To me, a scientific result is one that is obtained by following a scientific approach, so I think it would be scientific done the right way (regardless of the result), but anyway I get your point

_________________
Uti possidetis, ita possideatis.
May your beer be laid under an enchantment of surpassing excellence for seven years!
An online epic force creator: Armyforge

I will respectfully disagree. This is the kind of multi-formula, linear algebra statistical analysis that actuaries get paid to do.

The Florida Marlins (I think it was) changed American baseball when they won the World Series with the lowest payroll in professional baseball. They analyzed how each stat contributed to the winning of games and how each stat factored into the cost of player salaries. Then they went out and recruited players with the best total cost/benefit ratio and won the championship.

This could be done. I don't have the statistical skills to do it, but it is feasible.

But that's coming from a significant operational pool. ~30 teams, playing ~160 games a season for over a hundred years, all playing each other*, with detailed records, and with essentially the same skillset (ie, they're all playing the same army). Not including all the minor league games that were also likely factored into the analysis. And not taking into account the possibility that randomness or other factors was an influencing factor in the Marlins World Series victory.

* Most Epic players play relatively internally, due to diverse location. And as such, skews data fairly heavily. Marines don't have close to a 50% win rate in our metagame, for a variety of reasons.

We get similar numbers for this project, and the formula is robust enough, yes, it can be feasible. But if it's not, it's potentially worse than not having done it.

Morgan Vening

Well I have 1430 games in the database, but I can only guarantee lists for 952 of those games. Should be able to add another 50 games after 21st too.

I also have 608 where the objectives count has been recorded too.

It's in mySQL.

I actually agree. Have at it, folks! Somebody PM me when the numbers are crunched.

_________________
author of Syncing Forward and other stories...

It's a dog-eat-dog world, and I've got my Milkbone underwear on.

42

But apparently this forum won't allow the perfect answer!

It's impractical and not very feasible in my opinion.

Even the most sophisticated form of quantitative analysis cannot take account of all the variables associated with every individual battle to provide a comprehensive evaluation the performance or effectiveness of every formation. The conclusions reached would simply fall apart when assessing their reliability and validity, and that's even before you run into problems with the sample size.

Maths can work out averages, in terms of how many casualties a unit is likely to inflict, and the probability of units surviviing, hitting, winning assaults, and so on, but it cannot rank formations with any degree of confidence, because of the issue of variables I have alluded to above.

In short, I think that there are far more useful things which experienced members of this forum could put their minds to, and I would urge nobody to waste their time doing this, as it really isn't at all useful for inexperienced players of the current rules, such as myself, in my opinion.

_________________

It also runs the risk of producing 'results' that get used by people with limited understanding of stat-math in arguments.

I've heard it recommended for power-list-builders to keep detailed battle reports so after a bit they can check how units are performing, not relying on gutfeelings and vauge memories of improbable results.

We might be able to get statistical significance on some/most units, but since EPIC players cannot all be relied upon to be power gamers, the data is too skewed for results. The most succsefull players tend to experiment with 'weaker' units after a while anyway...

maybe if we trained up a few SAI's and had them Vassal each other a trilion times?