Explanation of Statistics sought

[slaaty]

I thought this might be in a FAQ but I couldn't find it.

My first question was answered by looking at two different players' stats and seeing that All players average & Winners average were identical, therefore those are the averages for each of those stats over ALL games played on the site, not just those involving the player in question.

Most of the individual stats are self explanatory, but I wonder how they are resolved when games are conceded. I see winners are always credited with 0 pip count and 15 checkers borne off, so either conceded games are not counted, or there is some formulation for calculating the losers' numbers in a conceded game which compensates for the fact that the game was not finished.

Doubles average must be number of doubles in a game (does it count doubles where no play is possible?), but mine is noticeably higher, which is likely due to my style of play focused on lots of hits and creating primes, which begs the question, what is my opponents' doubles average - I suspect similar to mine. I can see that blots hit by me and by my opponent are both quite a bit higher than average, not surprising given my style of play. Also times I can't play is much higher than average, but I suspect that is very true of my opponents as well. (Implied in this paragraph is a request for opponents' doubles and opponents' unable to play stats.)

I suspect the first three lines of data are more relevant for real-time games, whereas more than half of my games are turn-based, so I just ignore those.

[dschingis27]

I would also like to know how conceded games are counted for the stats.

Number of doubles is significantly higher for you (and most likely also for your opponents) because your overall number of throws is higher, obviously because of your playstyle. Number of doubles should more or less only depend on the number of throws you do. They can depend a little a bit on tendency to concede (i.e. conceding a game is a little bit more likely after a double by opponent just happened).

[CaractacusPots]

Doubles average must be number of doubles in a game (does it count doubles where no play is possible?), but mine is noticeably higher, which is likely due to my style of play focused on lots of hits and creating primes, which begs the question, what is my opponents' doubles average - I suspect similar to mine.

Frequency of doubles has been raised by more than one player here including myself. I've had a game with 18 doubles in it, another person has had over 20 both of which I think is ridiculous. Plus yesterday I played a game in which there were 11 doubles. Opponent was given 9 of them. I got just 2. I think doubles are probably the biggest issue with the game here and the second is getting men that are on the bar, back onto the board. Too many games where opponent only has the 6 point covered I get rolled 6-6 or when there is just the 6 and 1 point covered I get 6-1, 6-6, 1-1 etc.
Monitor those instances in your games and see what you think.

ETA:

Just played my latest game.

First 5 dice throws were: 6-4, 3-3, 6-6, 3-3, 6-6 (the 9th throw was also 3-3!)

4 doubles in 5 throws !! Just a bit silly TBH

Doubles should happen on average 1 time in 6 throws

In total there were 13 doubles of which my opponent got 8 (total dice throws 55)

In addition:

5-6 came up in 3 consecutive dice throws which is a 1 in 324 event (which has happened to me many times now in 80 odd games)

1-6 came up 3 times in 5 consecutive throws

2-5 came up 3 times in 5 consecutive throws


It's random Jim, but not as we know it . . . . . .!

[dschingis27]

Sure, we should totally ignore that OPs percentage of doubles is just 17.2%, and find some sneaky dubious explanation.

[Jest Phulin]

Let's also ignore CaractacusPots's other post where he listed out the number of doubles that he got for his games, and it was within 1% if what he claims the expected value should be.

I'm starting to see "I've got a list of 50 events. Each one only has a 1% chance of happening. But it seems like I see one of them happen about every other game that I play."

[slaaty]

I am really seeking an explanation of statistics: particularly, how statistics of conceded games are handled. Please do not hijack this thread to complain about luck. Start another thread if you have to do that.

Many thanks!

[euklid314]

I am really seeking an explanation of statistics: particularly, how statistics of conceded games are handled. Please do not hijack this thread to complain about luck. Start another thread if you have to do that.

Many thanks!

Lol. I like your clear words. By coincidence I happen to know that such a thread already exists - surprised?

The statistics page is not well documented as far as I know. But I have just checked with one player that has few games:

1) Games that are conceded are not considered in your statistics. Note that conceded games have their own statistic page (e.g.: https://boardgamearena.com/table?table=111607598 ), but these numbers are NOT included in your overall statistic data.

2) Games that are played in training mode (unrated) ARE considered in the statistics as long as they are not conceded.

3) Up to my knowledge there is no statistics page for your opponents. But since you can see the statistics of your opponent directly after each single finished game, you can fill an excel sheet with all necessary data if it is really important to you. You still have access to all the old data so you can do it for your old games too.

4) I am absolutely convinced that all doubles are counted - independ of the fact if you fan on them. But you can easily check that for a single game where it happens: Just replay the game and count the doubles and compare. Since you only have to do that for a single game, it is not that much of a work.
Update: I just checked it with one of your games (https://boardgamearena.com/table?table=111592104). You have rolled 4 doubles, you could move on 3 of them, 1 of them was a fan (the 5-5). The statistic page shows all of your 4 doubles, so all doubles are given.

5) As you wrote, your playing style gives longer games than average (number of moves is given in the statistics, so it is easily verifyable) which yields larger numbers in the other statistical data.

[slaaty]

Thank you Euklid314! Your clear and thorough answer is most appreciated!

[CaractacusPots]

1) Games that are conceded are not considered in your statistics. Note that conceded games have their own statistic page (e.g.: https://boardgamearena.com/table?table=111607598 ), but these numbers are NOT included in your overall statistic data.

This probably explains the various instances I have had where I am about to win and have one piece left to bear off and opponent quickly hits the concede button.

Seems somehow wrong. Means people can skew their stats, even though conceded games have their own set of stats.

A conceded game is a lost game any way you cut it. Especially if it were a Gammon or Backgammon (though those types of win are not currently coded to provide any additional win benefit like ELO points etc).

ALL games lost should surely factor into a player's statistics.

[euklid314]

Please note that a conceded game counts as a lost game in the win/loose-statistic.
The statistic of 323 wins at 620 games (15.9.2020) of slaaty (https://boardgamearena.com/playerstat?i ... 14&game=53) include conceded wins and conceded losses.

Conceded games just do not count toward the statistics in the Profile area (number of dice throws, number of doubles,...). I doubt that many players care about their numbers there to concede a game that they would have played on otherwise...