Forum adverts like this one are shown to any user who is not logged in. Join us by filling out a tiny 3 field form and you will get your own, free, dakka user account which gives a good range of benefits to you: No adverts like this in the forums anymore. Times and dates in your local timezone. Full tracking of what you have read so you can skip to your first unread post, easily see what has changed since you last logged in, and easily see what is new at a glance. Email notifications for threads you want to watch closely. Being a part of the oldest wargaming community on the net. If you are already a member then feel free to login now.

 JNAProductions wrote:

 Peregrine wrote:

 JNAProductions wrote:
I'm not a mathematician, nor a paid game designer. I'd hope that people who ARE paid to do this kind of thing are better than me at it.

It's not a question of skill, it's about the complexity of the problem relative to the alternative approach. It doesn't matter how good you are at math or game design when the task is unreasonable.

Are you a paid game designer or mathematician?

Because, to my knowledge, you're just a random 40k fan. I know that it'd be difficult to do, since I cannot do it myself, but your assumption that it's TOO difficult is something I'd believe when an expert says so.

It's actually easily realized with logic that algorithm leads to broken system. Value of movement for example changes a lot based on what kind of unit are we talking about. M1 h2h specialist? Worthless. M1 heavy gun toting wielder? Low movement isn't much of disadvantage. Especially if you then factor in unit-external things like ability to move unit in otherwise. For example how much lootas from orks should drop in points if you dropped them to M1? Any algorithm would say points. Any ork player would trade all the movement for point drop. They aren't moving all that much to begin with and if they do it's generally with da jump because you need hefty repositioning to ensure LOS. The loss of movement is irrelevant. But sure if you think it's possible try to figure out how to factor in that movement drop while then also making it work for say meganobs. How much current meganobs should go up in price if you upped their movement to 6?

Then factor in terrain, opponents faction, units etc which should affect point cost and kiss goodbye to algorithm.

General rumour and scuttlebutt is that they have a rough calculator which they combine with gut feel and then change around after some play-testing. There is not much doubt that in some previous editions a middle manager would occasionally intervene and mess that all up royally - hence the insane costing on a 7th ed Wraithknight.

I quite like the article Variance Hammer did on this whole subject http://variancehammer.com/2018/06/04/how-would-nasa-balance-40k/

Basically there is no algorithm available which could do an even half-decent job. If you have a few millions going spare and can interest some of the best simulation and AI computer scientists in the world then go for it, you may or may not be able to improve on what GW do.

Just read it. Good stuff.
As far as I understand it, the things he mentioned were the best ways of coming up with a grand solution. (Or problem, in this case).
I'm the only liberal art wonk in a family and friend network including physics majors, a member of the team that designed Oculus, a guy who left Lincoln Lab because he got bored, a CPA, and a controller for a major chemical company.
So, by the transitive property and the laws of physics, that naturally results in me knowing f-all about math more complex than standard geometry.
I know many of the concepts, but very little of the practice. I know what an algorithm is, and what it does, but I've never really seen one or have the faintest clue on how to design one.

After that wall of text, there is a point. I know what math can do, but not how to do it at that level. However, I'm sort of a pattern guy- I did stuff in the Army relating generally to pattern-recog., and I think many people tend to over complicate the basic question.

If one compiles all that he said (every stat line in the game, every possible interaction, all the bubbles, stratagems, traits, buffs, nerfs, ranges...everything) it would be Sisyphus pushing the dang rock up the hill... endless. (I think it was Sisyphus.. )

Anyway, I'm trying to find a formula in the potential, not the execution. I'm trying to find the empirical, not the predictive. So, leaving the interaction out and writing that part up to the acumen of the players, I want to try and see just how broken the rule set is, if it is. (Humor me, I know it is, but I'm trying to be devils advocate.)

I really think that the answer is attainable, within the boundaries I've set. I think that my friend (one of the physics dudes) and I will discover one of two things...

1. The stats are or were created around the same basic core design, and there is some sort of arithmetic or algebraic key. There may have been deviation over time, but the kernel is still there (MEQ).
2. The stats are pretty much all WAGs, and GW truly is just tossing poop on the wall and seeing what sticks. I really dont think that is the case, because in a game that relies so much on probabilities, ther simply HAS to be an origin point.

Danged if it ain't a deep rabbit hole, though.

I feel these threads always end up with "you need a super computer" - but to some extent I feel that's nonsense.

Sure, its hard to say "how many Assault Marines should a Storm Raven cost"?

But you can fairly easily look at two units that have essentially the same role - and are therefore likely to be an either/or on the table. Take say Fire Warriors and Breachers. Or Assault Marines and Vanguard Vets.

So taking the Fire Warriors - they do more damage outside of 10". In fact the breachers do no damage at all outside of 15". The Fire Warriors also synergise with the Fireblade, a cheap HQ which you are likely to want to take somewhere anyway - who buffs fire warriors up to being better outside of 5".

How often will breachers be able to get to shoot while within 5", versus every other situation imaginable? You don't need a super computer to tell you this - if breachers and fire warriors are the same price, you take fire warriors 100% of the time, all day every day.

I feel the analysis in the link is looking at things backwards. A Castelan isn't overpowered because its winning games - its winning games because its overpowered. Its overpowered because its a mathematically superior use of 600~ points. This can be modeled statistically versus whatever else you were going to spend those 600 points on.

I don't need to compute thousands of games. I just need to go "is it offensively superior, defensively superior and has better movement abilities than average for these points"? If thats the case either it should go up in points, or the units you are comparing to should go down to the point where the probability curves per point are roughly comparable. They don't have to be the same - they just have to be close enough.

Tyel wrote:
I feel these threads always end up with "you need a super computer" - but to some extent I feel that's nonsense.

I agree.

However, I think that the comparisons should be calculated horizontally, not vertically. Not so much "How is a Terminator better than a Tac Marine?", but rather "How is a tac marine better than a guardian?

If one accepts the premise that a basic tac marine is the control for marines, and possibly for the entire system, it might be better to compare the 'plain vanilla' of the other factions.

One thing I've noticed is the Tac marine attacks and wounds exactly the same in both ranged and melee.
So does the IG grunt.
So does the basic chaos marine.
I think the necron warriors do, as well.
I dont have the others committed to memory, but I know that a fire warrior skews to the ranged damage, but at the cost of melee, which to me seems like they are both skewing around a common point.

Worth looking into, I suppose.

Necron Warriors are better at Ranged. AP-1 as compared to AP0.

 Peregrine wrote:

 JNAProductions wrote:
I wouldn't say effectively impossible.

I'd agree that making a formula that creates, without fail, a well-balanced unit is effectively impossible, but a formula that gives a good baseline with which to work off of is probably doable.

Try doing it and you'll see what I mean. The complexity of even a moderately accurate formula is going to be way too high to be practical. For example, you might think of having a price per point in a stat, but BS 2+ is way more powerful on a shooting unit than on a unit with only a token bolt pistol for ranged weapons. Movement speed is far more valuable on melee units or units that want to move up to claim objectives than on a static gunline unit. Etc. Trying to capture any meaningful percentage of those differences inevitably leads to getting bogged down in a convoluted formula and tables of special-case exceptions and modifiers and such. It's certainly going to be way less efficient than just eyeballing the point costs based on experience playing the game and then going through a few rounds of iterative playtesting.

Why don't they just pre build list, then assign points so that those armies that are made efficient by GW fit only the points they assigned. there is ton of way they could force that, make extra units or gear of the same type cost more, or less depending on army ,unit or detachment. they could still have an open points system, the way they have now, but if someone would want an army that fits in to a "2000pts"slot they would have to build it the right way or they would end up with something bad. Then they could incentivise this by unit bundles, start collecting boxs etc. It is much easier to build an army and balance it against other armies first and then chop it up in to points later, as people will gravitate to the good builds anyway. And if they miss something and some combination is too OP? They just FAQ/errata it on the spot and change the rules in a CA. didn't think people would take 9 units of grechin, here you go they are now caped at 2. Flyers too powerful, same thing.

Less work, easier to fix, and everyone should be getting at least one working army this way. And not some armies having a ton of builds or units to switch out, and others thinking how to build a legal army that doesn't totaly suck.

I can see what you are saying. It would definitely take a lot of the guesswork when it comes to competitive games building. Three points, though, to think about:

1. GW is in the business of making money from model sales. The rules are a vehicle to promote this. Wouldn't these bundles of pre-made armies disincentivize the players from buying about 75% of the model range?

2. I love the centurion models. I think they are really cool and fit thematically with my Salamanders. However, from what I understand, they arent really a good investment in points due to various factors in the current meta. Should I throw them away? Many of the nice looking models simply arent competitive, and I think this would cause a similar problem for other people.

3. To my knowledge, the only 'deal' GW has ever offered money wise is the 'Start collecting' boxes. I'm not sure they would want to scale that kind of discount up to the point where there would be a commensurate discount on 2k point armies.

1. they have like 6 bundles for different primaris. What if each one of them added something to a good army. Stuff would be modular like an Ikea kitchen. They would be able to sell even more, as there wouldn't be any trap books that make people buy 2-3 untis, find out the army is bad and quit.

2. they aren't good now. They maybe wouldn't be good, as in general good, in a modular build army. But what stops GW from making a detachment with centurions that makes them good? nothing. As long as all armies would get some kind of such detachments, one specific unit being bad in a general sense wouldn't be a problem. Maybe strike scorpions suck, and no one wants to take them in an eldar army, but there is this scorpion theamed detachment which makes them work real fine.

3. Deals don't have to be monetary. Plus the start collecting aren't really good deals most of the time. Being given an HQ you wil never use and a unit or vehicle you will never use for free isn't a good deal. It is a trap for new players. But if a start collecting had 3-5 units people actually did want to use, or a deal would come with a nice rule bundle people would buy it. Just to avoid the hassle of hunting one specific unit.

I dunno. I just got the S.C. box for the skeletons in AoS, in which you pretty much get Arkhan for free...

 Peregrine wrote:

 JNAProductions wrote:
I wouldn't say effectively impossible.

I'd agree that making a formula that creates, without fail, a well-balanced unit is effectively impossible, but a formula that gives a good baseline with which to work off of is probably doable.

Try doing it and you'll see what I mean. The complexity of even a moderately accurate formula is going to be way too high to be practical. For example, you might think of having a price per point in a stat, but BS 2+ is way more powerful on a shooting unit than on a unit with only a token bolt pistol for ranged weapons. Movement speed is far more valuable on melee units or units that want to move up to claim objectives than on a static gunline unit. Etc. Trying to capture any meaningful percentage of those differences inevitably leads to getting bogged down in a convoluted formula and tables of special-case exceptions and modifiers and such. It's certainly going to be way less efficient than just eyeballing the point costs based on experience playing the game and then going through a few rounds of iterative playtesting.

The last game I saw that tried to properly take account of those sorts of interactions was Heavy Gear 2nd edition. That involved square and cube roots at various stages in the calculation, and you ended up with three points values for any unit - offensive, defensive and overall.

Karol wrote:
1. they have like 6 bundles for different primaris. What if each one of them added something to a good army. Stuff would be modular like an Ikea kitchen. They would be able to sell even more, as there wouldn't be any trap books that make people buy 2-3 untis, find out the army is bad and quit.

2. they aren't good now. They maybe wouldn't be good, as in general good, in a modular build army. But what stops GW from making a detachment with centurions that makes them good? nothing. As long as all armies would get some kind of such detachments, one specific unit being bad in a general sense wouldn't be a problem. Maybe strike scorpions suck, and no one wants to take them in an eldar army, but there is this scorpion theamed detachment which makes them work real fine.

3. Deals don't have to be monetary. Plus the start collecting aren't really good deals most of the time. Being given an HQ you wil never use and a unit or vehicle you will never use for free isn't a good deal. It is a trap for new players. But if a start collecting had 3-5 units people actually did want to use, or a deal would come with a nice rule bundle people would buy it. Just to avoid the hassle of hunting one specific unit.

1. IKEA? You said it would be easier to put together!!!

2. That's actually a good idea. Except for it skirts dangerously close to what PP does with Warmahordes, which is HERESY!

3. To be a deal, it does by definition have to save a bit of money. However, I see what you're saying. The problem is, to do this GW will have to at least tacitly admit that their game does not have balance, and that some units are better than others.

Tyel wrote:
I feel these threads always end up with "you need a super computer" - but to some extent I feel that's nonsense.

To a very real extent it's not.

I studied Physics - I can give you a pretty simple algorithm for gravity. You think therefore that if I want to calculate the movement of Jupiter's moons I could give you an algorithm for that; maybe a complex one with different terms for each of the moons. Not so. That problem cannot be solved by any algorithm we can come up with - but it can be very accurately solved by running a simulation on a computer.

There are simply some categories of problem that you cannot solved except by running a simulation. The n-body problem in celestial mechanics is probably the one which is most studied and one which I think has a lot in common with trying to balance 40K

We can perfectly solve the 2-body problem in celestial mechanics with an algorithm. For an n-body problem with more than n=2 we cannot. Not only that but while we can solve any number of 2-body problems we cannot combine them into a valid solution to the n-body problem.

Similarly we can solve any 2-unit balancing problem in 40K, we can also do that for an arbitrarily large number of 2-body combinations. What we cannot then do is combine all those solutions into a valid n-unit solution, the rules of maths and logic just do not work that way for us. In general the solution for n>2 is chaotic, which is super-exciting for people who study such systems but really not helpful for assigning an accurate points cost to a tactical marine

So it needs a computer simulation. As the number of interactions is rather high it has a huge computational cost, which is why those who know about such things talk in terms of it needing a very big computer.

tl;dr algorithmic solutions have limits, this is beyond those limits.

happy_inquisitor wrote:

Tyel wrote:
I feel these threads always end up with "you need a super computer" - but to some extent I feel that's nonsense.

To a very real extent it's not.

(Very intelligent position)

tl;dr algorithmic solutions have limits, this is beyond those limits.

I believe that when we get to that level of dissection, the numbers cease to have meaning to us as gamers. Within the context of the game mechanics, there would very likely not be any practicable use for the results defined by such a process. I would be very interested to see what resulted, but beyond intellectual interest, I dont feel it would help us determine if and to what degree the game is balanced.

Essentially, it was created by some gamer buddies sitting around a table at GW brainstorming some epic sh*t. On the Beasts of War interview about the kickstarter for Antares, Rick Priestly pretty much says this.

Now that I think about it, though, what my friend and I are trying to do is more or less for intellectual interest, though...

The point is, I'd like to see if we can come up with a model that can be used to convert 7th ed (Horus Heresy) stats to 8th ed stats that casual opponents won't feel are broken or gratuitously OP.

Hell, it may allow us to convert even older characters from previous editions of 40k that are plausibly balanced.

I know this is a very convoluted way of going about it, but if we can come up with a sort of formula, it might even benefit GW in the fact that some of the badass FW models might sell slightly better when people can convert them to play in the newer versions.

 CATACLYSMUS wrote:


I believe that when we get to that level of dissection, the numbers cease to have meaning to us as gamers. Within the context of the game mechanics, there would very likely not be any practicable use for the results defined by such a process. I would be very interested to see what resulted, but beyond intellectual interest, I dont feel it would help us determine if and to what degree the game is balanced.

Essentially, it was created by some gamer buddies sitting around a table at GW brainstorming some epic sh*t. On the Beasts of War interview about the kickstarter for Antares, Rick Priestly pretty much says this.

Now that I think about it, though, what my friend and I are trying to do is more or less for intellectual interest, though...

The point is, I'd like to see if we can come up with a model that can be used to convert 7th ed (Horus Heresy) stats to 8th ed stats that casual opponents won't feel are broken or gratuitously OP.

Hell, it may allow us to convert even older characters from previous editions of 40k that are plausibly balanced.

I know this is a very convoluted way of going about it, but if we can come up with a sort of formula, it might even benefit GW in the fact that some of the badass FW models might sell slightly better when people can convert them to play in the newer versions.

I rather think that GW already have some form of rough guide to points values to which they then add human judgement and some feedback from play testing. Essentially what they have is a process rather than just an algorithm and it is one which uses human judgement rather than supercomputers.

There is an article about points values in AoS in next month's white dwarf, they might give us a little more insight. I expect the process across the two games is fundamentally similar.

tneva82 wrote:

 JNAProductions wrote:

 Peregrine wrote:

 JNAProductions wrote:
I'm not a mathematician, nor a paid game designer. I'd hope that people who ARE paid to do this kind of thing are better than me at it.

It's not a question of skill, it's about the complexity of the problem relative to the alternative approach. It doesn't matter how good you are at math or game design when the task is unreasonable.

Are you a paid game designer or mathematician?

Because, to my knowledge, you're just a random 40k fan. I know that it'd be difficult to do, since I cannot do it myself, but your assumption that it's TOO difficult is something I'd believe when an expert says so.

It's actually easily realized with logic that algorithm leads to broken system. Value of movement for example changes a lot based on what kind of unit are we talking about. M1 h2h specialist? Worthless. M1 heavy gun toting wielder? Low movement isn't much of disadvantage. Especially if you then factor in unit-external things like ability to move unit in otherwise. For example how much lootas from orks should drop in points if you dropped them to M1? Any algorithm would say points. Any ork player would trade all the movement for point drop. They aren't moving all that much to begin with and if they do it's generally with da jump because you need hefty repositioning to ensure LOS. The loss of movement is irrelevant. But sure if you think it's possible try to figure out how to factor in that movement drop while then also making it work for say meganobs. How much current meganobs should go up in price if you upped their movement to 6?

Then factor in terrain, opponents faction, units etc which should affect point cost and kiss goodbye to algorithm.

The easy way to correctly make an algorithm here would probably not to use movement, as the stats dictate, but threat range.
As in : a heavy bolter totting guardsman with a Mov of 1'', has a threat range of 37''. A mega-mutilator engine with only melee weapon and a Mov of 1, has a threat range of more or less 8.5.

I think you may be right about the way they do things with the points values. Moreover, it seems that in 8th, they removed some variables (s/t table, no more initiative) to make things more discrete.
I'm looking forward to reading that article. Especially since the 40k rules have come more in parallel with the AoS rules.

The threat range idea is exactly what I am saying. That, rather than calculating all potential moves would pull the estimates down from the supercomputer range.

VoidSempai wrote:

tneva82 wrote:

 JNAProductions wrote:

 Peregrine wrote:

 JNAProductions wrote:
I'm not a mathematician, nor a paid game designer. I'd hope that people who ARE paid to do this kind of thing are better than me at it.

It's not a question of skill, it's about the complexity of the problem relative to the alternative approach. It doesn't matter how good you are at math or game design when the task is unreasonable.

Are you a paid game designer or mathematician?

Because, to my knowledge, you're just a random 40k fan. I know that it'd be difficult to do, since I cannot do it myself, but your assumption that it's TOO difficult is something I'd believe when an expert says so.

It's actually easily realized with logic that algorithm leads to broken system. Value of movement for example changes a lot based on what kind of unit are we talking about. M1 h2h specialist? Worthless. M1 heavy gun toting wielder? Low movement isn't much of disadvantage. Especially if you then factor in unit-external things like ability to move unit in otherwise. For example how much lootas from orks should drop in points if you dropped them to M1? Any algorithm would say points. Any ork player would trade all the movement for point drop. They aren't moving all that much to begin with and if they do it's generally with da jump because you need hefty repositioning to ensure LOS. The loss of movement is irrelevant. But sure if you think it's possible try to figure out how to factor in that movement drop while then also making it work for say meganobs. How much current meganobs should go up in price if you upped their movement to 6?

Then factor in terrain, opponents faction, units etc which should affect point cost and kiss goodbye to algorithm.

The easy way to correctly make an algorithm here would probably not to use movement, as the stats dictate, but threat range.
As in : a heavy bolter totting guardsman with a Mov of 1'', has a threat range of 37''. A mega-mutilator engine with only melee weapon and a Mov of 1, has a threat range of more or less 8.5.

That doesn't work, since threat range is interrupted by LoS -- if a heavy Bolter with a mov of 1 wants to see around a corner, he's going to have to advance to move that fast assuming he can still advance, while the melta gun with the threat range of 12 and move 6, will still be able to move 6 and then fire 12 inches away and get past cover.

Movement is more than just moving towards the opponent, it's repositionjng and the ability to open and close firelines. Which is why it's a hard stat to value properly, especially since Orks have the Jump, GK have GOI, and etc.

You can't make a good algorithm for the sheer scale of 40k -- one that can handle a ratlings sniper with s2 and t2 and then an imperial fortress with t10 and 40(?) Wounds.

Game design is tricky because as others have mentioned it is non-trivial to create a system that accounts for every special action, weapon being used, and force multiplier.

Ultimately the best guide is existing units and then measuring point efficacy against those. Also, with more play stats available from tourneys they can also see better which combox are being favored over another.

Personally I'd just suggest looking at some Blizzard videos when they are discussing balancing and the reasons they do X instead of Y. They see a problem, they create a treatment, test it internally, and then throw it into the wild where it may work as intended or create something horrible instead; which is always a problem when you have a limited team of n people against the horde of players who will use everything they can to best the system.

This topic is very near to the mammoth task I had set out for myself.

Every new edition I have to "convert" our tired old 2nd edition Squat codex to the new, in this case, 8th has been better and hard all in one.
I am pretty clever with Excel so the hard part is getting all the info into tables without it taking an eternity.
The folks who support Battlescribe 40k have my utmost respect, digging into some of the notes in there helped with this project.

How do we figure out the "proper" points value so my friend's Squats are not super-powered or garbage? = Comparative or equivalent point costing.

So I list off the various weapons with all the insane stats and rules and their points values.
Problem is, they are costed by what BS or WS the likely used has (and are stated differently sometimes exactly so). <edit> BUT I got a direct comparison weighting of points vs BS/WS values.
This alone is problematic.
Then also some weapons are "0 points" and their cost is rolled into the model cost so you ignore it or go down that rabbit hole.
Lastly, you "try" to fabricate an algorithm that takes into account all the factors and tracks reasonably well with the existing points values (applying weighting of stats etc.).
This then allows you to re-create unsupported weapons and assign a "reasonable" points cost which at least will emulate the assigned costing of the moment.
Multiple profiles on a weapon was a particularly hard item to nail down.
Funny enough, I found any flame weapon to have the most variation and inconsistencies in points from army to army (roughly tracking with bad BS = higher cost).

Weapons are the "easier" part.

Where my brain almost melted is when you get all these special rules with units, the keywords are an awesome thing but make costing almost impossible (for me) to adequately account for.
What poses the most resistance to costing is units with aura abilities.
They are a huge force multiplier that is the linchpin for any competitive army.
Their costing is insane on the variables to account for largely based on what stat they buff and how "needed" it is (+1BS on a shooty army is huge +1WS not quite as much).

Anyway... I am sure a "reasonable" approximation can be fabricated but I have not seen or created anything that fits perfectly with their points scheme.
<edit>The "project" got us most of the way there to help shape our "arguments" and then we "eyeballed" the rest as best we could.

There is no master algorithm I can discern but does not mean it is not there.
"Random" can mean an algorithm so complex it might as well be unpredictable outcomes (further "cheating" with a seed randomizer thrown in).

Karol wrote:
the start collecting aren't really good deals most of the time. Being given an HQ you wil never use and a unit or vehicle you will never use for free isn't a good deal. It is a trap for new players.

I don't know about that. I have the Start Collecting! SM box and I still use the HQ that came with that. A Terminator Captain with Iron Resolve, the Sanctic Halo and a relic blade makes for a pretty tough warlord with a nice re-roll aura. In recent games he's been more than capable of taking on a Daemon Prince, a Necron Overlord and and an Ork Warboss single-handed.

Karol wrote:
... Plus the start collecting aren't really good deals most of the time. Being given an HQ you wil never use and a unit or vehicle you will never use for free isn't a good deal. It is a trap for new players. But if a start collecting had 3-5 units people actually did want to use, or a deal would come with a nice rule bundle people would buy it. Just to avoid the hassle of hunting one specific unit.

Off the top of my head Artemis is the only model in any Start Collecting box I actually never see on the table (mostly because Deathwatch lists tend to want a Watch-Master for their first HQ, a Librarian or a budget Captain (possibly packing the Dominus Aegis) for their second, and no third HQ because they don't have the budget for more HQs after you get to a Battalion). You'll usually find some kind of use for one of everything in your army's Start Collecting, and if you're playing Tau, Daemons, Dark Eldar, Guard, MT, Necrons, or AdMech the box represents at least small savings over buying all of the non-HQ components that you might use at least two of.

there is no use for any of the marine HQs in termintor armor. And there is one in the marine, BA and CSM set. The SW one has TWC, the sb leader can at least be turned in to some sort of RP or something, but you will never use the TWC.
no idea about other starter sets though, maybe they are good. I was told they are a trap, because they give you a "free" model you will never use.


Automatically Appended Next Post:

 Brother Castor wrote:

Karol wrote:
the start collecting aren't really good deals most of the time. Being given an HQ you wil never use and a unit or vehicle you will never use for free isn't a good deal. It is a trap for new players.

I don't know about that. I have the Start Collecting! SM box and I still use the HQ that came with that. A Terminator Captain with Iron Resolve, the Sanctic Halo and a relic blade makes for a pretty tough warlord with a nice re-roll aura. In recent games he's been more than capable of taking on a Daemon Prince, a Necron Overlord and and an Ork Warboss single-handed.

If you run a car park then azrael is better, or Gulliman. Even if you do take a chapter master, why give him a termintor armor? Now GK don't have the option to not take one, but everyone else would go for the power armored version. The smaller base alone is better, because it is easier to hide and block from charges .

Karol wrote:
there is no use for any of the marine HQs in termintor armor. And there is one in the marine, BA and CSM set. The SW one has TWC, the sb leader can at least be turned in to some sort of RP or something, but you will never use the TWC.
no idea about other starter sets though, maybe they are good. I was told they are a trap, because they give you a "free" model you will never use...

If you find the 21pt difference between an 85pt Captain with relic blade and storm bolter and a 106pt Terminator Captain with relic blade and storm bolter takes something from being "fine" to "completely unusable" I think you might want to take a step back from the tournament optimization for a minute and try pretending to think like a new player.

As for the other starter sets most of them tend to be about $10 cheaper than the components without the HQ, so whether or not the HQ is a trap (and most of them aren't) is immaterial. The Tau used to be a trap because the Crisis suits were always worse than equivalent points of Ghostkeels/Riptides in the same slot, but the price drop means they're actually somewhat usable.

Karol wrote:
there is no use for any of the marine HQs in termintor armor.

...

If you run a car park then azrael is better, or Gulliman.

Azrael isn't an Ultramarine and of course Guilliman is better - he costs 400 points vs 106 points

 AnomanderRake wrote:
If you find the 21pt difference between an 85pt Captain with relic blade and storm bolter and a 106pt Terminator Captain with relic blade and storm bolter takes something from being "fine" to "completely unusable" I think you might want to take a step back from the tournament optimization for a minute

Agreed. For club nights and casual play a Terminator Captain is just fine.

tneva82 wrote:

Then factor in terrain, opponents faction, units etc which should affect point cost and kiss goodbye to algorithm.

You shouldn't need to factor in terrain or opponent factions.

These are the variables that makes a game instead of two people rolling dice at each others faces for 2 hours.

IMHO its entirely possible to algorithm out points assuming you are starting mostly from scratch or basic concepts. sure you could set options for m1 or m infinity but you also should be curating and adjusting to close enough through play testing and a wee bit of critical thinking not just taking the robots word for it.

For a predictive formula to work it needs to be a combination of several equally complex formulae.
Formula 1: durability value. A function of toughness armour save and wound. Possibly also model size and unit size.
Formula 2: offensive melee capability. A function of strength, attacks, weapon skill, movement and weapon stats.
Formula 3: offensive ranged capability. A function of ballistic skill, movement, and weapon stats. Possibly also model size.
Formula 4: special rules

Then the true algorithm compares all the above to predict a points value.