I've always thought it would be amazing to do an analysis of the webdip database. In Game of Diplomacy Richard Sharp has statistics from postal games from the 70s and I think it would be so awesome to do an up-to-date analysis with the giant database we have.

Exactly what I'm thinking. The Agar statistics from Dip Archive look to be at least 20 years old, and based on my small sample size of games from Webdip, in serious need of updating.

I've asked a couple times for a copy of the database; I can probably crunch the numbers but not without the raw data

Number crunching would be easy, it's just a matter of getting the data.

Didn't TruNinja already do this?

Yes, in a very comprehensive manner.
Here's a little sample: results by power (W-D-Surv-Elim).

R 13,9% 14,9% 23,1% 48,2%
T 12,4% 16,8% 43,8% 27,0%
F 10,7% 19,3% 36,0% 34,0%
E 9,8% 17,2% 34,7% 38,8%
G 9,8% 16,6% 26,9% 46,7%
A 6,8% 11,1% 21,3% 60,8%
I 5,0% 13,2% 34,5% 47,3%

What are these numbers for, obviously not all games on this site (too few total games), but are these for a particular set of opening moves?

Did Tru do anything with opening move breakdowns?

No:
"A game was not calculated if it met one or more of the following:
• Games that ended with 3 or more players resigning
• Games in which any country failed to submit orders at any point in the first full year
• Games that demonstrated sufficient evidence of a player controlling more than one country
• The game was a gunboat game (no messaging allowed)
• The game was a public press game (public messaging only)
• The game was a variant of the classic game
• The game ended prior to 1905
With these controls in place, games were selected from a simple random sample and collected to study games as they were completed, beginning with the newest game and working to the oldest game. Over 5,000 games were examined over the course of this study..."

@ 2ndWhiteLine: Yes, but then you should ask for the full paper.

Leif, those are the win-draw-survived-eliminated stats for each country, given the qualifiers Jesus mentioned.

I see my confusion. I've been using periods for decimal for so long I forget that a good portion of the world uses a comma.. I somehow imagined the first number was total games and that Jesus was saying Tru only found ~100 games (looking at russia) that fit that criteria, and I was in disbelief..

Oh, yes, I've copied directly from an Excel sheet which has language settings for Brazilian Portuguese. I forgot to edit the commas to periods.

no problem.. I can track with the stats now..

The opening moves breakdowns are for the most common openings. I've found problems with the opening moves listed on the Dip Pouch site. Specifically, a lot of the moves overlap each other, and with statistics, you really need clear-cut lines on where one opening begins and one ends. For example, The Diplomatic Pouch site names several opening moves based on the movement of 2 units (like German openings) and others with 3 and there's some overlap between the two groups where the same opening could be construed as either opener.

Also, sometimes they assume opening moves based on how the fall moves play out, and if you extend an "opening move" beyond the spring, then the possibilities are in the tens of thousands of possibilities.

My breakdown takes about 3 or 4 of the most common for each country based on this site with all other openings lumped under the heading of "other" which is considered 'nonstandard'. Also, the breakdowns for opening moves identify general trends on overall success, not as a percentage, but as avg number of centers controlled by the game's end and compares the confidence intervals to isolate the population mean. It's an effort to identify if one set of moves is generally better than another, or if standard moves are better than nonstandard ones.

threadID=760600

I previously posted a statistical survey of only tournament/league games.

I need to update it with some recently finished games.

It's interesting to see how different the ratio of solos/draws is in tournaments and in overall games.