I don't get how Germany wins 16% in gunboat...

vaft, see my pm

Gobbledook, 16% is only a 3% increase on the press games.

I think a weak russia helps germany a lot.

Don't worry vaft, SQL is a cinch and wonderfully elegant, by Cod!

I'm finding the stats you'vew done already very interesting - especailly how much france and turkey benefit from no press - not really a fair game then!

Yonni, just because it was easier than most stuff people requested, I wrote a quick piece of code to check that:

2006: 100% wins, 74 wins in 74 games
2007: 98.548297040759% wins, 1765 wins in 1791 games
2008: 85.374554102259% wins, 2872 wins in 3364 games
2009: 68.593955321945% wins, 3654 wins in 5327 games
2010: 58.337818925611% wins, 5419 wins in 9289 games
2011: 55.420271280572% wins, 5189 wins in 9363 games
2012: 51.921627731726% wins, 1378 wins in 2654 games

Although I'm not even sure when the draw function was added :D

But yes, we have become soft.

Fine, so those old farts do have something to go by when they complain about the prevalence of draws. However, as you said, it may just be a function of the draw vote being added.

clearly the draw function was added at some point in 2007, which still means you have 85% in 2008 reducing to the current 51%, that is pretty surprising. Though i would argue that people are playing better, and strong opposition makes it harder to earn a solo.

Going by the database, the draw function seems to have been added in September 2007. Does anyone know if there is a changelog for functionality implemented here? I've tried looking at the code of the different releases, but I can't find one. The point system seems to have been added in August 2007, going by a kestas post on the dev forum. I just want to know when the site became close enough to what it is today to make a comparison meaningful. Clearly a game without a draw function and without points can't be compared to one played today.

The draw function was by mod request for a long time before voting was introduced.

I was looking through the spreadsheet, and I noticed that Russia has played one less game than all of the other countries. That seems a bit odd.

@Ora: That's not what I was noting. It's a matter of sample, not the numbers themselves. It's sort of like if you perform a study on vehicle safety looking at Hondas versus Hondas that are less than 5 years old. Either case, your results sum to 100%, but the sample is different.

I'm saying that Vaft's sample contains games that give inaccurate results. For example, games that have multiaccounters tend to have the countries played by multis do better than those that arent if the multi's aren't detected early. If they are and they're banned in the middle of the game, some countries had an unfair disadvantage and got eliminated before the ban while the remaining countries received an unfair advantage after the ban with easy to obtain centers.

I've found that quite a large amount of games have unbalanced results and throw the data off.

But any multis are randomly assigned, so this bias should be evenly distributed across games.

Also i think handling a CD can be considered a part of diplomacy. It happens in f2f and there are rules to cover it. Again which country goes CD is probably evenly distribured (though weaker countries may get wiped out sooner and CD with one/two SC left - however those CDs have the smallest/most neglidible impact)

This could be filtered from the data to see how it contrasts and to see if my assumptions have merit.

Granted, while multis are randomly assigned, the effects of having multis in a game can be rather significant. When CD's are rampant, they shape games. Players win with 3 centers, countries have unfair advantages or disadvantages, games that might end in a draw might end in a win instead, or might have unfair two-way draws or other things of that nature.

Simply because a "bias" might be assumed to be evenly distributed doesn't mean that it doesn't shape the data in some significant fashion.

That's the reason I calculate all of my statistics by hand. Sure it takes a while but you get a more true feel for how the nature of the game is.

Also, when dealing with CD's, they shape games far less after the first year if the country is small. The reason is as follows: If a country CD's within the first year, even if someone takes over the country, there are often alliances that have already formed making it more difficult for the fill-in to do as well as if they had started the game out fresh. In addition, it gives countries an unfair advantage early on, especially where neutral centers are concerned.
If a CD occurs after the first year and the country is small (a few centers since the player quits assuming they have no chance left) then its a result of alliances that have already worn down the country and effects will be minimal (the victorious countries will take the centers a bit quicker than normal). However, if the country that CD's is larger, it gives major advantages to their neighbors and often the game ends with the CD'd country having listed as a "resign" with a few centers left. These games skew data heavily if they're not thrown out.

More stats required ....

Vaftrudner: I haven't been following the thread, but you do know that your 'standard' games probably include many no press (gunboat) or public press games from before those game types were created, right?

JECE, I do. This is an alpha version. There are a few issues to work out before I can produce quality data, but right now, I'm keeping it off the forums :) I'll get back to you once things have been worked out.

A once in a lifetime opportunity! Help me test some new code! Just PM me your e-mail and I will make on of these for you:

http://i.imgur.com/pU6MQ.png

I'll customise it (although it currently only does standard games). All I ask in return is that you check if it seems to be correct so I don't have to do it myself :)

By customise I mean that I can remove any of the tables or categories (you may not want to see all your resigns!), and I should add that it is only correct up to the beginning of April (the 8th I think), but still!

Bump for volunteers!

I've added some more data now.

http://i.imgur.com/ExJPy.png

I can still make one for you if you PM me your e-mail, and I'll take any ideas for what else can be added!

Want one of these? http://i.imgur.com/ExJPy.png

PM me your e-mail and I'll send it over in html form via e-mail, it's correct up to April 8th, I can remove any tables or rows to customise it for you, and if you have any ideas for good data to add, I'll add it.