"I determine sucess as a better result for those involved had the alliance not happened"
This determinant is begging the question. Consider: Country A can only be in an alliance with another power, or in no alliances at all. Discounting no alliances at all, which is out of the scope of the question, Country A is choosing between alliances with different powers. If Alliance 1 does not happen (Country A x Country B), then Alliance 2 will happen (Country A x Country C); if neither Alliances 1 nor 2 happen then Alliance 3 will happen (Country A x Country D); and so forth.
Therefore, we are supposed to determine which alliances are best by looking at... which alliances of alliance 1, 2, 3, etc. are best. We're being asked to pick the most successful alliances, with the determining criteria of success being "the best result." But we're being asked to pick the best result anyhow.
"@ Eden, one can pose the question like this (slightly statistical, I'm afraid). In all games where a given alliance is played at least for a couple of years, what is the average result of both players involved? In other words, how much does playing this alliance increase the chances that one of the participants wins the game?"
Eh. Good idea in theory, but I'm seeing issues. An EF can be played until Germany is quickly KOed and both powers receive ample spoils from the German lands, and then France stabs England and wipes him out en route to the win. France won, England died, which would indicate a neutral result. But it clearly had benefits, because both of them profited from the short-term gains in Germany.
All of THAT said, this criteria could be used for long-term alliances. Longer than a couple years. I would expect pairings like France/Italy, Russia/Italy, Germany/Austria, France/Russia, England/Turkey, and Germany/Austria/Italy to fit well here.
)
Jus soli
)
)
)
)