So, on paper...

- Swarm beats Turret Tanks

- Turret Tanks beats High PS arc dodgers

- High PS arc dodgers beats swarm...

...any more for any more?

Wave 6 has the spocks and lizards.

Punch to the neck beats pointless inane comments...........

Any more for anymore?

Turret Tanks can be separated into those with Gunner/Luke, and those without.

Turret Tanks without Gunner / Luke do not necessarily counter higher PS ACD Phantoms.

So you really have 2 categories of turret tanks there. Turret Tanks without gunner generally counter Turret Tanks with gunner, because they only have 0-1 agility.

So I don't know that it easily breaks down into a 5 way split like paper rock scissors lizard Spock, but... it's something like that.

Overly simplistic and valid in only the absolute broadest of senses.

Player skill trumps all...

Player skill trumps all this FotM rock/paper/scissor stuff. If the game was that simple it would be boring.

Overly simplistic and valid in only the absolute broadest is senses.

That's the nature of a theory - it's expressly a simplification. Now, if you have a salient permutation or different paradigm, we're all ears.

Player skill trumps all this FotM rock/paper/scissor stuff. If the game was that simple it would be boring.

Granted, but it would be interesting to know by how much. How much better do the scissors need to be to beat the rock?

Just saying 'skill matters more' is just a conversation stopper. What more do you have?

I went into two tournaments this weekend really worried about rock-paper-scissors type matchups but it never ended up that way. Falcons might have an advantage but it isn't instant loss for the arc-dodgers. You still have to fly it right.

Player skill trumps all...

.... in chess

Player skill trumps all...

I dunno, I mean skill is the biggest factor by far, but the list matters a little. Pretty sure a top ranked player would struggle to beat any average player if he had to fly 6 turretless Rebel Operatives, and the average player could choose his list.

Overly simplistic and valid in only the absolute broadest is senses.

That's the nature of a theory - it's expressly a simplification. Now, if you have a salient permutation or different paradigm, we're all ears.

There we go. I guess what I'm trying to say in an incredibly obtuse manner is that no matter what kind of list you build, there is another list out there which is going to give you kittens.

2 ship Falcon lists do NOT enjoy facing half a dozen or more low-priced ships. Swarms do not like facing ships they find almost impossible to block or pin down in arc. Arc dodgers do not like facing lists which have turrets.

Sure it's simplistic, but whichever way you go about tailoring a list there's going to a similarly tailored list which you can't account for, or at the very least will face an uphill struggle against.

Overly simplistic and valid in only the absolute broadest is senses.

That's the nature of a theory - it's expressly a simplification. Now, if you have a salient permutation or different paradigm, we're all ears.

For me, the issue (and it's unfortunately also a conversation stopper) is that "swarm", "turret tank", and "high-PS arc dodger" are all ill-defined. Is a list with 8 TIEs a swarm? How about a list with 5 TIEs and Howlrunner? Because there's not actually a great deal of similarity in how they deploy or typically operate.

And clearly a list with Chewie + Draw Their Fire and Lando + Draw Their Fire is a turret-tank list, but what about Chewie and two or three Headhunters? What if the Headhunters have missiles? What about Han + Corran? All of those lists feature at least one Falcon, but they're not very similar in a strategic or tactical sense; you can't very meaningfully lump them all together as "Falcon lists".

And of course the same can be said of high-PS arc-dodgers--and of course there are archetypes that weren't included in the initial list. Three-ship Rebel elites are more of a thing now than they were, and I recently lost a match to BBXX, and I've seen variations on the Imperial 9-9-9 start going around again, and one of the top 8 at GenCon was running three Lambdas and a Firespray. You can argue about how prominent and/or successful any of those list archetypes are right now, but I think oddball lists make up at least a non-negligible part of the metagame.

But even more problematic is when you start hybridizing between those three majors. If I'm running a stripped-down Chewie and four Bandits, does that count as a turret-tank list or a swarm? What if I'm running Whisper with Howlrunner and three Academy Pilots? What if I'm running Kenkirk + Echo?

Rock-paper-scissors isn't a very interesting game, but it works because the list of choices is exclusive and exhaustive: you can choose only one option from the list, and there are no options that aren't on the list. Given that definition and the diversity available even among successful tournament lists that "fit" the archetypes laid out in the OP, it seems to me that the X-wing metagame isn't well represented by the rock-paper-scissors model.

Player skill trumps all...

.... in chess

I guess at really high tier tournaments in the final rounds, where only the best of the best are left playing, you could compare the lists themselves in a rock/paper/scissor manner, but I can't help think there's more to it than that.

You could say that skill trumps all. There are also people who say dice trump all. Undoubtedly, there's an underlying truth to both theories, and the degree to which you embrace either might just be a function of how your mother treated you as a child.

But, where the OP is going with this is the 'ceteris paribus' approach: all other things held constant. Sure, skill and dice matter, but there's no point in theorizing about them, except to try and isolate how much they matter compared to other things.

Vorpal Sword, as usual, has the more important critique. To what extent are these archetypical lists a part of a rigid typology? I think that's definitely a question worthy of exploring. I do think there's utility in making the typology so as to fit it into a rock<paper<scissors<rock model of the game.

I also think that we might actually be able to test the proposition using MajorJuggler's data, though MJ would be a better judge of how that data might be used, because he's 1) so much more familiar with it, and 2) 100x the mathematician that I am.

You could say that skill trumps all. There are also people who say dice trump all. Undoubtedly, there's an underlying truth to both theories, and the degree to which you embrace either might just be a function of how your mother treated you as a child.

Just checking... are you serious here?

You could say that skill trumps all. There are also people who say dice trump all. Undoubtedly, there's an underlying truth to both theories, and the degree to which you embrace either might just be a function of how your mother treated you as a child.

Just checking... are you serious here?

Yes, mostly.

I also think that we might actually be able to test the proposition using MajorJuggler's data, though MJ would be a better judge of how that data might be used, because he's 1) so much more familiar with it, and 2) 100x the mathematician that I am.

Player skill is still an unknown variable. This is a strong argument for only using the Top Third + Final Cut matches to get as similar a baseline as possible. However if I had all the data (and I'm trying to for Nationals) I will gladly slice it each and every possible way, eventually.

The variety of archetypes as Vorpal said is staggering. My analysis only looks at a relatively small slice, because I only differentiate between ships and nothing else. That will change once I get some scripts running.

The main problem is that we don't have a full record of what lists lost to what. The only place I have this recorded, is for the elimination matches in the more recent tournaments. The Holy Grail of data collection would be to get the match by match breakdown (including MoV) of every single match during Swiss. The next step is going to be for me to interface more directly with the tournament software (and coders) that's out there. If we get a "standard" software package that everyone uses, this will be much easier. Email me the results, I run scripts, it spits out stats. However, I need to graduate first, and I'm already working full-time.

All hail the mighty dice

Player skill is still an unknown variable. This is a strong argument for only using the Top Third + Final Cut matches to get as similar a baseline as possible. However if I had all the data (and I'm trying to for Nationals) I will gladly slice it each and every possible way, eventually.

Right, I think the approach, if possible, would be to see how reliably certain types of lists beat other types of list. Beyond that, we might assume luck as an unbaised error term.

The variety of archetypes as Vorpal said is staggering. My analysis only looks at a relatively small slice, because I only differentiate between ships and nothing else. That will change once I get some scripts running.

Would it be possible to do something like a factor analysis to see if there are underlying correlations beyond the factional?

The main problem is that we don't have a full record of what lists lost to what. The only place I have this recorded, is for the elimination matches in the more recent tournaments. The Holy Grail of data collection would be to get the match by match breakdown (including MoV) of every single match during Swiss. The next step is going to be for me to interface more directly with the tournament software (and coders) that's out there. If we get a "standard" software package that everyone uses, this will be much easier. Email me the results, I run scripts, it spits out stats. However, I need to graduate first, and I'm already working full-time.

That's what I feared. It would be awesome if we did have data at the dyad level.

Here's the thing. I wonder if it's possible to build an online engine that can record results. I've already cogitated (a little) on that idea in the context of something for DagobahDave's galactic campaign, so as to relieve him of the burden of collection and data entry. I'm certainly not a magnificent coder or anything, but I could possibly figure something out over time (The fall semester just began and I've got more students than ever, so this is not a short-term proposition.) I do have the webspace for it, however.

Player skill is still an unknown variable. This is a strong argument for only using the Top Third + Final Cut matches to get as similar a baseline as possible. However if I had all the data (and I'm trying to for Nationals) I will gladly slice it each and every possible way, eventually.

Right, I think the approach, if possible, would be to see how reliably certain types of lists beat other types of list. Beyond that, we might assume luck as an unbaised error term.

The variety of archetypes as Vorpal said is staggering. My analysis only looks at a relatively small slice, because I only differentiate between ships and nothing else. That will change once I get some scripts running.

Would it be possible to do something like a factor analysis to see if there are underlying correlations beyond the factional?

I have some ideas about this, although it requires getting 100% of the data, and is the Holy Grail approach. I can't explain it all now, but I'll pin the thought and discuss it on the next Nova Squadron Radio. They invited me onto their next podcast and we are recording Thursday.

I have some ideas about this, although it requires getting 100% of the data, and is the Holy Grail approach. I can't explain it all now, but I'll pin the thought and discuss it on the next Nova Squadron Radio. They invited me onto their next podcast and we are recording Thursday.

Oh, cool. I hadn't heard of that podcast before.

*Googling*... *Finding...*

Good lord those episodes are long!

Player skill is still an unknown variable. This is a strong argument for only using the Top Third + Final Cut matches to get as similar a baseline as possible. However if I had all the data (and I'm trying to for Nationals) I will gladly slice it each and every possible way, eventually.

Right, I think the approach, if possible, would be to see how reliably certain types of lists beat other types of list. Beyond that, we might assume luck as an unbaised error term.

The variety of archetypes as Vorpal said is staggering. My analysis only looks at a relatively small slice, because I only differentiate between ships and nothing else. That will change once I get some scripts running.

Would it be possible to do something like a factor analysis to see if there are underlying correlations beyond the factional?

I have some ideas about this, although it requires getting 100% of the data, and is the Holy Grail approach. I can't explain it all now, but I'll pin the thought and discuss it on the next Nova Squadron Radio. They invited me onto their next podcast and we are recording Thursday.

I'd like to see dyadic data for my own purposes--it should be possible to position each player on a continuum of overall performance, and that would do two things: first, I could match a "true skill" estimate up with each player's margin of victory and (old-format) strength of schedule.

Second, and far more interestingly, I think I can play with some psychometric toys that will let me place error bars on the tournament's "estimate" of each player's skill level. If I'm right, then it would be an empirical example of how good a major tournament is at resolving players from one another--is the outcome of the top 8 at Nationals approximately deterministic, or are the players clustered closely enough that the outcomes are relatively random? And if it's the latter, what's the first obvious gap--where's the first position where a player is significantly different from the next-best player?

I'd like to see dyadic data for my own purposes--...

How's your skill at coding?

I'd like to see dyadic data for my own purposes--...

How's your skill at coding?

Depends on the language, but I did finish a bachelor's in electrical engineering, which required work in Java and C++ as well as RISC assembly. Most of my coding (or, perhaps, "coding") these days is in SAS and SPSS, but I know enough of a couple of different scripting languages to at least get myself in trouble.