Long term success in a league

[dode74]

See, that's the thing. It's not defending the ball against Skaven, it's defending the ball (and your team) against Skaven inducements.

I consider that a part of "defending against skaven", what with that being the game and everything

So skaven are efficient at low TV - I don't have an issue with that. It's yet another reason not to MM by TV.

Anyway, we're not going to agree on this. We have very different experiences of this and the OP has already got what he needs from this thread, I suspect. You can carry on with your theorybowl all you want, or describing how good the rat players are in your league (if that is what you are doing) but I'd rather look at stats from real games from a range of coaches, because that's what tells me how the game is balanced, not your small sample.

[mattgslater]

Quit with the straw men. You concede my point, pretend I'm making some other point, and use that made-up argument you put in my mouth to snipe at me. Not cool.

I'm not complaining that Skaven are unbeatable. They're popular here, but our best coaches mostly prefer Orcs, High Elves, Pro Elves, Halflings, and Nurgle, with only one good Skaven coach and two average ones (one exclusive). (I'm 8-1-1 against Skaven since we started our current format, though 5-0-0 on par or as an underdog and only 3-1-1 as a massive overdog. But thanks for your concern.) Notably, our bottom-feeders don't coach Skaven (one likes Amazons, one likes Necromantic), so that might skew things a little. But I think that's true of bottom-feeder coaches in general, who usually either get hung up on one race, give up on record and play T3 teams, or gamble on cheese (usually killstack or Amazon, but there are a fair number of third-rate old-timers who didn't get the memo and still try to run the Green Wall), after a season or two.

What I really was saying was that low-value Skaven are, all else considered, more effective than high-value Skaven. I brought up Skaven in response to your claim that higher TV teams tend to be better than lower ones: my point was that there are no absolute rules for TV curve, and it's only if you aggregate the data to the point of decontextualization that you see much of a relationship between TV and win%.

In fact, I'd argue that this discrepancy is more than made up for by a post-hoc noise factor, an incidental noise factor, and a legitimate, but highly conditional, element.

First, there's the tendency among young teams (up to 12-15 games) to spike in TV as they win and not as they lose, as losing teams tend to take more early damage and earn fewer SPP and less cash. After awhile, it doesn't really matter: the losing teams hit their curve too. But if you aggregate a bunch of leagues at all points in their development, you'll get a bunch of that noise. I don't think this is sufficient, but it does mean you have to factor out games between young teams in your considerations.

Second, all that about experienced losing teams eventually peaking aside, bad coaches play at slightly lower average TV, in the aggregate. Simply put, bad teams take more damage than good ones, and are more likely to be playing with MNGs and to retire players. Armour and ClawPOMB teams want to peak at high values, but if you're losing and having guys miss games, your TV will fluctuate a whole lot, and your high-peaking bash team will play (and lose) underdog games quite a lot, when you'd naturally want to be favored. By contrast, team trimming is a much easier skill for bad coaches with agility teams (in fact, it happens naturally once they learn to live with just enough TRRs), and they too will play as underdogs and lose games, because they're bad coaches: this just factors them out, and doesn't counterbalance all those Orcs that just can't hold it together.

Finally, a major element of TV is skills, but TV doesn't notice skill quality or combo power. I guess this is a way of saying that teams that hit and stay at their sweet spot are rewarded more than teams on either side of it, but it's probably better to go over than to go under. Once you've got all your good combos and your power skills and your toolbox, you're not getting better by getting bigger. Until then, all the stuff you're taking is better than inducement quality, and sometimes much better. It's like the arguments about wealth and happiness. If you make $20k a year, you'll be unhappy because you're poor. If you make $80k or $8Ma year, you won't. The first $60k gap is huge, but the next $7,920,000 yields very little psychological gain on average. Ditto TV. Every team has a set of skills and positionals and RRs and staff and bench that it needs, and anything beyond that is somewhere between a mixed blessing and a frivolous indulgence. (Except Fan Factor: that's a rich man's problem.)

So, yeah, you can handicap games by TV, but it takes a complex patchwork of matchups and racial considerations. Number of players, bench value, race @ TV, race @ handicap for/against (1.8M High Elves give away 150k, don't care; 1.8M Amazons give away 150k, shaking in boots).

[dode74]

Quit with the straw men. You concede my point, pretend I'm making some other point, and use that made-up argument you put in my mouth to snipe at me. Not cool.

Where, exactly? The TV-based MM thing was my own point, not yours, and nor did I pretend it was.

my point was that there are no absolute rules for TV curve, and it's only if you aggregate the data to the point of decontextualization that you see much of a relationship between TV and win%.

Argue it all you like, that kind of aggregation was how the BBRC set the tiers. You may want something different - you may want all the teams to be equivalently efficient or competitive at all TVs, for example - but that is not how the balance was worked - it was done on aggregated results (win%, not win% vs team X at TV Y).
I've also not stated it was an absolute rule, so you can quit with the straw man as well. It's a tendency and was an aim.
As an aside, if you deconstruct data to too low a level then it also becomes meaningless. It's only with aggregation that the bell curve for 2d6 becomes apparent, for example; and it's only by looking at the overall win% at all TVs that the fact that the balance is working becomes apparent.

In fact, I'd argue that this discrepancy is more than made up for by a post-hoc noise factor, an incidental noise factor, and a legitimate, but highly conditional, element.

Maybe. As I have said before, the correlation between having a higher TV and winning is fairly weak, but there is a correlation.

Reference your 3 points, I tend to agree in general. From your first point, in MM teams which are unsuccessful early tend to get cut: that should filter out that noise. Reference your second point, I'm not sure where "all those Orcs who just can't hold it together" are coming from - I've seen plenty of highly competitive Orc teams in both leagues and MM. Reference your final point, I agree. Sweet spot teams are better rewarded and beyond that it's bloat - another reason not to TV-base MM; I already quoted Galak somewhere else where he stated that some teams were meant to take inducements to enable them to better compete with teams which peak at higher TVs - it was an argument against TV-based MM.
None of that changes what was my point, though: in general, the underdog has a lower chance of success than the TV overdog. You simply think that statement is meaningless, whereas I see it for what it was: a design goal for the BBRC.

I don't know what you want this game to be, Matt, but you're really not liking it for some reason. Maybe because it's not LRB4 any more, maybe it's because you're going too deeply into the theory, but it might be worth taking a step back

[Aliboon]

Interesting stuff. Can't say that I've found skaven inducements to be that good, yep the wizard is great for them, BBabes are good, Fezglitch can be nice, if he works (if...) Glart is pretty good against AV9/wrestle heavy teams, the rest, meh. I'd rather be an elf and get the wizard and Edril tbh, but then maybe I run my skaven too deep. I just don't think that 12 is enough, the linos aren't very good at dodging away, it's too difficult to keep all the grunners safe when rats down, so one or two will get crumped by the end of the game (which makes life a lot tougher). Yeah the game may be won by then, but with likely only 11 for the next game, then things are only gonna be worse next game.

I will however try to keep my new Pro Elf team @ 12 players (the best team for it imo - 4 star catchers, 2 star blitzers, a half decent thrower and a leader caddy, the rest make weights) to see how that works.

True about "optimal" TV, but although I'm sure it does exist, the game is designed that the overdog wins 65% of the time, and tbh I can't agree that 15% is all down to noise, wherever it comes from, some yeah, but not 15%.

Plasmoid had some stats regarding a fair few leagues (TT and online), which I think were for CRP (or pretty much CRP) so if you ask him nicely, then he might pass them on.

Oh, and please don't bracket all online leagues as lacking in inducements, us at MBBL have everything! (and mattslater, I think you'd like it there, no time limits, so as much time to plan your turns as you'd like without getting timed out, although a few skills aren't quite what they might be, side-step being the prime example...).

[Megr1m]

QWhat I really was saying was that low-value Skaven are, all else considered, more effective than high-value Skaven.

As much respect as I have for your tactical skill, there are a couple of problems with this. Your play examples rely far too strongly on a single Fireball working in favour of the Skaven team, along with (relatively) cheap star players who you assume will remain on the pitch. Competent bash team can and will target the best pieces - and the problem with rats is that as soon as you lose that buffer, your Gutter Runners become available. And if those start to go, 99% of all Skaven teams ever have nothing left.

The other issue is that you're looking at it from the point of view of a single game. Yes, having low tv and massed inducements is arguably a good thing for one match. However, in a league, and especially in a league with multiple bully teams, you need bodies to feed into the grinder if you expect to keep the momentum going over a season. In a league with well-developed bash teams, this is doubly so.

As a corollary, and I may be wrong about this, but keeping the TV low seems to imply the very typical pattern of having a few overdeveloped pieces, with the rest of the team being supporting chaff. In the case of Skaven this will be the GR, which in turn leads to the aforementioned problem. Losing those key players can mean going into a downward spiral of not being able to win games, losing more players, not having enough money to replaces pieces, etc... Conversely, a higher-value but more well-developed all around Skaven team will still be vulnerable to the usual attrition over time, but significantly less so.

[mattgslater]

See, where I think the designers went wrong was the belief that inducements could yield a 65/35 ratio that you could point to, meaningfully. For some matchups it's 35/65, or 50/50, or 75/25. Saying that the designers only looked at the aggregate may explain the source of the problem, but doesn't make it any better.

@ dode: My point was that your little piece of jujitsu was unappreciated. We seem to agree that TV matching is a poor way of generating matchups. I was trying to add to that that if you don't do TV matching you quickly see that the 65/35 ratio holds between developed and undeveloped teams (though the relative success of so many rookie teams in FUMBBL tournaments suggests otherwise), but not between developed teams of different races in similar positions with respect to their own curves.

@ Megr1m: I like your thinking, and I remember wistfully the days when it was the right lens through which to view the game. Look at any of my teams, and you'll see how I have stubbornly held to my "spread the love" philosophy. You'll also see that many of my losses are against old teams with a few stars, teams that my old-timer mindset says should have been easy rollovers.

Losing key players no longer hurts Skaven. There's no amount of handicap value that can't turn into one more Bribe, one more Babe, one more WApo, one more Dirty Player Linerat (great with Fezglitch + Bribe; acts almost like a free Bribe for T8 if you take just a little damage) or interim Guard Vermin, one more whatever you need. And a rookie GR is MA9/AG4/Dodge; you'll find something to do with him. Never again do you have to go into a match man-down. In fact, seldom will you go into a match against a team at your level without Star Players and/or Mercs to take you over 11 men.

[dode74]

See, where I think the designers went wrong was the belief that inducements could yield a 65/35 ratio that you could point to, meaningfully. For some matchups it's 35/65, or 50/50, or 75/25. Saying that the designers only looked at the aggregate may explain the source of the problem, but doesn't make it any better.

I don't think it's a problem, personally. To give the correct win% for the underdog across the board would be both incredibly difficult (due to the complexity of the game) and, imo, undesirable because it would remove some of that wonderful unpredictability. Aggregates work when designing systems where you want unpredictability - using the 2d6 bell curve to set desired aggregate outcomes is another example of this. To expect that the same should hold in the smaller scale is both unreasonable and an overlay of your own expectations rather than an assessment of the game.

@ dode: My point was that your little piece of jujitsu was unappreciated. We seem to agree that TV matching is a poor way of generating matchups.

It wasn't jujitsu - I was using the comment to make a general point that TV-based MM is a bad thing. I'm glad we agree. Don't be so defensive

I was trying to add to that that if you don't do TV matching you quickly see that the 65/35 ratio holds between developed and undeveloped teams (though the relative success of so many rookie teams in FUMBBL tournaments suggests otherwise), but not between developed teams of different races in similar positions with respect to their own curves.

My big question to this is "so what?" As I said above, that it doesn't hold in small samples but does in aggregate is, as stated above, a good thing.

As an aside, it's not a 65/35 ratio. It's that "the low TV team should have no worse than a 30% win%". Not trying to score points, just aiming for accuracy.

[mattgslater]

Aggregates work when designing systems where you want unpredictability - using the 2d6 bell curve to set desired aggregate outcomes is another example of this. To expect that the same should hold in the smaller scale is both unreasonable and an overlay of your own expectations rather than an assessment of the game.

Two things.

1) In design, there's a time and a place for aggregation. When you're talking about games built around a single roll of 2d6, like Craps, tactical Backgammon, and some little facets of BB viewed in isolation all work that way. There's plenty of room for aggregating stats to get a good rough picture of the impact of your solution. But don't mistake that primitive map for the territory.

2) We're not creating anything with these aggregates: we're observing them. They're not "generating unpredictability": they're a willful mask for collectible data. Matchup between two teams at two different TVs has a lot of dependent variables, and the outcomes of the matches are highly dependent on numerous variables, but the on-paper matchup itself is a concrete, if subjective, thing. To grab thousands of them and refuse to analyze them for their traits you create a variable, but only in the sense that if you throw your keys out into the street and then blindfold yourself and spin around, they could be anywhere.

My big question to this is "so what?" As I said above, that it doesn't hold in small samples but does in aggregate is, as stated above, a good thing.

I never said it was a bad thing, not at all. What I mean is that teams with low sweet spots (Skaven being the best, but not only, example) can't be crippled, but teams with high sweet spots, which rely on spread-out skills and don't get good inducement bargains, can be rendered irrelevant for a long time with a streak of bad luck, much like those light teams used to be subject to crippling in the days before Journeymen.

As an aside, it's not a 65/35 ratio. It's that "the low TV team should have no worse than a 30% win%". Not trying to score points, just aiming for accuracy.

Glad to hear. On that count it works. I think that before you consider coaching, no team has much less than a 1-in-3 chance to win any given game, no matter what TV. In some cases, underdogs have better than 50/50!

[DoubleSkulls]

See, where I think the designers went wrong was the belief that inducements could yield a 65/35 ratio that you could point to, meaningfully. For some matchups it's 35/65, or 50/50, or 75/25. Saying that the designers only looked at the aggregate may explain the source of the problem, but doesn't make it any better.

We didn't have a lot of statistical data to look at to ensure that the inducements worked for every racial match up at every gross and net inducement level. Even for LRB6 the volume of data was comparatively low.

[DoubleSkulls]

The target was for a tier 1 team to have a floor of around .300 even if it were a massive underdog - so underdogs always had a chance of winning a game.

It seems there is a bit of noise about wizards being a bit too effective. Plasmoid wanted to drop lightning bolt and matt you think the fireball is too good for skaven (and presumably other fast teams). It makes me think perhaps wizards should be unchanged but 200k.

[dode74]

1) In design, there's a time and a place for aggregation. When you're talking about games built around a single roll of 2d6, like Craps, tactical Backgammon, and some little facets of BB viewed in isolation all work that way. There's plenty of room for aggregating stats to get a good rough picture of the impact of your solution. But don't mistake that primitive map for the territory.

The "rough picture" was the target.

2) We're not creating anything with these aggregates: we're observing them. They're not "generating unpredictability": they're a willful mask for collectible data. Matchup between two teams at two different TVs has a lot of dependent variables, and the outcomes of the matches are highly dependent on numerous variables, but the on-paper matchup itself is a concrete, if subjective, thing. To grab thousands of them and refuse to analyze them for their traits you create a variable, but only in the sense that if you throw your keys out into the street and then blindfold yourself and spin around, they could be anywhere.

What you're actually talking about here is the ability of TV to accurately reflect on-pitch capability against a wide range of opponents. That's impossible, even at 1000TV. Skills are only as valuable as their usage allows. At 1000TV tackle on all those Dwarves is wonderful value against Amazons, but completely wasted against Orcs, for example; at 2000+ TV claw is superb on Chaos against Orcs, but completely wasted against Amazons. TV itself is an aggregation because of the variable utility of skills. It might be possible, with enough data, to come up with an "accurate" matchup system which compared teams and applied the "correct" amount of inducements based on skill selections etc, but I suspect that might be quite the task.

I never said it was a bad thing, not at all. What I mean is that teams with low sweet spots (Skaven being the best, but not only, example) can't be crippled, but teams with high sweet spots, which rely on spread-out skills and don't get good inducement bargains, can be rendered irrelevant for a long time with a streak of bad luck, much like those light teams used to be subject to crippling in the days before Journeymen.

I disagree. Skaven most certainly can be crippled (literally as well as figuratively) through effective play. Certainly the wizard can be effective, but it can also be a total damp squib. I think the teams which can struggle are those with players such as BOBs which have a lot of potential but take time to make solid.

Glad to hear. On that count it works. I think that before you consider coaching, no team has much less than a 1-in-3 chance to win any given game, no matter what TV. In some cases, underdogs have better than 50/50!

Again, so what? That's part of the fun, and part of the management aspect. The underdog having better than 50:50 is hardly cut-and-dried, and I still think that coaching ability will almost invariably have a greater impact than the historical records of those two teams at whatever TVs with varying builds and different coaches.

I agree with DS on the wizard. Currently it's a no-brainer with a 150TV difference, and I know that people will actually cut rookies to ensure they get one.

[the.tok]

@ Megr1m: I like your thinking, and I remember wistfully the days when it was the right lens through which to view the game. Look at any of my teams, and you'll see how I have stubbornly held to my "spread the love" philosophy. You'll also see that many of my losses are against old teams with a few stars, teams that my old-timer mindset says should have been easy rollovers.

I think every player from 3ed/LRB4 has a really hard time to stop "spreading the love" as you said. I used to be such an important part of the game to give TD to your linemen to skill them up. I often have to hold back from handing off from a star, I still have the feeling I'm doing something bad if I let him hog SPP.

Quote:
2) We're not creating anything with these aggregates: we're observing them. They're not "generating unpredictability": they're a willful mask for collectible data. Matchup between two teams at two different TVs has a lot of dependent variables, and the outcomes of the matches are highly dependent on numerous variables, but the on-paper matchup itself is a concrete, if subjective, thing. To grab thousands of them and refuse to analyze them for their traits you create a variable, but only in the sense that if you throw your keys out into the street and then blindfold yourself and spin around, they could be anywhere.
What you're actually talking about here is the ability of TV to accurately reflect on-pitch capability against a wide range of opponents. That's impossible, even at 1000TV. Skills are only as valuable as their usage allows. At 1000TV tackle on all those Dwarves is wonderful value against Amazons, but completely wasted against Orcs, for example; at 2000+ TV claw is superb on Chaos against Orcs, but completely wasted against Amazons. TV itself is an aggregation because of the variable utility of skills. It might be possible, with enough data, to come up with an "accurate" matchup system which compared teams and applied the "correct" amount of inducements based on skill selections etc, but I suspect that might be quite the task.

I don't think it must be a purpose to have each individual matchup fair.
The only thing is to have a playable metagame.
I have no problem if low TV Skaven vs high TV orc is a 65-35, as long as another team gives that low TV skaven gets 35-65 by another reasonably viable team (any TV for that matters). They could even be 90-10 matchups, as long as they are spread among teams.

But honestly, with the data I've seen from now, I have yet to see a clear match-up chart, the way other games community do it (versus fighting for example, that's how they define tiers mainly).
For BB, it would be pretty complicated, cause you'd have do make at least TV100 125 150 175 200 for each team, wich makes roughly 5000 Matchups to discuss.

But when that is done, all arguments are settled (and what about the 135 undead )


And you would end up with a tier list, with probably...
125TV skaven, 150 TV wood elves and 200TV chaos for S tier
or maybe not?

[DoubleSkulls]

There used to be very comprehensive stats like those on FUMBBL. Not sure where they are now or if they are still around.

[Chris]

The target was for a tier 1 team to have a floor of around .300 even if it were a massive underdog - so underdogs always had a chance of winning a game.

It seems there is a bit of noise about wizards being a bit too effective. Plasmoid wanted to drop lightning bolt and matt you think the fireball is too good for skaven (and presumably other fast teams). It makes me think perhaps wizards should be unchanged but 200k.

Well presumably the way to solve things for teams that can make more use of a wizard than others is to list it on the team sheet with the star players. So wizards become a form of starplayer effectively. Not rostered because then you could always have it, but variable cost inducement.

[the.tok]

There used to be very comprehensive stats like those on FUMBBL. Not sure where they are now or if they are still around.

Couldn't find them.
I'm not very familiar with Fumbbl though, so if someone knows, I would be very interested to see those datas. With CR it would be even better.