The problem is finding a fixed point to study. How do I systematically isolate critical points in a game? Do I go by centre count, any unit passing the stalemate line, year, countries defeated? I'd love to jump into that mess of factors. This is an early prototype with a convenient fixed point, mostly written as an exercise in SQL and php, to form a programming blueprint for later, more useful research. That's what usually happens when I throw some random data in here. I had an idea and saw if I could write a program. I'm learning and refining my methods.

If anyone's interested, here are centre ownership for each country in A1901 and what it lead to:

http://draugnar.com/VaftStats/growingisgood.html

I reiterate my main conclusion: Having SCs is good. Not having SCs is bad. I believe this is a minor revolution in our understanding of the game.

It goes back to what i was saying a while ago. Its really difficult at best to trace a path to the solo as a trend for any country. What you end up with is a series of paths that vary quite differently from one another. For example, Englan might start out taking norway in 01 and move on to France then Italy and Austria/Germany or he might start at Germany and go to Russia and dip into Austria never really touching France.

How does england get holland in 533 A1901s ?

The difference between france taking belguim and not taking bel (while also taking spain and portugal) is 10% of solos.

That is more than double the win rate when bel falls to france in A1901, which shoild send a signal to G/E players, never give france bel in '01

Any solution which gives france venice deserves a pat on the back (well that italy deserves a pat on the head...)

Italy does VERY poorly with just tunis (the safe move) and is much improved by taking trieste or trieste AND greece... This jumps from 5 to 10 to over 20... pulling it off however may be a bit of a task...

Belguim or not belguim plays a similar role in german performance - with a 9 to 17 % gain... And it appears only moves which take sweden or warsaw perform better. Note to germans, stop playing so Pro-Russian.

Austrian stats are rather poor for the weakest nation... Serbia and greece or serbia amd rumania are the strongest regular openings, and the difference of not managing a second build seems massive.

Not a good sign for the Archduke...

Similarily it doesn't seem to matter where turkey get his second build, after bul greece rum or sev will do. In any case improvig the solo rates, notably three centers does even better.

Russia, in contrast have a steady run of games with no builds, and loses the vast majority of them. Blocking either rumania or sweden cuts the russian solo rate in three from 18 to 6, definitely a consideration for most neighbours... And for russia taking the third center in '01 really makes a huge difference, with norway or vienna (along with sweden and rum) netting massive solo rates.

One must also keep in mind, of course, that some of these could just be correlation. For example, taking Belgium as Germany may be due usually to more active or aggressive diplomacy, and the resulting higher performance may really be coming from that.

An interesting thing to control for would be the average message count of the people playing various powers. (Obviously, you could only get their average across all games, but that might be a good enough indicator).

And yes as smeck says, these all correlate with wins because if you can pull off the stronger opening you are probably a stronger player.

But i still think this could be interesting for new players to get an idea from looking at their first winter to see how likely each country is to win (as an indication of how strong that player is)

Agreed Semck, seeing if there was a messages/year vs win rate correlation would be interesting. I imagine there is.

orathaic, Tru Ninja has made a model for this, check his user page for functions that correlate messages and success. Perhaps I can integrate this. It would take some creativity.

Tru looked at messages/game which is inherently skewed by doing well and surviving longer in the game. More appropriate, IMHO, would be to show messages/year.

That's something I can't do. My database dump is stripped of in-game messages. That's private data and I have no business there.

Fair. Any idea how true got that though?

Any chance kestas would give you a dump with messages reduced to zeros? (so you would have message lenght of every single message, but nothing about the content)