Lemon Knuckles Posted November 17, 2018 Share Posted November 17, 2018 Alright lads. Let's get to work. Spreadsheet attached, and full definitions included. Will try to keep this post updated with questions, changes, etc. as we go. Project Description: Spoiler The premise of this project is simple: look to create a method to the madness. Unit point costs are a foundational part of the game, but no one seems to know why units cost what they do, and consequently, few can agree on what they think units should cost. Much of the discussion on evaluating points costs is based on intuition, circumstance, differing experiences in-game, and haphazard comparisons with other units subjectively considered to be better or worse. As the number of available units expands through time, as warscrolls change, and as periodic and somewhat haphazard points adjustments are rolled out, the situation threatens to get even more confusing. It doesn’t have to be this way. This project is not about “balancing the game,” as that would be well beyond the mechanics of points alone. At its most ambitious, however, it could aspire to be about helping to remove a level of imbalance from the game that comes from flawed point costing. But that kind of discussion is entirely contingent on the immediate effort being successful, and so is a discussion best left to another time. F.A.Q.s Spoiler What’s the point? Exactly. Haha, funny guy. No really, what’s the point of this project? This project is meant to explore the possibility of creating a rational and consistent basis for points costing. It is first and foremost an intellectual puzzle and a curiosity. Right now, going through the process is the point: what insights and learning can be gained from the journey. How can I help? AWESOME! That’s really appreciated, and very needed! Let’s see… Are you: Mathematically inclined? You could help validate the formulas, suggest better formulations, help refine key user-inputed variables, and just generally identify ways to make it better, more accurate and more elegant… A whiz with a spreadsheet? You could help make the damn thing easier to manage, more automated, better organized, more user friendly, more future-proof… A computer programmer or project manager? You could help make sure that our methodology is logical, improve our process and generally protect us from ourselves… An experienced AoS player? You could help evaluate the output points values for reasonability, identify things that make no sense, suggest units that should be in our sample list, make sure we are interpreting warscrolls correctly… Have nothing but scorn for this project? Well, you can still help by not trolling or griefing or creating unnecessary drama in this thread! Design Principles Spoiler Elegance: strive to articulate the math in its simplest and simultaneously most effective form. Fidelity: strive to engineer the formula in a way that renders it as faithful as possible to current GHB18 values. We want to improve the consistency and fix any obvious errors with the current points costing; we do not want to create an alternate and competing system. Proposed Stages Spoiler Project proceeds through three levels. Levels must be approached sequentially, and a level must be resolved to consensus satisfaction prior to engaging the next level. Level One Scope: Warscroll data only. No consideration of unique tactical abilities that do not quantify directly to the warscroll’s statline. Level Two: Warscroll data only. Consideration of unique tactical abilities that do not quantify directly to the warscroll’s statline. Level Three: Additional non-warscroll data considered (e.g. allegiance abilities and synergies, base size, etc.). Proposed Level One Methodology Spoiler Formula-driven All points values should be derived from one consistent and universal formula. No introducing variance by tweaking values or doing things differently on a warscroll by warscroll basis. Changing the points value of a particular unit by tweaking an assumption or variable in the formula therefore has a cascading effect across the entire range of units. Think about a water-balloon. If you push in one area, another area of the balloon bulges out in response. The lack of this kind of internal integrity is the principle flaw with the current pointing methodology. The one exception to this are unique special abilities. These are generally tactical abilities that should have points value, but do not clearly correlate to a unit’s fundamental stat-line. Think of Khinerai Heartrenders Descend to Battle ability as an example. There are placeholders in the formula for these kinds of abilities, but determining their value is a process of trial and error, gameplay experience and community consensus, and the changing of this kind of variable does not have a cascading effect on the calculated costs of other units. Recommended to leave these considerations to level two. Limit the variables We consider only a unit’s warscroll for the purposes of the formula. Its stat line and its listed abilities. We do not consider allegiance abilities, artefacts, etc. Potential buffs that could effect the unit coming from a different warscroll should be calculated for the unit whose warscroll has the ability listed, not the unit who could potentially be the target. Limit the test units Create a workable sample list of 30-50 (??) units to evaluate the formula output. The list should contain a representative variety of unit types (horde, monster, range, melee, caster, flyer, fast, slow, elite, different alliances and factions). The sample list should have enough unit variety to properly evaluate the reasonability of the formula in sufficient context without becoming overwhelming or unmanageable. Evaluate Review the output points values for the sample list units and evaluate. Look for patterns among any units that seem obviously over/under costed (unit type, ability types, etc). Use any insights to inform targeted incremental changes to key formula assumptions. End Continue to Iterate the formula by incrementally tweaking key assumptions until an output is produced that: Does not radically deviate from current GHB18 unit points values (in the overall view; individual outliers are to be expected, and are OK!) Achieves an informed consensus view of the output “that makes sense/looks good” (aka passes the sniff test) The Formula Spoiler ((Expected Damage Output + Effective Wounds) + SpellCasting + Command Ability + Other Ability) * (1 + Movement Modifier + Fly Modifier) Key User Inputs Spoiler These are the values that ultimately determine how unit stats and abilities convert into points. These are the values to be adjusted to calibrate the points-costing. PtsD3: Point Value per point of Expected Damage vs Saves of 3+ PtsD4: Point Value per point of Expected Damage vs Saves of 4+ PtsD5: Point Value per point of Expected Damage vs Saves of 5+ PtsD6: Point Value per point of Expected Damage vs Saves of 6+ PtsW0: Point Value per Effective Wound versus Rend 0 PtsW1: Point Value per Effective Wound versus Rend 1 PtsW2: Point Value per Effective Wound versus Rend 2 RM: Modifier for points value of Expected Damage based on range MoveBase: Baseline move value at which no points modifier is applied MoveModValue: Variable used to determine points modifier for movement above/below Move Base, expressed as denominator FlyModValue: Variable used to determine increased movement points modifier for units with fly SpellValue: Points value per spellcast per turn CommandValue: Points value for presence of a Command Ability Weighting: Used to determine what weight a particular unit state is given in terms of the warscroll as a whole (for example, different weapon options, abilities which modify a weapon’s stat-line “after the unit completed a charge this turn,” etc., etc.). Changelog Spoiler Version 1.03 Change Summary: Changed formula for calculating Variance%(GunkPoints-GHB18Points)/GHB18 Points Added buckets for aggregating degrees of variance%>25% over/under = Very over/under-costed, 10%-25% = Over/under-costed, 5%-10% = Slightly Over/under-costed, 0-5% = Spot On Added Reference fieldsAllegiance, Faction, Hero, Wizard, Monster, Battle-line, Artillery, Unique, GHB17 points cost, change, variance, rating Added units: Grot Rock Lobber, Orruk Ardboyz, Orruk Brutes, Spirit Hosts, Dragon Ogres, Chaos Mauraders, Crypt Ghouls, Sisters of the Thorn, Evocators, Kroxigor, Ripperdactyl Riders, Fungoid Cave-Shaman, Necromancer, Beastlord, Battlemage, Freeguild General, Hammerers, Abhorrent Ghoul King, Wight King on Skeletal Steed, Vampire Lord on Nightmare, Arch-Warlock, Butcher Version 1.02 Change Summary: All Sheets/Rename Unit Morghast to Morghast Archai Effective Wounds/Rename field Ward Save to Damage Save Effective Wounds/Create new field MW Save, express as a % Effective Wounds/Create new field WoundsMW, formula= (Models*Wounds)/(1-MW Save) Formula Values/Create new assumption PtsMW (starting value = 12?) Effective Wounds/Create new field PVMW, formula = (WoundsMW*PtsMW) Effective Wounds/Modify formula of field PV Wounds to include PVMW Added Units: Sequitors, Ironblaster, Huskard on Thundertusk, Arkanaut Company, Tzaangor Skyfires, Judicators ReferenceConceptLabelCurrent Value 1Point Value per point of Expected Damage vs Saves of 3+PtsD350 2Point Value per point of Expected Damage vs Saves of 4+PtsD438 3Point Value per point of Expected Damage vs Saves of 5+PtsD526 4Point Value per point of Expected Damage vs Saves of 6+PtsD614 5Point Value per Effective Wound versus Rend 0PtsW06 6Point Value per Effective Wound versus Rend 1PtsW18 7Point Value per Effective Wound versus Rend 2PtsW210 8Modifier for points value of expected damage based on rangeRM20 9Baseline move value at which no points modifier is appliedMove Base8 10Variable used to determine points modifier for movement above/below Move Base, expressed as denominatorMoveModValue40 11Variable used to determine increased movement points modifier for units with flyFlyModValue20% 12Points value per spellcast per turnSpellValue50 13Points value for presence of a Command AbilityCommandValue25 14Points per MWPtsMW12 ReferenceUnitNameTotalGH2018DifferenceVar% 1Morghast Archai180220-4022% 2Grave Guard82802-2% 3Grimghast Reapers17514035-20% 4Hexwraiths1671607-4% 5Black Knights1231203-3% 6Akhelian Morrsarr Guard18914049-26% 7Witch Elves13210032-24% 8Liberators1051005-5% 9Celestar Ballista1011001-1% 10Bloodletter105120-1514% 11Great Unclean One326340-144% 12Megaboss on Maw-krusha336440-10431% 13Coven Throne28226022-8% 14Mortis Engine150180-3020% 15Bloodseeker Palanquin236320-8436% 16Blood Knights225240-157% 17Fulminators25924019-8% 18Dire Wolves766016-21% 19Kurnoth Hunters21720017-8% 20Endrinriggers14512025-17% 21Tzangor Enlightened1631603-2% 22Savage Boar Boys83100-1721% 23Akhelian Allopexes134140-64% 24Putrid Blightkings1621602-1% 25Vampire Lord on Zombie Dragon437440-31% 26Irondrakes76180-104136% 27Stormvermin138140-21% 28Bestigors16312043-27% 29Namarti Thralls14114010% 30Vargheists17116011-6% 31Savage Orruk Arrowboys15714017-11% 32Sequitors14312023-16% 33Ironblaster115120-54% 34Huskard on Thundertusk297360-6321% 35Arkanaut Company15312033-21% 36Tzaangor Skyfires200220-2010% 37Judicators159160-11% Version 1.1 Current Key Assumption Values: PtsD3: 50 PtsD4: 38 PtsD5: 26 PtsD6: 14 PtsW0: 6 PtsW1: 8 PtsW2: 10 RM: 20 Move Base: 8 MoveModValue: 40 FlyModValue: 20% SpellValue: 50 CommandValue: 25 Current Output: UnitName/Total/GH2018/Difference/Var% Morghast/180/220/-40/22% Grave Guard/82/80/2/-2% Grimghast Reapers/175/140/35/-20% Hexwraiths/167/160/7/-4% Black Knights/123/120/3/-3% Akhelian Morrsarr Guard/189/140/49/-26% Witch Elves/132/100/32/-24% Liberators/105/100/5/-5% Celestar Ballista/101/100/1/-1% Bloodletter/105/120/-15/14% Great Unclean One/326/340/-14/4% Megaboss on Maw-krusha/336/440/-104/31% Coven Throne/282/260/22/-8% Mortis Engine/150/180/-30/20% Bloodseeker Palanquin/236/320/-84/36% Blood Knights/225/240/-15/7% Fulminators/259/240/19/-8% Dire Wolves/76/60/16/-21% Kurnoth Hunters/217/200/17/-8% Endrinriggers/145/120/25/-17% Tzangor Enlightened/163/160/3/-2% Savage Boar Boys/83/100/-17/21% Akhelian Allopexes/134/140/-6/4% Putrid Blightkings/162/160/2/-1% Vampire Lord on Zombie Dragon/437/440/-31% Irondrakes/76/180/-104/136% Stormvermin/138/140/-2/1% Bestigors/163/120/43/-27% Namarti Thralls/141/140/1/0% Vargheists/171/160/11/-6% Savage Orruk Arrowboys/157/140/17/-11% Current Issues: Need a better list of sample units; too many units with unique abilities that are currently uncosted and skewing results Something wrong with Irondrakes? AoS Gunkulator v1-03.xlsx Link to comment Share on other sites More sharing options...
The Jabber Tzeentch Posted November 17, 2018 Share Posted November 17, 2018 Whilst I admire the scope of this and the very well put together post, it has been attempted by many people (including myself) many times. I don’t want to say it’s impossible but... It’s hugely subjective with a lot of the warscroll rules, and some units are only good at certain sizes and when combined with other units which may or may not be in the same allegiance, and then are better against some armies but not others... etc... It’s really evident from design that a deep mathematical approach has not been used to create the GHB point values. I really don’t think it would help to be honest it would only move the bar to another area, making different units optimal. But enjoy the exercise of doing it, it’s fun and I still tinker with my spreadsheets when I’m mathhammering lists but I don’t use it as a be all and end all. Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 17, 2018 Author Share Posted November 17, 2018 Hi Jabber. I debated whether to respond to this, not because it's an unfair comment (it isn't), but simply because 1) this being the internet and 2) hobby passion being what it is, I don't know that I want to risk derailing this particular thread by turning it into a meta-discussion about the merits or no of the enterprise. Variations of that conversation have occurred recently in some other threads, and I'd be happy to discuss them further elsewhere if needed. Preferably, I'd like to keep this thread reasonably focused on collaborating with others trying to solve real problems and questions as they arise and trying to build something. Ultimately I've deciding to respond anyway, for the benefit of those who may think similarly to you, and also for whatever it's worth to those who might be curious about the project, but discouraged by your comments. I'll touch briefly on a few of the key issues you raised. 3 hours ago, The Jabber Tzeentch said: Whilst I admire the scope of this and the very well put together post, Thanks! I did put effort into that post. 3 hours ago, The Jabber Tzeentch said: I don’t want to say it’s impossible but... It's definitely is not impossible at all. It's just a cipher, a way of translating from one language (warscroll values) to another (points values). There is nothing impossible about it in the least. Maybe you mean it's hard to make it useful. That's a possibility, but I don't think so. Maybe you mean it will be hard to get people to agree. That's definitely a possibility, but I would say that about almost anything nowadays. I think you probably mean that it will be impossible for points to capture all the circumstances of the game and to create a point system where a unit worth 200 points is always better than a unit worth 180 points. I definitely agree with that, but that's not at all the point of the exercise. We have a points system. I want to explore it deeper, understand it more and hopefully even find ways to make it better. 3 hours ago, The Jabber Tzeentch said: It’s hugely subjective with a lot of the warscroll rules, and some units are only good at certain sizes and when combined with other units which may or may not be in the same allegiance, and then are better against some armies but not others... etc... Yes... and no. How do you eat an elephant? One bite at a time. Compartmentalize the problem. One of the benefits of this kind of exercise is that it forces you to really think through the game. For example, when I started to think about how I could build a cypher like this, I tried to distill things down to what's really most important, and why? Ultimately I came to three things: the ability to exert pressure, the ability to withstand pressure and the capacity to extend those abilities through space. Most of the things in the game can be thought through in relation to these three things. It makes each bite easier. 3 hours ago, The Jabber Tzeentch said: But enjoy the exercise of doing it, it’s fun and I still tinker with my spreadsheets when I’m mathhammering lists but I don’t use it as a be all and end all. Thanks! And that is really the point right now. Going through the process. It is insightful, it is interesting, it is thought-provoking. I'm not worried about where it will end up; I'm just looking forward to the ride. Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 17, 2018 Author Share Posted November 17, 2018 Recommendation (Location/Action): All Sheets/Rename Unit Morghast to Morghast Archai Effective Wounds/Rename field Ward Save to Damage Save Effective Wounds/Create new field MW Save, express as a % Effective Wounds/Create new field WoundsMW, formula= (Models*Wounds)/(1-MW Save) Formula Values/Create new assumption PtsMW (starting value = 12?) Effective Wounds/Create new field PVMW, formula = (WoundsMW*PtsMW) Effective Wounds/Modify formula of field PV Wounds to include PVMW Reference Cases: Morghast Archai, Ebon-wrought Armor Great Unclean One, Blubber and Bile Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 18, 2018 Author Share Posted November 18, 2018 Input Needed Need suggestions on what to include in a functional list of test units to evaluate ongoing iterations of the gunkulator. 30-50 (??) units total The list should contain a representative variety of unit types (horde, monster, range, melee, caster, flyer, fast, slow, elite, different alliances and factions) in order to help identify and isolate patterns with the output (e.g. are we over-valuing range, under-valuing flying, etc.) Units that have broad consensus of opinion around their current GHB18 points values (either as correct, or as over/under costed) are especially useful to help the process Should not nominate units that have particularly abstract unique special abilities that are hard to quantify (see Khinerai Heartrenders, Desccend to Battle). These are better left to Phase Two. Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 18, 2018 Author Share Posted November 18, 2018 Validation Needed Jabber raises a real issue: 19 hours ago, The Jabber Tzeentch said: It’s hugely subjective with a lot of the warscroll rules, and some units are only good at certain sizes and when combined with other units which may or may not be in the same allegiance, and then are better against some armies but not others... etc... The gunkulator currently attempts to deal with this through weighting. Example and discussion in spoiler: Spoiler A single unit is capable of have multiple states, with each state carrying a different stat-line. This gives one unit a potential range of possible points values depending on game state. The simplest example is when a unit has different weapon load-out options (Grave Guard). This is captured in the Weapon Profile field. Other kinds of examples of state difference are captured in the Dimension field, and include proximity to other units (Bloodletter, Locus of Fury), unit size (Bloodletter, Murderous Tide) and after certain actions have been taken (Blood Knights, Martial Fury). By weighting each state, we are able to collapse the range of points values for a unit to a single, calculated value. I think this is a reasonable way of tackling the problem, and I think it is faithful to actual gameplay. Good generals will try to keep their Bloodletters within 8" of a Demon, or try to avoid getting their Blood Knights bogged down in combat without first getting off a charge. Using weighting to point-cost units reflects this: I am effectively putting extra points on the tabletop the more I am able to trigger the conditions that make my most valuable states operative. I am effectively losing points the more I fail to do so. Looking for validation, thoughts, suggestions around weighting and current weighting values: Weapon Profiles: equal weighting across all available profiles Grave Guard (50% Wight Blade and Crypt Shield, 50% Great Wight Blade) Liberator (not prime) (25% Warhammer & Shield, 25% Warblade & Shield, 25% pair of Warhammers, 25%, pair of Warbaldes) Target Unit Generic: 50% Grimghast Reapers, Reaped Like Corn (50%) Namarti Thralls, Sweeping Blows (33.3%, 33.3%, 33.3%) Liberators, Lay Low the Tyrants (50%) Tzaangor Enlightened, Guided by the Past (50%) Target Unit Specific: 25% weighting Savage Orruk Arrowboys, Aim For Its Eyes (25%) Bestigors, Despoilers (vs. Order part) (25%) Unit Size: 50% Bloodletters, Murderous Tide Savage Orruk Arrowboys, Loadsa Arrows Unit Action: 33.3% Black Knights, Deathly Charge Bestigors, Beastial Charge Link to comment Share on other sites More sharing options...
Kramer Posted November 18, 2018 Share Posted November 18, 2018 Read it all, cool idea well thought out. But one thing I keep doubting about stuff like this... how do you value a specific roll in an army that has a focus on something else. So practical example. There was a conversation until the destruction thread about the ogor ironblaster. How does a high potential gun value in a close combat focussed army? In my mind a cannon in a melee focussed faction should be relatively more expensive. Likewise in a glass army anything with a high save should be relatively more expensive. So a 16” range 4+ 4+ archer in a khorne army should be more expensive than in a free guild army, do you agree and how do you want to value these element? Because it should be in tactical stage 2 but I feel it’s a good thing GW doesn’t try to quantify these things as it will lead to all factions doing the same. Link to comment Share on other sites More sharing options...
Retro Posted November 18, 2018 Share Posted November 18, 2018 24 minutes ago, Kramer said: Read it all, cool idea well thought out. But one thing I keep doubting about stuff like this... how do you value a specific roll in an army that has a focus on something else. So practical example. There was a conversation until the destruction thread about the ogor ironblaster. How does a high potential gun value in a close combat focussed army? In my mind a cannon in a melee focussed faction should be relatively more expensive. Likewise in a glass army anything with a high save should be relatively more expensive. So a 16” range 4+ 4+ archer in a khorne army should be more expensive than in a free guild army, do you agree and how do you want to value these element? Because it should be in tactical stage 2 but I feel it’s a good thing GW doesn’t try to quantify these things as it will lead to all factions doing the same. I've put a bit of thought into that myself. Personally I thought having a factor of some sort to apply to units that quite obviously exist outside of an armies core strengths (or align with for that matter) For example, you could use a factor of 0.9 for aligns with armies theme and 1.1 for clashing with the theme and 1.0 for neutral/jack of all trades. To use your example of ranged units in a freeguild army vs khorne. Gunkulator gives 100 points for 10 models Khorne: 110 points Freeguild: 90 points Stormcast: 100 points Or something along those lines anyway Link to comment Share on other sites More sharing options...
Kramer Posted November 18, 2018 Share Posted November 18, 2018 Just now, Retro said: I've put a bit of thought into that myself. Personally I thought having a factor of some sort to apply to units that quite obviously exist outside of an armies core strengths (or align with for that matter) For example, you could use a factor of 0.9 for aligns with armies theme and 1.1 for clashing with the theme and 1.0 for neutral/jack of all trades. To use your example of ranged units in a freeguild army vs khorne. Gunkulator gives 100 points for 10 models Khorne: 110 points Freeguild: 90 points Stormcast: 100 points Or something along those lines anyway hmm interesting. It's good to know you thought about it. But (as someone who plays KO) let me try to make it more difficult it How about a gunhauler (mobile cannon/chariot) vs. the Ogor Ironblaster (also mobile cannon/chariot)? Again I love the idea of the project even though I do not necessarily believe your goal is attainable, you still should try to get as close as possible. So more power to you. Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 18, 2018 Author Share Posted November 18, 2018 1 hour ago, Kramer said: Read it all, cool idea well thought out. But one thing I keep doubting about stuff like this... how do you value a specific roll in an army that has a focus on something else. So practical example. There was a conversation until the destruction thread about the ogor ironblaster. How does a high potential gun value in a close combat focussed army? In my mind a cannon in a melee focussed faction should be relatively more expensive. Likewise in a glass army anything with a high save should be relatively more expensive. So a 16” range 4+ 4+ archer in a khorne army should be more expensive than in a free guild army, do you agree and how do you want to value these element? Because it should be in tactical stage 2 but I feel it’s a good thing GW doesn’t try to quantify these things as it will lead to all factions doing the same. GREAT questions and thoughts! Sometimes in the context of this project, I talk about Level One adjusting and Level Two adjusting. Level One is really just the vanilla nuts and bolts. Making the thing predictive, consistent and reasonable. Level Two is what you are talking about: customizing for flavor, adding variety and interest. The possibilities here are huge, and I think that building a Level Two system on top of a stable Level One system allows you to really explore the possibilities of variation in a way that is consistent, manageable, and allows for much easier balancing and troubleshooting down the road. Here's how I would probably go about tackling your specific challenge: There are 13 unique variables in the current gunkulator that ultimately do the work of converting gameplay variables into points values. One of them is Range Modifier (RM). This is a value that is used to modify the Expected Damage Output (vs Saves of 3/4/5/6) of each weighted Weapon line and Dimension. It basically functions as a "points tax" that causes the damage done by a 18" attack to be worth more points than the same damage done by a 3" attack, which in turn costs more than the same damage done by a 1" attack. In your scenario where you want a gun unit in a melee focused faction to cost more points than a gun unit in a range focused faction (which I agree with!), I would simply suggest we start by differentiating the RM value for those factions (effectively, make the range tax higher for the destruction unit and lighter for the KO unit). And that's just for range. You can literally have custom values for any mix of the variables, across any mix of differentiation you want (Faction, Keyword, etc.). It would push the Rock/Paper/Scissors aspects of the game further. Link to comment Share on other sites More sharing options...
Retro Posted November 18, 2018 Share Posted November 18, 2018 Worked out the irondrake issue. The warscroll says 5 or more but they are pointed as being units of 10 for 180. If you adjust the values it comes out with 150/180/-30/20% 👌 Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 18, 2018 Author Share Posted November 18, 2018 OMG thank you! It was driving me nuts! Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 18, 2018 Author Share Posted November 18, 2018 Re: Ironblaster and modifying the RM variable... Gunkulator currently values Ironblaster at 115 points. Default RM value is 20. Lower numbers increase the range tax, higher numbers decrease it. Here are some of the results of tweaking RM on Ironblaster points: RM=5 188pts RM=10 139pts RM=20 (default) 115 pts RM=30 107pts RM=40 103pts Link to comment Share on other sites More sharing options...
Retro Posted November 19, 2018 Share Posted November 19, 2018 So I might have missed something to change, but on my copy of the gunkulator all I changed for irondrakes was the 4s to 9 in the expected damage and the 5 to 10 in the wounds sections. The result gave me 150. Is there another parameter I missed? Also, it might be worth having different range factors depending on the unit type or even attack type (not for level 1 development I'd say). For example, melee attacks having a different scaling value than ranged attacks or having a different RM value for melee attacks in horde units vs small elite ones. I figure that 40 skeletons with 2" range matters more than 2 morghast, seeing as it would be very rare that the morghast don't get both models into combat and the only benefit of range would be hitting over enemy bubble wraps Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 19, 2018 Author Share Posted November 19, 2018 Great thoughts. Re: Irondrakes, I'm not sure. I don't know the unit well, and I remember struggling a bit understanding the war scroll, so I may have translated it poorly. Or it may be an outlier. It's also relevant that we have not pointed Cinderblast Bomb. We also haven't pointed banners or musicians yet for any of these types of units, but their banner ability seems more powerful than most. (As an aside, I wonder if GW points banners and musicians at all? I remember thinking that it was weird when the Blood Knight banner changed from model recursion to -1 bravery, but their points cost didn't change at all). Having different range modifiers for melee and for range is a potentially really good idea. Thanks for suggesting it here so we have a running record for documentation. From a methodology standpoint, I think I want to get the right-sized sample list done and dusted, and then re-evaluate where we have problems. For instance, if we see that we have fidelity with GHB18 values in the majority of cases, but range units tend to be an outlier, or elite units tend to be an outlier, then we can focus on tweaking those. Your idea of different RM for range and melee is great, and probably theoretically better than one value for both, but if that's not faithful to how GW is currently doing it, then I don't think I want to go there right now as it will make calibrating a nightmare. Make sense? Link to comment Share on other sites More sharing options...
Retro Posted November 19, 2018 Share Posted November 19, 2018 Definitely makes sense, I was mostly posting it here so that it was swimming in the brains of others as well, theres a good to fair chance I would forget those kinds of things in a week or 2 Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 19, 2018 Author Share Posted November 19, 2018 Yep perfect. Link to comment Share on other sites More sharing options...
VBS Posted November 19, 2018 Share Posted November 19, 2018 Greetings, I feel inclined to post some rambling on this, as I'm generally interested in these type of projects (happen to work as project manager in something similar, just not with fantasy miniatures haha!). Please note, can't help much on statistics as I pretend to know more about methodology 🤣 So, random thoughts: The Proposed Stages seem quite on spot, it is a nice breakdown. And although stages 1 & 2 seem feasable, stage 3 is what seems most "impossible" to solve due to multiple random variables but also given "non-measurable" variables. What I mean by "non-measurable" variable are factors that can alter a units performs depending on external reasons/subjective use of units by people. Further explanation: one increases point cost for a unit based on its synergies alliance abilties, etc... as one might reason that the unit can/will benefit from it. But someone that uses the unit without making use of these factors that have an impact on the units point cost --> from that individual's point of view/unit usage, it will not reflect an accurate cost. The non-uniformity of unit usage between players cannot be grasped by statistics, and I'm assuming that a single "way of usage" (they get these benefits overall with alliance, X hero synergy, etc) will be used for the sake of clarity/simplicity. A way around this would be to make subgroup calculations for different scenarios, but that is just making things more complicated! To add another slab, as mentionned above, armies unique aspects such as being more melee oriented could have an impact on certain units that are out of that aspects (shooters in melee oriented factions). However, the grand alliance and ally system breaks to some extend this factor. Should this variable be added? If so, it goes back to the idea of subjective use by players (in this case, list building). The examples above is one of many others, such as the business/marketing factors. Obvious example: big units that get a bonus with X models, yet get a point reduction with X++ models. The massive regiment idea is a contradiction motivated by the desire to sell more models, and will be an obstacle to come up with a logical "statistical" method. On the polar opposite, as an illustration, you have what T9A did by mixing fixed minumum cost + increased individual additions, which would be a more realistic way of approaching a units performance that gets bonus rules based on numbers. I don't know if you thought how to approach such issues, but could require quite some changes to the regular method. I noticed when AoS added points that many of the points seemed to follow a somewhat "unit type" point distribution logic, more or less. Examples: Infantry: weaker 60-80p. per 10 / regular about 100p. per 10 / elite 120-160p. per 10 Cavalry: Light/weaker about 120p. per 5 / Heavy-specialized 140-160p. per 5 Heroes: based around 100p. aproximately, with +/- but overall, quite homogenous (weapon stats, etc...). Monsters: wider range but could still see a pattern fleeting around 200-300p., similar damage table, etc. This makes me think that for the sake of making the analysis clearer and sharper, making variables individually by regrouping unit type could help. This would result in subgroups of variables instead of just a huge bulk regrouping everything. Given the somewhat consistency in Warscrolls and bonuses per additional rules (synergies, etc) per unit types, this could eventually be a different approach. Well, these are my initial random thoughts. Could be a bit too much leaning to Stages 2 & 3, though. Feel free to think it relevant or not! Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 20, 2018 Author Share Posted November 20, 2018 16 hours ago, VBS said: Greetings, I feel inclined to post some rambling on this, as I'm generally interested in these type of projects (happen to work as project manager in something similar, just not with fantasy miniatures haha!). Please note, can't help much on statistics as I pretend to know more about methodology 🤣 Greetings to you too! Your experience and brainpower are welcome! 16 hours ago, VBS said: And although stages 1 & 2 seem feasable, stage 3 is what seems most "impossible" to solve due to multiple random variables but also given "non-measurable" variables. Part of the reason it's stage 3 is that it scares the s--t out of me, so I want to kick it down the road as far as possible. It's possible that it will appear less complex once 1 and 2 are more fully locked down. It's also very possible that it is neither necessary nor desirable to take it that far. 16 hours ago, VBS said: What I mean by "non-measurable" variable are factors that can alter a units performs depending on external reasons/subjective use of units by people. Further explanation: one increases point cost for a unit based on its synergies alliance abilties, etc... as one might reason that the unit can/will benefit from it. But someone that uses the unit without making use of these factors that have an impact on the units point cost --> from that individual's point of view/unit usage, it will not reflect an accurate cost. The non-uniformity of unit usage between players cannot be grasped by statistics, and I'm assuming that a single "way of usage" (they get these benefits overall with alliance, X hero synergy, etc) will be used for the sake of clarity/simplicity. A way around this would be to make subgroup calculations for different scenarios, but that is just making things more complicated! Agree. And usage is also complicated by redundancy. For example, it is useful to have a fast, mobile unit for grabbing objectives. But that use value decreases somewhat the more of those types of units I have. There are only so many objectives after all! It kind of reinforces to me where we should set the limits of what we are trying to do. List building is a skill in this game. Setting aside the practical impossibility of building a points system that accounts for these "non-measurable" variables, we also need to face the fact that even if it were possible, it probably wouldn't be desirable, since all we would be doing is effectively reducing the impact of skilled play on the game. 16 hours ago, VBS said: The examples above is one of many others, such as the business/marketing factors. Obvious example: big units that get a bonus with X models, yet get a point reduction with X++ models. The massive regiment idea is a contradiction motivated by the desire to sell more models, and will be an obstacle to come up with a logical "statistical" method. On the polar opposite, as an illustration, you have what T9A did by mixing fixed minumum cost + increased individual additions, which would be a more realistic way of approaching a units performance that gets bonus rules based on numbers. I don't know if you thought how to approach such issues, but could require quite some changes to the regular method. My initial reaction to something like massive regiment is that its not a problem. It's a known variable and it's consistent across factions. It would be more of an issue I think if it was hidden and arbitrary. 16 hours ago, VBS said: I noticed when AoS added points that many of the points seemed to follow a somewhat "unit type" point distribution logic, more or less. Examples: Infantry: weaker 60-80p. per 10 / regular about 100p. per 10 / elite 120-160p. per 10 Cavalry: Light/weaker about 120p. per 5 / Heavy-specialized 140-160p. per 5 Heroes: based around 100p. aproximately, with +/- but overall, quite homogenous (weapon stats, etc...). Monsters: wider range but could still see a pattern fleeting around 200-300p., similar damage table, etc. This makes me think that for the sake of making the analysis clearer and sharper, making variables individually by regrouping unit type could help. This would result in subgroups of variables instead of just a huge bulk regrouping everything. Given the somewhat consistency in Warscrolls and bonuses per additional rules (synergies, etc) per unit types, this could eventually be a different approach. Yes, exactly, and I am very curious about unpacking that. I want to know if there is a rational method to it, or if it's all somewhat of a "fudge factor." Do they have a consistent basis for how the cost, say, an infantry unit, or do they have a constrained but still arbitrary process (e.g. infantry will have stats in this range, abilities of this type, and a points cost in this range, and we just cobble it all together a la carte rather than try to cost it out in a consistent manner). 16 hours ago, VBS said: Well, these are my initial random thoughts. Could be a bit too much leaning to Stages 2 & 3, though. Feel free to think it relevant or not! Super relevant! Thanks for sharing. Link to comment Share on other sites More sharing options...
Retro Posted November 20, 2018 Share Posted November 20, 2018 22 hours ago, VBS said: To add another slab, as mentionned above, armies unique aspects such as being more melee oriented could have an impact on certain units that are out of that aspects (shooters in melee oriented factions). However, the grand alliance and ally system breaks to some extend this factor. Should this variable be added? If so, it goes back to the idea of subjective use by players (in this case, list building). Hey, thanks for your input, it's good to have as much insight as possible! In regards to what I quoted above, I feel that to a certain extent the loss of allegiance abilities/artifacts/spells etc would balance that out somewhat (hopefully). This is working on the assumption that the army in question has allegiance abilities at all and that they are superior to the generic grand alliance ones available. Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 20, 2018 Author Share Posted November 20, 2018 Hey Retro, do you have a copy of GHB17? I only have '18, and I am wondering if there might be diagnostic value in having any previous unit costs available to look at. I've added in some more units to the tool, and some of the output is bang on, but some of it is a bit puzzling. I'll try to get an updated sheet up later today. Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 21, 2018 Author Share Posted November 21, 2018 Updated to version 1.02 Change Summary Spoiler Version 1.02 Change Summary: All Sheets/Rename Unit Morghast to Morghast Archai Effective Wounds/Rename field Ward Save to Damage Save Effective Wounds/Create new field MW Save, express as a % Effective Wounds/Create new field WoundsMW, formula= (Models*Wounds)/(1-MW Save) Formula Values/Create new assumption PtsMW (starting value = 12?) Effective Wounds/Create new field PVMW, formula = (WoundsMW*PtsMW) Effective Wounds/Modify formula of field PV Wounds to include PVMW Added Units: Sequitors, Ironblaster, Huskard on Thundertusk, Arkanaut Company, Tzaangor Skyfires, Judicators Analysis Spoiler Green = calibrated to within 10% of GHB18 Blue = Outlier Orange = Incomplete (contains Stage 2 variables that are not currently costed) Currently calibrated on 58% of test units, excluding Stage 2 exceptions. Some of the outliers are probably to be expected based on Community sentiment (Morrsarr Guard, Grimghast Reapers). But some of the outliers seem strange. I can't see an apparent pattern. I thought maybe Range was an issue given the skewing of KO units, but both Judicators and Celestar Ballista are spot on, Tzaangor Skyfires are within threshold (would love to compare to GHB17 cost to validate if original value was more in line), and Irondrakes are a significant outlier in the opposite direction. We need more units! Would love to get suggestions for more good candidate units to add to the sample size. Would also love to hear thoughts on any possible patterns or suggestions regarding the current output. Link to comment Share on other sites More sharing options...
Superninja Posted November 21, 2018 Share Posted November 21, 2018 I would like to see ironjawz units put through the gunkalator. We feel behind the curve. Celestar ballista is spot on?!? It has always seemed overpowered/underpriced to me. oh well, I'll chalk that up to me needing to play more/learn more/play better. Part of that feeling was the ability to drop it anywhere(allegiance abilities)/the other part was the d6 hits on each hit and the -2 rend. I wonder how it compares to gitmob rock lobber? The rock lobber has great range, no need for line of sight, -2 rend as well, and 3 damage, but still seems lesser to the ballista. (at least in my mind) Link to comment Share on other sites More sharing options...
Lemon Knuckles Posted November 21, 2018 Author Share Posted November 21, 2018 I'll run a couple of different ironjawz units and the rock lobber through it. Thanks for the suggestions. It's useful when people have particular units in mind that stand out to them because of comparisons they've made to other units. And I should clarify, by "spot on" I mean that the tool is currently solving for GHB18 values. It shouldn't be interpreted as a statement about whether those values are right or wrong. But we can check whether the same system that solves for ballista can also solve for rock lobber. If it can't, that might mean that your intuitions are correct. Link to comment Share on other sites More sharing options...
Superninja Posted November 21, 2018 Share Posted November 21, 2018 yeah i understand what you meant and I am still surprised by the outcome. no worries though. I enjoy the work you are doing. Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.