1 minute ago, Derrault said:

Oh so you did proper data collection this time?

Of course. You know we had all this data last year too? And published it...

Don’t worry, I got the link for when you come back saying “nah, I’m full of it”.

https://thefifthtrooper.com/lvo-the-data-article-part-one/

It’s probably invader you were thinking of, but I’d already mentioned not having the systems in place to check and QC every list. Really looking forward to the LVO 2020 deep dive - it’s gonna be a hoot! The pre-event breakdown (Units/Upgrades) is already you for comparison purposes.

Y’all keep on doin you. Missed this place 😘

And we missed you too @TalkPolite

1 hour ago, TalkPolite said:

Don’t worry, I got the link for when you come back saying “nah, I’m full of it”.

https://thefifthtrooper.com/lvo-the-data-article-part-one/

It’s probably invader you were thinking of, but I’d already mentioned not having the systems in place to check and QC every list. Really looking forward to the LVO 2020 deep dive - it’s gonna be a hoot! The pre-event breakdown (Units/Upgrades) is already you for comparison purposes.

Y’all keep on doin you. Missed this place 😘

As a heads up, the first excel excerpt on Objective/Deployment/Condition has Played as the first column for Obj/Dep and Cut for Condition.

Based on the context of the write up, it looks like that should also be Played.

edit: of course, then I go back and check the date and see it’s last year 🙄

*edit* Crisis adverted!

On 1/30/2020 at 9:50 PM, syrath said:

ust take a look at the variety on the imperial side I the top 8 and the lack of variety on rebel variety in the top 8 tells me that other variations aren't viable just now.

So here's a thought....not sure if it amounts to much, but it might be worth exploring.

So, Rebels only have 3 Units with red dice saves. Luke, Luke and Sabine.

Conversely Imperials have 7 with white dice saves : Krennic, Officer, Bossk, Scout Troopers (reg), Scout Troopers (strike team), Bikes and AT-ST.
The Empire also has a stark range advantage and 2 of the 7 vulnerable targets have a built in range 4 weapon (4 of the 7 if we're just talking snipers)

My hypothesis is that the lack of armor choice on the rebel side is what has led primarily to the lack of diversity in units. Tauntauns are the clear solid choice because they have great mobility as well as their overflowing back of tricks. Rebel troopers/rebel vets become cheap cappers because they are squishy, might not survive the attempt to close the range gap and suppress/panic easily so why waste the points on them? Wookiees may be braver, but they are squishy and have an even harder time closing the range gap (as well as being a melee unit that has a ranged weapon as its special). The heavies aren't as solid a choice because of the prevalence of impact and critical X on the other side of the gun line. So the rebels have simply fewer units that can expect to survive to do good damage to the enemy. That leads to stagnation and saminess in lists.

5 hours ago, Derrault said:

1) I didn’t say they’re mutually exclusive, I said they aren’t the same .

2) The perception of something being good isn’t actually proof it is.

You don’t get a special pass to not have evidence of your claims. Your individual and limited experience has exactly zero bearing on the question of whether a particular unit is performing at expectations, under, or over in tournament settings.

Yeah, no.

You may not have a functional understanding of how seeding impacts probability, but it works like this:

If you put 64 marbles in a jar, and 63 are red; it’s a safe assumption on pulling 8 that they will all be red.

If you change the proportion, the expected draw also changes. Ie if 8/64 are yellow, one of the drawn 8 can be expected to be yellow.

Conversely, if 0/64 are blue, then it’s a certainty that none of those pulled will be blue.

The only way to demonstrably prove the theory that unit X is overpowered is not to look at the top 8 alone and say, “Look, it was present in all these lists that ended up winning!”, its to compare those outcomes to the seed total, the field of 64 possibilities.

If it’s proportional, then guess what? It’s just what should be happening. Again, it’s not what’s winning, it’s what that is compared to what got entered to try and win in the first place. If a majority of the entered lists are of a composition, and they don’t get proportionally represented, it’s safe to say it’s underperforming. If a minority of lists are of that composition, and over represent in the top 8, then (and only then) can we say that it’s overperforming.

Sure? Nobody said point costs don’t matter, especially in making units appear more or less attractive at first blush, and without crunching any numbers.

1) You implied that.

2) People don't need unequivocal proof in a wargame. Just enough experience to understand what works and what does not.

3) I get a pass because this is man dollies. 'Limited' is subjective, therefore an irrelevant comment.

4) I understand seeding. Do you understand abstraction? Tactics? Are you able to quantify the benefit of activation control, range control, terrain layout and use, synergy of units, force balance, and activation efficiency?

5) In a wargame you don't need a large sample size to see these patterns emerge. After only a few games I could see that big vehicles were bad. Once you understand a game's mechanics, you can see what can/cannot work.

6) You argued in older threads that the T-47 and AT-ST were fine. I argued against that.

7) I think we may have to start agreeing to disagreeing soon.

6 hours ago, Derrault said:

The only way to demonstrably prove the theory that unit X is overpowered is not to look at the top 8 alone and say, “Look, it was present in all these lists that ended up winning!”, its to compare those outcomes to the seed total, the field of 64 possibilities.

If you wanted to go all the way with this, you'd need to account for players' varying skill too. Theoretically a unit might end up being over represented in the top 8 not because it's that strong but because it was played by better players, who would likely finish in the top spots no matter what they played (well, almost).

So yeah, you're technically correct that in order to prove units' strength with scientific rigor, you'd need a lot more than just a peek at top 8. The thing is that my many years of (admittedly anecdotal) wargaming experience have taught me that applying scientific rigor to analyse meta is neither reasonable nor necessary.

It's not reasonable because samples are too small and complexity of the matter too high to get solid results. In addition to the players' varying skill levels, dice variance can significantly pollute the results in case of less popular units and then there's the random factor of people getting paired with lists that do or do not counter them. You'd need much larger samples to filter all of this out.

And it's not necessary either. While it's technically possible that people in general misjudge the units' actual potential, when you browse through the top 8 lists you get the idea of what the best players consider to be good. Those people are usually very commited to the game, spend a lot of time both online and off testing various strategies and typically have IQ significantly above average. The chances that all of them are wrong are negligible and probably lower than the chances you're going to get flawed results from the statistical analysis for all the reasons I mentioned. If there's a consensus at the top tables that shores, tauns, snipers, Leia and Sabine are head and shoulders above the rest of the pack I'm willing to believe them without trying to confirm it by the means of statistical analysis. Call me lazy if you want to, but I'm probably right too.

So on some levels, I think Derrault has a point. If you just go "ah everybody used X unit, it's clearly really good", that kind of falls into tyranny of the majority - you assume the unit is really good or too good or whatever when the only thing the data told you was "it was popular". Similarly, if your entire tournament only includes some units, juts them making it to the top 8 doesn't indicate anything, it was going to happen regardless of the quality of the unit because somebody has to win and move on. Now, when based on personal experience or armchair commentary or just reading the cards, you look at a unit and see it is good at things, or how it can play well, and then you see also based on data it is popular, you can extrapolate "ah, obviously this is a pretty good choice". But that's technically outside the scope of just the raw numbers. Think of it this way; if you looked at early tournament results, you'd see a lot of units represented in them which are no longer currently represented. Are those units "good"? Well, if they were in Top 8s or winning they probably weren't awful, but if they disappear later, that probably means they are not as efficient, or new tools exist to counter them, or people just found something to do better what they wanted. To interpolate that say, Shoretroopers or Tauntauns are good, you can't just see their representation, you also have to bring in other data. We can reasonably conclude Tauntauns must be pretty good, based on what we know of the unit, and how it performs but not just by sampling the Top 8. That's a confirmation bias, we already assume Tauntauns will do well, and then tauntaun players rise to the top, but there's also E-Webs in the Top 8, are you gonna rush out to slap them in every list now? That's also then not to say when people further interpret these results to "Tauntaun OP Plz nerf" - but that's a whole different debate about people want out of that.

To understand why people build a list, you'd have to ask them, and understand their choices, which is a human element not reflected in just the numbers. Why did they take Shores. And there is a difference between "I really like shores and think they're good and practiced with them" to "internet man said Shores were good, and said E-Web bad". There is obviously some technical depth here, as the lists are not all exactly the same. But, this does go up to Lightrock's point above which is basically now we're in a way too complex set of variables to mathematically calculate everything because even if you sampled the whole tournament, you're not accounting for all the variables because winning a round of Legion is not just picking a marble from the jar. When similar styles of lists conflict, there will be elements like terrain, skill, objectives to consider.

So tl;dr I guess everybody is right and wrong? Don't go "Only shoretroopers in the Top 8 this PROVES shores are too good!" but also the idea of statistically proving Shores are too good is way to much of an effort. The Top 8 doesn't prove it inherently, but we can interpret based on lots of other factors.

@TalkPolite Still on the topic of LVO data, does the data collected include rounds completed for all games? The latest Notorious Scoundrels podcast spent a good chunk of time talking about games not going past round 4, use of chess clocks, and that players were issued slow-play warnings even in the final cut.

21 minutes ago, NeonWolf said:

@TalkPolite Still on the topic of LVO data, does the data collected include rounds completed for all games? The latest Notorious Scoundrels podcast spent a good chunk of time talking about games not going past round 4, use of chess clocks, and that players were issued slow-play warnings even in the final cut.

Yes, the forms collected what round games ended on, and if it was a table/concession. We’ll dig in to that and give average round times just like we did for LVO 2019 and Adepticon 2019

1) You implied that.

2) People don't need unequivocal proof in a wargame. Just enough experience to understand what works and what does not.

3) I get a pass because this is man dollies. 'Limited' is subjective, therefore an irrelevant comment.

4) I understand seeding. Do you understand abstraction? Tactics? Are you able to quantify the benefit of activation control, range control, terrain layout and use, synergy of units, force balance, and activation efficiency?

5) In a wargame you don't need a large sample size to see these patterns emerge. After only a few games I could see that big vehicles were bad. Once you understand a game's mechanics, you can see what can/cannot work.

I did not imply that at all. I said they weren't the same.
(Effective != Popular) != (Effective XOR Popular).

A thing can be effective, a thing can be popular, it can be both, or it can be neither. Neither property is tied to the other.

Nobody gets a pass when they try to pass off opinion as fact. Downplaying the importance of the topic in discussion seems silly; if it's not important to you, why keep making unsupported assertions?

The term limited isn't subjective, you're one person, and you're talking about a less than statistically significant number of events. Unless you are now purporting to have cloned yourself ala multiplicity?

If you truly understand seeding, then this shouldn't be a debate. Game strategy has nothing at all to do with analysis of unit performance at the tournament level. I mean, we can always have those discussions in their own threads...but they're neither here nor there when it comes to asking: "Is this unit overperforming in tournaments?"

People often claim to see patterns in randomized events as well, that's called pareidolia. Just because something occurred on a limited basis for you (limited being a statistically insignificant number of events, of course) doesn't mean that holds true at a larger scale, or for everyone. This is why larger studies are more meaningful than small studies.

Looking at tournaments has at least the potential to allow for extrapolation of information, but you need the whole set of data, not just the data points for 8 users.

6) You argued in older threads that the T-47 and AT-ST were fine. I argued against that.

7) I think we may have to start agreeing to disagreeing soon.

Ah, I see? Why does this matter?
If you don't want to engage in the discussion, I won't hold it against you.

If you wanted to go all the way with this, you'd need to account for players' varying skill too. Theoretically a unit might end up being over represented in the top 8 not because it's that strong but because it was played by better players, who would likely finish in the top spots no matter what they played (well, almost).

So yeah, you're technically correct that in order to prove units' strength with scientific rigor, you'd need a lot more than just a peek at top 8. The thing is that my many years of (admittedly anecdotal) wargaming experience have taught me that applying scientific rigor to analyse meta is neither reasonable nor necessary.

It's not reasonable because samples are too small and complexity of the matter too high to get solid results. In addition to the players' varying skill levels, dice variance can significantly pollute the results in case of less popular units and then there's the random factor of people getting paired with lists that do or do not counter them. You'd need much larger samples to filter all of this out.

And it's not necessary either. While it's technically possible that people in general misjudge the units' actual potential, when you browse through the top 8 lists you get the idea of what the best players consider to be good. Those people are usually very commited to the game, spend a lot of time both online and off testing various strategies and typically have IQ significantly above average. The chances that all of them are wrong are negligible and probably lower than the chances you're going to get flawed results from the statistical analysis for all the reasons I mentioned. If there's a consensus at the top tables that shores, tauns, snipers, Leia and Sabine are head and shoulders above the rest of the pack I'm willing to believe them without trying to confirm it by the means of statistical analysis. Call me lazy if you want to, but I'm probably right too.

I largely agree with you. I mean, it's conceivable that the top 8 are just the best at playing the game, regardless of their ultimate list choices; to control for that we'd need to see them all playing the same list in mirror match series. That seems like an implausible ask given how long games take, and the distance some people have to travel to play in the first place.

My understanding from the fragmentary reporting of prior tournaments is that there's very frequently been a significant amount of groupthink in list building, where all the rebels and/or all the imperials were advancing nearly the same lists. In which case, the top 8 list type results are pretty much pre-ordained. When you have only a couple lists deviating from the norm, it's much less probable that those lists are able to get through.

2 Opportunities to go 3-0 vs 62 opportunities. The safe money is on the 62, all things being equal.
The line I'd draw is when it becomes fundamental attribution error. A unit isn't good because the world champion runs or likes it. Not questioning the meta is a certain way to be consumed by the meta.

And, for that matter, what became the top 8 is a function of what came before. i.e. What was the 64? If the 64 is more of the top 8...ok? Then it doesn't tell us anything. If the top 8 are different than the rest of the field, then looking at the top 8 provides a meaningful answer to that. Otherwise, you're looking at 56 losers who had the same lists as the 8 winners. 56 > 8, no?

So tl;dr I guess everybody is right and wrong? Don't go "Only shoretroopers in the Top 8 this PROVES shores are too good!" but also the idea of statistically proving Shores are too good is way to much of an effort. The Top 8 doesn't prove it inherently, but we can interpret based on lots of other factors.

This is pretty much my position. Seeing the top 8 raises the question of if those units are too good; but confirming it requires the top 64 (and also seeing what lists had to fight what lists on the way up; since not everyone played everyone else). I don't agree that this is too much effort. All you need is to scan/data entry the lists into excel, somethings the tournament organizers would already be doing (i.e. Collecting lists, that's required + entering win/losses to account for the next round pairings).

The problem that drives the objection, is that it this always seems to devolve into: Unit X needs nerfing! But, if they can't prove that it's OP on an empirical basis, then why the call to nerf? It's just perception management, not a proven problem.

@TalkPolite are all the lists entered into the tournament available to show what the competition for the top 8 were?

15 hours ago, Polda said:

Lol

Shores are not as bad as the Jumpmaster. Funny meme though