doubles in Castles of burgundy

[mizzou95]

oh, i'm chilled out. i actually am proud of the fact that my double rate is 1 in 750000 chance of happening and i'm still at a 300 ETO. I'm at the point where it is actually a game now to see how high my doubles rate can go in a game that clearly is a disadvantage to rolling doubles, while maintaining or increasing my ELO. Fun! BTW, my rate since my last post has gone to 17.34%, which take 3900 games X 25 rolls a game is approaching 90000 rolls. The chance of rolling 17.34% or more doubles on 90,000 rolls of two standard six-sided dice is approximately 3.56 × 10⁻⁸, which translates to about 1 in 28,089,887 (or a 0.00000356% chance). Because the sample size is incredibly large, the results tightly cluster around the statistical average, making a deviation this high exceptionally rare.

I feel like you skipped a lot of the discussion on this post.

The statistics only cover the last 500 games, so you got 17.34% doubles on 12,500 rolls, not 90k. And the odds of that are 2.3% (not 1 in 750k) for any given player, of which there are thousands and ergo this is pretty much a certainty for somebody. It happens to be you.

Last 500 games huh? I'm surprised that BGA would do this - it makes zero sense and makes counting any stats irrelevant. In fact, for a board gamer community I would have to say I'm very skeptical still. Most board gamers understand stats and realize counting 500 games is practically useless for a good sample size. I mean, you guys have said that yourself - "chill out it's only 500 games." I mean, does it take much more bandwidth to just keep counting all the stats? Asking for a friend.

[FrankJones]

Wow, I was trying to be funny and it went straight over your head. Just forget it dude... I mean, this means absolutely nothing in the grand scheme of things, and you are obviously butt hurt. Sorry I blew your actuarial mind and you went off writing 6 or 8 posts. BTW, if the stats on BGA are only for 500 games, then by your reasoning - WHAT WOULD BE THE POINT OF LISTING THE STATS AT ALL, AS THEY WOULD REALLY MEAN NOTHING. (I'm not convinced it's only the last 500 games, but someone on the internet said so - so it must be right... Right?

BarnardsStar actually did all the work for you, and the actual probability of what happened during your last 500 games is approximately 1 out of 43.

If/when we get to the "1 in 28 million" you keep talking about, then we can resume the conversation.

If you were trying to be funny and the joke missed, maybe your style of humor doesn't translate well in written text on the internet. I know mine sometimes doesn't.

Here is why the stats calculate only the last 500 games:

If a player has played 5000 games, those stats would be badly misleading for a player who has dramatically shifted his play-style.

A prefect example: In Ark Nova, upgrading build used to be the meta. A top-10 player had stats showing "build upgraded" in the 80% ballpark. Maybe higher.

Then the meta shifted, and most top players avoid upgrading build most of the time. This same player went on to upgrade build only 20% of the time. His "lifetime" stats reflect this shift in his play style. But if the stats were actually always "full lifetime", these stats would never reflect such changes, and would be heavily weighted toward how the person played in his first few thousand games.

So, to recap: If a player has played a couple thousand games, and has a "build grade" of 80%, and then 500 games later has a "build upgrade" of 20%, you can see how that's mathematically impossible (by a lot), unless those stats are only showing the most recent X number of games (in this case, 500).

Should BGA make this clearer? Yes.

Would it be a good idea to show actual lifetime stats AND in a separate column, "most recent 500 games" stats? Yes.

I'm not "butthurt", I just care about BGA, and unfounded accusations of dice rigging are potentially harmful to BGA.

My posts (and others' similar posts) here are informative for people who actually believe this nonsense.

My apologies for not realizing you were joking around.

[mizzou95]

Wow, I was trying to be funny and it went straight over your head. Just forget it dude... I mean, this means absolutely nothing in the grand scheme of things, and you are obviously butt hurt. Sorry I blew your actuarial mind and you went off writing 6 or 8 posts. BTW, if the stats on BGA are only for 500 games, then by your reasoning - WHAT WOULD BE THE POINT OF LISTING THE STATS AT ALL, AS THEY WOULD REALLY MEAN NOTHING. (I'm not convinced it's only the last 500 games, but someone on the internet said so - so it must be right... Right?

BarnardsStar actually did all the work for you, and the actual probability of what happened during your last 500 games is approximately 1 out of 43.

If/when we get to the "1 in 28 million" you keep talking about, then we can resume the conversation.

If you were trying to be funny and the joke missed, maybe your style of humor doesn't translate well in written text on the internet. I know mine sometimes doesn't.

Here is why the stats calculate only the last 500 games:

If ap layer has played 5000 games, those stats would be badly misleading for a player who has dramatically shifted his play-style.

A prefect example: In Ark Nova, upgrading build used to be the meta. A top-10 player had stats showing "build upgraded" in the 80% ballpark. Maybe higher.

Then the meta shifted, and most top players avoid upgrading build most of the time. This same player went on to upgrade build only 20% of the time. His "lifetime" stats reflect this shift in his play style. But if the stats were actually always "full lifetime", these stats would never reflect such changes, and would be heavily weighted toward how the person played in his first few thousand games.

So, to recap: If a player has played a couple thousand games, and has a "build grade" of 80%, and then 500 games later has a "build upgrade" of 20%, you can see how that's mathematically impossible (by a lot), unless those stats are only showing the most recent X number of games (in this case, 500).

Should BGA make this clearer? Yes.

Would it be a good idea to show actual lifetime stats AND in a separate column, "most recent 500 games" stats? Yes.

I'm not "butthurt", I just care about BGA, and unfounded accusations of dice rigging are potentially harmful to BGA.

My posts (and others' similar posts) here are informative for people who actually believe this nonsense.

My apologies for not realizing you were joking around.

ok, have a great day!

[FrankJones]

Last 500 games huh? I'm surprised that BGA would do this - it makes zero sense and makes counting any stats irrelevant. In fact, for a board gamer community I would have to say I'm very skeptical still. Most board gamers understand stats and realize counting 500 games is practically useless for a good sample size. I mean, you guys have said that yourself - "chill out it's only 500 games." I mean, does it take much more bandwidth to just keep counting all the stats? Asking for a friend.

The issue is that BGA does not make this clear.

See my above post for an explanation of why true lifetime stats can also be useless and misleading.

Short answer: What a person did over his last 500 games is more informative than what he did in his first 2000 games. People adapt, improve, learn, and the overall game meta shifts.

As for bandwidth, I'm not sure. Maybe when a player clicks "view stats", the function has to tediously (relatively speaking, for a computer program) count each stat. I hope not; I would assume the stats are just updated to a database after each game, and then it' just a matter of grabbing the value from each entry in the DB.

Also, don't forget the other issue, some stats are not logged properly for abandoned or conceded games.

That might be the most relevant and important fact in this entire topic.

For example, I just completed a game in which the end game stats show I gained 17 seeds, when in fact, I gained 9. That's a HUGE error.

[tbhp]

If a player has played 5000 games, those stats would be badly misleading for a player who has dramatically shifted his play-style.

That's why I think they should show both all-time stats and narrowed down stats where you can manually pick the time window.

Now, I don't know if they've kept all the stats from since the site exists, or if the older stats are just gone. In the latter case, all-time would be impossible to provide, except by starting from this point in time.

It would be very interesting to see how your stats have evolved over time, instead of them being limited to just your current playstyle (or current meta when looking at the general stats).

[Jellby]

Last 500 games huh? I'm surprised that BGA would do this

You wouldn't be surprised if you knew that the "all time" ranking includes only those who have played in the last 2 months, or that "friends" are not necessarily people you know, or that you can be a "guru" for a game you have only played once (and lost), or that you can get a "5 victories" achievement in a game without ever getting the "1st victory" achievement.

it makes zero sense and makes counting any stats irrelevant.

It makes a lot of sense from a programming point of view, if you want to limit the use of resources.
But yes, stats are mostly irrelevant anyway.

[Meeplelowda]

or that you can be a "guru" for a game you have only played once (and lost)

I feel seen.

[tbhp]

It makes a lot of sense from a programming point of view, if you want to limit the use of resources.

That doesn't make much sense, saving all this data probably takes up less storage space than saving one high quality picture.

[FrankJones]

Last 500 games huh? I'm surprised that BGA would do this

You wouldn't be surprised if you knew that the "all time" ranking includes only those who have played in the last 2 months, or that "friends" are not necessarily people you know, or that you can be a "guru" for a game you have only played once (and lost), or that you can get a "5 victories" achievement in a game without ever getting the "1st victory" achievement.

it makes zero sense and makes counting any stats irrelevant.

It makes a lot of sense from a programming point of view, if you want to limit the use of resources.
But yes, stats are mostly irrelevant anyway.

That's for sure!

I just won a game and simultaneously got 3 achievements: "First win", "Big bang", and "never 2 without 3".

[Paddles]

It makes a lot of sense from a programming point of view, if you want to limit the use of resources.

That doesn't make much sense, saving all this data probably takes up less storage space than saving one high quality picture.

It's not the storing of the data that's the issue, but the amount of time required to calculate the various figures it every time someone wants to see the stats - because each time the database has to go back, find the hundreds or thousands of rows of data, and add up all the numbers again. Alternatively, if the long-term numbers are being precalculated and stored, then you've got hundreds of thousands of calculations that have to be recalculated and stored on a regular basis. All while new data is being added constantly.

There's an adage in IT - it takes a lot of work to make something look simple.