Current top4 Regionals Finishers vs Regionals Attendance

By Baltanok, in Star Wars: Armada

because @stonestokes asked nicely:

Fleet Descriptor ( Top 4 finishes / All finishes)

Any Fleet (24/103)

Imperial (10/57)

Rebel (14/46)

Ackbar (3/15)

Cracken (1/3)

Dodonna (3/6)

Garm (1/3)

Leia (0/2)

Madine (1/4)

Mothma (2/6)

Rieekan (3/5)

Sato (0/2)

Jerry (5/15)

Motti (2/17)

Ozzel (0/1)

Sloane (3/18)

Vader (0/6)

Excellent! Thanks @Baltanok!

A good way to compare them is to employ the rule of succession. <--- LINK

Here is a table of the results ordered by their performance:

26111912_10156105236697608_3204303011771

The Laplace-Bayes' Estimator is the best mathematical guess as to the probability of finishing in the top 4 with a given admiral, all else being equal.

Edited by stonestokes

I think everybody can get on board the Nerf Garm train.

Just when I've settled on Madine, you're telling me Vader is 0 for 6?!

7 hours ago, stonestokes said:

Excellent! Thanks @Baltanok!

A good way to compare them is to employ the rule of succession. <--- LINK

Here is a table of the results ordered by their performance:

The Laplace-Bayes' Estimator is the best mathematical guess as to the probability of finishing in the top 4 with a given admiral, all else being equal.

Neat! How would you determine the confidence level for the LBE?

I think it's interesting that all of the Rebel Commanders are considered viable for tournaments while nearly half of the Imperial Commanders don't even get fielded at all. Poor Tarkin, Screed, Konstantine, and Tagge.

Judging by how good the LBE looks for Ozzel, I say that Tarkin is clearly overpowered since no one took him.

Edited by RobertK

I think instead of just ranking the commanders by LBE you need to compare their LBE to 23%. That is the chance to just make the top 4 unweighted by commander. That Ozzel, Leia, and Sato all outperform this benchmark without even making a top 4 at all makes me question the analysis a bit.

Edited by RobertK
7 minutes ago, RobertK said:

I think instead of just ranking the commanders by LBE you need to compare their LBE to 23%. That is the chance to just make the top 4 unweighted by commander. That Ozzel, Leia, and Sato all outperform this benchmark without even making a top 4 at all makes me question the analysis a bit.

Only Jerrod, Ackbar, Sloane and Motti have enough data points to be that meaningful.

Rieekan, Dodonna, Garm, Cracken, Mothma and Madine have been taken to regionals by known regular high ranking players.

13 minutes ago, RobertK said:

I think instead of just ranking the commanders by LBE you need to compare their LBE to 23%. That is the chance to just make the top 4 unweighted by commander. That Ozzel, Leia, and Sato all outperform this benchmark without even making a top 4 at all makes me question the analysis a bit.

Haven't done it yet, because sick dog is demanding love and affection, but I want to understand more about confidence levels when using LBE. Is LBE saying "this is the win probability" or "the win probability is no higher than this"?

As great as this data is, and kudos to those involved, I think taking anything away from this as far as what commanders are good is pointless. Player skill has more to do with it than we like to admit.

11 minutes ago, Baltanok said:

Haven't done it yet, because sick dog is demanding love and affection, but I want to understand more about confidence levels when using LBE. Is LBE saying "this is the top4 probability" or "the top4 probability is no higher than this"?

The likelihood of a top4 result upon the next attempt is X.

It comes partially from Beta distribution and makes the assumption that top 4 and not top 4 is possible for all admirals.

I'm no statitician so take this with a pinch of salt. Laplace gives me bad memories.

It seems like an additional metric needs tracked, like individual player performance over time. Beyond a ranking system, perhaps a "how many Top 8's" at a regional or larger event does the player have.

3 minutes ago, comatose said:

It seems like an additional metric needs tracked, like individual player performance over time. Beyond a ranking system, perhaps a "how many Top 8's" at a regional or larger event does the player have.

Noooooo.....

We just need more results.

Do we think that the horse is not behind the cart possibly?

If you look at the Rebel Admirals you could conclude that Ackbar, Reeikan, and Dodonna are all equally setting at 3 top 4 finishers each and say that the Rebel meta is diverse. However, fleet builds often dictate the admiral as much as the admiral dictates the fleet build. By which I mean that the Reeikan and Dodonna lists are almost certainly all 2-3 Rebel bomber wings. I think we should look by fleet archtype to get a better idea of what is winning.

Viewed through that lens, 2-3 Rebel bombers has a clear edge on the rest of the field, which I think we all suspected anyway.

Also of consideration, Sloane is super popular, as is Motti and Jerry. Of the 3, only Sloane significantly changes the way a fleet is built. I have always said that Rebel Admirals change the way a fleet plays, where Imperial Admirals tend to simply cover up weaknesses in a fleet. I think that is why you do not see a single Imperial Admiral or fleet build rising to outperform the others, because Imps are Imps, the Admiral just lets you key in on and eliminate a key weakness, or enhance a key strength in your list.

Most Imp. Admirals are interchangeable list to list, rebel lists typically are not. Put Ackbar in a Dodonna list and you may have a hard time...

#Can'tBuyAnArticle :-(

3 minutes ago, Space_Cowboy17 said:

By which I mean that the Reeikan and Dodonna lists are almost certainly all 2-3 Rebel bomber wings.

I haven't kept up since Atlanta, but I don't think that's the case.

What fleet archtype gels well and is being run with Dodonna besides Rebel Bombers?

Someone might take him as a useless "cheapest admiral," in a token farming Super Pickle build, but I doubt it. Reeikan, or basically any other admiral would be much better in that fleet.

25 minutes ago, Space_Cowboy17 said:

By which I mean that the Reeikan and Dodonna lists are almost certainly all 2-3 Rebel bomber wings.

Maybe? I'm not disagreeing but I do want to see the data to check that statement.

25 minutes ago, Space_Cowboy17 said:

I think we should look by fleet archtype to get a better idea of what is winning.

Agreed, if possible.

It is a lot more work for the stats people to do. This is more a quick and dirty indicator, but thank you for this. It is always nice to see what is happening in the rest of the world.

Those 6 Dodonna fleets:

(place/total)
ships
squadrons

(1/32)
GH/Yav/75/75/75
B/Ten/Corran/VCX/VCX/Luke/Dutch/Gold

(2/24)
Admo/Yav/HH/75/75
Ten/Dagger/Jan/Biggs/Gold/Y/1300/1300

(4/32)
GH/Yav/75/75/75
B/B/Ten/Corran/Jan/Luke/X/Y

(5/14)
Pelta/HH/HH/HH/75/75
Jan/Luke/Wedge/X/Norra/Gold/Y/Y/1300

(13/32)
80/90/90/75
Green/Shara/B/B/HWK/Rogue/Norra/1300

(16/32)
GH/Yav/Admo/75
HWK/Biggs/Luke/Wedge/X/Norra/Gold

Ah, seems I was mistaken.

(But the original line I quoted was "2-3 Rebel bomber wings", not Rebel bomber lists in general - which means I was still technically correct, the best kind of correct.)

1 hour ago, Space_Cowboy17 said:

What fleet archtype gels well and is being run with Dodonna besides Rebel Bombers?

Someone might take him as a useless "cheapest admiral," in a token farming Super Pickle build, but I doubt it. Reeikan, or basically any other admiral would be much better in that fleet.

I ran a 3-2 instead of a 2-3..... Does that count? Lol

4 hours ago, Baltanok said:

Neat! How would you determine the confidence level for the LBE?

Honestly, I'm not sure. Suffice it to say that it isn't very high. The estimators for Jerjerrod, Sloane, Ackbar, and Motti are probably pretty good, but the others are just guesses.

Also, this methodology is somewhat flawed because the variables are not conditionally independent — as one admiral performs well, the others must necessarily perform worse, relatively speaking. And that is not even considering confounding variables such as player skill.

That said, it is a better estimate than looking at the list of Top 4's and asking how many times a given admiral is represented. The clearest example of this difference is in Jerjerrod. He has the most Top-4 finishes (5) of any admiral, which might lead one to believe that he is the best admiral to take. Yet the rule of succession only predicts about a 35% chance of placing Top 4 with him, much lower than Rieekan's 57%.

Oops, I forgot what the whole point of this analysis was in the first place! Sloane, it seems, is not stand-out strong as some have claimed. In fact, she seems to be one of the weaker admirals currently, with only 1 in 6 Sloane fleets making Top-4 placings.

Edit to add: Though she doesn't seem stand-out bad either, as others have claimed.

Edited by stonestokes