.019
 
General:
Placement on top row:         27618     0.325
Placement on 3rd row:         45516     0.535
Placement on 2nd row:         10423    0.122
Placement on bottom row:      1399    0.016
 
Spy:
Placement on top row:          11136    0.131
Placement on 3rd row:         36206    0.426
Placement on 2nd row:         33567    0.395
Placement on bottom row:      4047    0.047
 
Flag:
Placement on top row:           1482      0.017
Placement on 3rd row:           1708      0.020
Placement on 2nd row:           6691     0.078
Placement on bottom row:    75075     0.883
 
So there's an 88% chance the flag is placed on the back row.  There's less than a 2% chance that a piece from the back row is a Marshal or General.   If you want to be really good at Stratego, it helps to remember not only what pieces have moved, but also where they came from at the start.  Then just play the percentages.  Unfortunately for me, my memory is not that good anymore.  :)

  • The Prof likes this

My Youtube Stratego Channel: https://www.youtube....cGDvlZZkGbgq0LA


#20 Dobby125

Dobby125

    Sergeant

  • Members
  • PipPipPipPipPipPip
  • 359 posts
  • Coat of arms
  • Silver Major

Posted 07 September 2016 - 09:30 PM

I finally calculated the winning percentages(i didn't count ties in the winning % calculation) of the most played game setups from the gravon archive classic stratego files.

 

Here's a link to the list(it might take a few seconds to load):

https://www.dropbox....ravon1.txt?dl=0

 

I only listed game setups with 10 or more games played to keep the file short.

 

Here's the top 20. I used the American Stratego number system this time on the grids.

9531992479
7B7965S3B7
8494B5896B
896BFB6858
Wins = 261 Losses = 181 Ties = 10 Games Played = 452 Winning Percentage = 59.05%

693249147B
9BBS6995B7
B768795359
89B584F868
Wins = 263 Losses = 86 Ties = 7 Games Played = 356 Winning Percentage = 75.36%

979B794768
B9969B2319
78B865S384
B986B5554F
Wins = 196 Losses = 88 Ties = 11 Games Played = 295 Winning Percentage = 69.01%

9619964995
538752S379
7468848B95
BB8BFB6B79
Wins = 114 Losses = 77 Ties = 12 Games Played = 203 Winning Percentage = 59.69%

97BS455679
5968127B95
8638B83B78
994BFB4699
Wins = 138 Losses = 55 Ties = 4 Games Played = 197 Winning Percentage = 71.50%

9758592659
76B496S839
48B5B39199
6B7BFB8784
Wins = 149 Losses = 38 Ties = 0 Games Played = 187 Winning Percentage = 79.68%

99B5998499
753S456256
67168B937B
84B897B8BF
Wins = 90 Losses = 77 Ties = 2 Games Played = 169 Winning Percentage = 53.89%

59649957B9
314S6235B6
989B896479
57BFB887B8
Wins = 130 Losses = 21 Ties = 7 Games Played = 158 Winning Percentage = 86.09%

7683592469
541465S359
BB699B99BB
7FB8879887
Wins = 97 Losses = 54 Ties = 7 Games Played = 158 Winning Percentage = 64.24%

9649599354
9B3S298661
75B76B7894
8B75BFB898
Wins = 99 Losses = 39 Ties = 6 Games Played = 144 Winning Percentage = 71.74%

7964799169
9986952847
54B838S35B
97B6B58BBF
Wins = 65 Losses = 64 Ties = 5 Games Played = 134 Winning Percentage = 50.39%

4694967596
5391539294
9987688S85
B8BBF7BB7B
Wins = 62 Losses = 65 Ties = 7 Games Played = 134 Winning Percentage = 48.82%

7997B95163
5958B6288F
899B7B453S
6996B7B484
Wins = 68 Losses = 59 Ties = 5 Games Played = 132 Winning Percentage = 53.54%

764B79B859
9B98B58195
B863S34592
76867BF499
Wins = 76 Losses = 47 Ties = 1 Games Played = 124 Winning Percentage = 61.79%

6762761597
953S549853
B98949849B
B8B8F9B76B
Wins = 49 Losses = 65 Ties = 3 Games Played = 117 Winning Percentage = 42.98%

5964695697
319S9293B5
984B688479
57BBFB87B8
Wins = 80 Losses = 25 Ties = 7 Games Played = 112 Winning Percentage = 76.19%

699367B359
8525948996
976S19478B
8BB485FBB7
Wins = 85 Losses = 23 Ties = 0 Games Played = 108 Winning Percentage = 78.70%

7999762846
9599B3S751
B99864348B
786B585BBF
Wins = 58 Losses = 47 Ties = 2 Games Played = 107 Winning Percentage = 55.24%

97BS397B59
5B791946B6
B48398526B
689589F487
Wins = 74 Losses = 25 Ties = 5 Games Played = 104 Winning Percentage = 74.75%

964745S269
9598139976
6B7B88B493
98B75BFB85
Wins = 66 Losses = 30 Ties = 7 Games Played = 103 Winning Percentage = 68.75%
 


Edited by Dobby125, 07 September 2016 - 09:37 PM.

  • General J. Wolfe likes this

My Youtube Stratego Channel: https://www.youtube....cGDvlZZkGbgq0LA





0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users