Switch Theme:

2000 tyranids versus orks competitive (pics)  [RSS] Share on facebook Share on Twitter Submit to Reddit
»
Author Message
Advert


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.




Made in us
Longtime Dakkanaut




Shep wrote:[ ..stuff..


Interesting incites

This message was edited 2 times. Last update was at 2010/02/19 04:11:12


 
   
Made in us
Devious Space Marine dedicated to Tzeentch




@nurglitch:

In your latest analysis...


An expected value of 0.71 attacking a cruising Land Raider.


An expected value of 0.83 attacking a cruising Land Raider.


An expected value of 1.29 attacking a cruising Land Raider.

An expected value of 0.99 attacking a cruising Land Raider.

An expected value of 1.16 attacking a cruising Land Raider.


What are theses numbers? The expected value of what? You're multiplying it by the number of attacks so I assume it's some sort of value assigned based on chance to hit? But I can't figure out where it's coming from or what exactly is being calculated.

Could you fill me in on what this step is? I'm with you otherwise.

Edit: Just realized this looked more confrontational than I meant. I was just curious. I love numbers and math, just wanting to follow your logic.

@Danger-boy:

Shep: so are you saying that foot nids (hive guard/tervs/etc.) can't cut it against mech builds in general, or just that they have a major uphill battle against a guard/wolf gunline that goes first. How many armies can say that they are excited to see that across from them though? From the way you are talking, it sounds like nid players need to make a decision about what they want to be vulnerable to (gunline/hammernators/etc.) and hope to not see those at the table. Does that about sum it up?


That's kind of the idea, but here's the big issue with the way I see it. You'll never build an army that's super strong against everything. But it's all about meta-game calls. My Tzeentch Daemons are much stronger against IG/SW gunline than any Nid list I've been able to come up with. On the other hand, I'd just about roll over and die to Nob Bikerz and have an uphill fight against Lash chaos. The Nid lists I've been managing to come up with lately basically flip-flop those.

The problem is, the way IG and SW have shifted the competitive metagame around, I would be much more comfortable going to a tournament with an army that's weak against Nob Bikerz and Lash than one that's weak against mech IG and mech SW. The meching up of the meta has weakened Lash somewhat, and IG (and the Lash armies still around) have made Nob Bikerz almost a non-issue. The weaknesses that I see in the Nids are weaknesses against the armies I think are likely to be out in the most force at a tournament like Adepticon. It's not like I'm playing an army that rolls over and dies to double Seer Council and I hope not to play it...compare the odds of running into double Seer Council with the odds of running into IG with Chimera spam+Vendettas and various tanks, and I just don't like having an army that's hoping to dodge one of the most likely opponents (or at least my perception of what's likely to be around).

That's what I'm trying to get past with Nids, and failing. Not shoring up all weaknesses...but trying to shore up the weaknesses against the most likely armies. That's what frustrating to me. Shep may be in a slightly different boat, but I've gotten the feeling we're seeing things from the same perspective.

This message was edited 1 time. Last update was at 2010/02/19 14:23:21


 
   
Made in fr
Guardsman with Flashlight



Paris, France

Nurglitch wrote:ED209:

Those red numbers are the expected value, or reliability, of the different options weighted by their potential.

A Carnifex with no close combat weaponry has:
A4+1 charging, for 5 attacks
An expected value of 0.71 attacking a cruising Land Raider.
Therefore: (5)(0.71) = 3.55 overall effectiveness vs LR

If i understand correctly, you multiply the expected number of penetrating hits (already including the number of attacks) by the number of attacks?
Doesn't that double account for the number of attacks?
   
Made in us
Longtime Dakkanaut





Orlando, Florida

I only post this here because it's a natural progression of what Shep has been testing, and basically my own "take" on his testing lists...

After some thought I modified the list I am testing on Saturday:

-HQ-

Hive Tyrant w/ Lashwhip-Bonesword, Scything Talons, Ancient Enemy, Life Leach, Paroxysm, Regeneration = 215

2 Tyrant Guard = 120

-Elites-

3 Zoanthropes = 180

3 Zoanthropes = 180

3 Hive Guard = 150

-Troops-

10 Termagaunts = 50

10 Termagaunts = 50

Tervigon w/ Catalyst, Onslaught, Toxin, Regeneration, Cluster Spines = 230

Tervigon w/ Catalyst, Onslaught, Toxin, Regeneration, Cluster Spines = 230

-Heavy Support-

Tyrannofex w/ Rupture Cannon, Regeneration, Thorax Swarm w/ Shreddershard Beatles, Cluster Spines = 295

Tyrannofex w/ Rupture Cannon, Regeneration, Thorax Swarm w/ Shreddershard Beatles, Cluster Spines = 295

Total = 1995

Basically I dropped the Venomthropes and a unit of Hive Guard to put 2 units of three Zoanthropes on the table. After a lot of reading describing the problems with spamming Hive Guard, I though it would be interesting to see what happens if you have a large contingent of Zoanthropes on the table.

For some reason, even though they can be insta-killed, I still have greater faith in the Zoanthropes survivability. I don't need a Gaunt screen to rely on cover saves with them, and their effective range is 24". If I cast Onslaught on them from the Tervigons that effective range could almost push 30". Plus the model is big enough that a unit of three should be able to give me cover to the Tyrannofexes.

Against aggressive armies I have a better chance of popping heavy transports with 10 str. 10 shots. And I have Ancient Enemy Poisoned Gaunts and a Hive Tyrant to do my own assault heavy lifting.

I still have regeneration on everything. At 1850, I would simply drop the regeneration.

Thoughts?

This message was edited 1 time. Last update was at 2010/02/19 14:35:50


Current Armies: Blood Angels, Imperial Guard (40k), Skorne, Retribution (Warmachine), Vampire Counts (Fantasy)

 
   
Made in de
Regular Dakkanaut






Nurglitch wrote:ED209:

Those red numbers are the expected value, or reliability, of the different options weighted by their potential.

A Carnifex with no close combat weaponry has:
A4+1 charging, for 5 attacks
An expected value of 0.71 attacking a cruising Land Raider.
Therefore: (5)(0.71) = 3.55 overall effectiveness vs LR



This is wrong math, the 0.71hit is the final result of average dice roll ,no mulitplie after.
   
Made in us
Devious Space Marine dedicated to Tzeentch




Love to see how that list goes Mahu. And I agree on the Zoanthropes...I love them. My biggest issue with them is actually the anti-psyker stuff more than anything else. I'm a big fan of using them as moving cover. And the option to throw the blast around if there are no good hard armor targets.

I'm still questionable on the Tfexes...let me know if you feel like they earn their 300 point price tag.
   
Made in us
Longtime Dakkanaut





Orlando, Florida

And I agree on the Zoanthropes...I love them. My biggest issue with them is actually the anti-psyker stuff more than anything else. I'm a big fan of using them as moving cover. And the option to throw the blast around if there are no good hard armor targets.


It will be interesting, there is a lot of trade-offs there and my primary concern is with range and the idea that they can be insta-killed and wounded easier.

With 6 of them, I am less concerned about anti-psyker stuff. Another thing that put them in my radar besides their consistent armor saves, is the fact that I have greater Shadow of the Warp and Synapse coverage, which would help against Space Wolves.

I'm still questionable on the Tfexes...let me know if you feel like they earn their 300 point price tag.


In my previous playtesting, they haven't done well, but I was only running one, it will be interesting to see what two do.

Here is some math I am looking at concerning the trade off:

3 Hive Guard versus AV 12 = 6 Shots, 4 Hits, 1.32 Penetrate, .44 Wrecked or Destroyed

3 Zoeanthropes versus AV 12 = 3 Tests, 2.4 Shots, 1.6 Hit, 1.32 Penetrate, 0.65 Wrecked or Destroyed

3 Vendetta's shooting at a unit of 3 Hive Guard in cover = 9 Lascannon shots, 6.75 Hit, 5.6 Wounds, 2.8 Wounds caused after 4+ cover, 1 and a half dead Hive Guard

3 Vendetta's shooting at a unit of 3 Zoanthropes = 9 Lascannon shots, 6.75 Hit, 5.6 Wounds, 1.8 Wounds caused after 3+ save, 1 or 2 dead Zoanthropes

Hydra Squadron of 2 shooting at 3 Hive Guard in cover = 8 Autocannon Shots, 6 Hits, 3.09 Wounds, 1.54 wounds after cover

Hydra Squadron of 2 shooting at 3 Zoanthropes = 8 Autocannon Shots, 6 Hits, 4.99 Wounds, 1.64 wounds after saves

Current Armies: Blood Angels, Imperial Guard (40k), Skorne, Retribution (Warmachine), Vampire Counts (Fantasy)

 
   
Made in us
Dakar



Arlington, VA

sirisaacnuton wrote:@nurglitch:

In your latest analysis...


An expected value of 0.71 attacking a cruising Land Raider.
An expected value of 0.83 attacking a cruising Land Raider.
An expected value of 1.29 attacking a cruising Land Raider.
An expected value of 0.99 attacking a cruising Land Raider.
An expected value of 1.16 attacking a cruising Land Raider.


What are theses numbers? The expected value of what? You're multiplying it by the number of attacks so I assume it's some sort of value assigned based on chance to hit? But I can't figure out where it's coming from or what exactly is being calculated.


http://en.wikipedia.org/wiki/Expected_value

In probability theory and statistics, the expected value (or expectation value, or mathematical expectation, or mean, or first moment) of a random variable is the integral of the random variable with respect to its probability measure.

For discrete random variables this is equivalent to the probability-weighted sum of the possible values.

The term "expected value" can be misleading. It must not be confused with the "most probable value." The expected value is in general not a typical value that the random variable can take on. It is often helpful to interpret the expected value of a random variable as the long-run average value of the variable over many independent repetitions of an experiment.

The expected value may be intuitively understood by the law of large numbers: The expected value, when it exists, is almost surely the limit of the sample mean as sample size grows to infinity. The value may not be expected in the general sense — the "expected value" itself may be unlikely or even impossible (such as having 2.5 children), just like the sample mean. The expected value does not exist for all distributions, such as the Cauchy distribution.

It is possible to construct an expected value equal to the probability of an event by taking the expectation of an indicator function that is one if the event has occurred and zero otherwise. This relationship can be used to translate properties of expected values into properties of probabilities, e.g. using the law of large numbers to justify estimating probabilities by frequencies.


Automatically Appended Next Post:
ED209 wrote:
Nurglitch wrote:ED209:

Those red numbers are the expected value, or reliability, of the different options weighted by their potential.

A Carnifex with no close combat weaponry has:
A4+1 charging, for 5 attacks
An expected value of 0.71 attacking a cruising Land Raider.
Therefore: (5)(0.71) = 3.55 overall effectiveness vs LR



This is wrong math, the 0.71hit is the final result of average dice roll ,no mulitplie after.


That's what he's saying, the E(x) Expected value (read arithmetic mean of all probable results as sample size goes to infinity) is .71 he's multiplying that by the sample size (5) to get an "effectiveness" for the sample (5) of successes in attacking AV14.




Automatically Appended Next Post:
Mahu wrote:


Here is some math I am looking at concerning the trade off:

3 Hive Guard versus AV 12 = 6 Shots, 4 Hits, 1.32 Penetrate, .44 Wrecked or Destroyed

3 Zoeanthropes versus AV 12 = 3 Tests, 2.4 Shots, 1.6 Hit, 1.32 Penetrate, 0.65 Wrecked or Destroyed

3 Vendetta's shooting at a unit of 3 Hive Guard in cover = 9 Lascannon shots, 6.75 Hit, 5.6 Wounds, 2.8 Wounds caused after 4+ cover, 1 and a half dead Hive Guard

3 Vendetta's shooting at a unit of 3 Zoanthropes = 9 Lascannon shots, 6.75 Hit, 5.6 Wounds, 1.8 Wounds caused after 3+ save, 1 or 2 dead Zoanthropes

Hydra Squadron of 2 shooting at 3 Hive Guard in cover = 8 Autocannon Shots, 6 Hits, 3.09 Wounds, 1.54 wounds after cover

Hydra Squadron of 2 shooting at 3 Zoanthropes = 8 Autocannon Shots, 6 Hits, 4.99 Wounds, 1.64 wounds after saves


I'm not seeing enough of a difference to justify spending points on Zoanthropes here. Now if you add in the blast, and ability to work on AV 14, It does change things.

I'm floored by Shep's results (That you simply cannot line up enough T6 wounds across from shooty-guard to hit their lines in numbers). With the Lack of ability to (reliably) manipulate the reserves game in the early turns (T2); Reserving your army (especially if you want to get into CC) appears to make targeting decisions even easier for your opponent.


This message was edited 3 times. Last update was at 2010/02/19 17:05:55


 
   
Made in us
Devious Space Marine dedicated to Tzeentch




I think you misunderstood me... I'm familiar with what an expectation value is, I was asking what that was an expectation value of. Number of hits? Number of pens? Number of wrecked/destroyed?

Since he then multiplied it by the number of attacks, my initial assumption was that it was the expectation value of something occurring per attack, but the numbers were far too big to be an expectation value of anything for one attack. So it seemed to be the expectation value for (say) 5 attacks with 2xST, or 6-8 attacks with 1xST, or something. But that's not something you would then multiply by the number of attacks again, as you're then double-counting the number of attacks. It skews the results toward more attacks to do that, thus inflating the perceived benefit of more attacks (i.e. Crushing Claws).

I just wanted to make sure I wasn't missing something, so I was asking what outcome's expectation value was being calculated in those lines.
   
Made in us
Longtime Dakkanaut





Orlando, Florida

I'm not seeing enough of a difference to justify spending points on Zoanthropes here. Now if you add in the blast, and ability to work on AV 14, It does change things.


Well, that is what I am figuring, it's a lateral shift against most AV 12 heavy Guard, but it is a positive shift if we are talking about Leman Russ chasis, Battlewagons, and Land Raiders.

Plus, it answers the question in the list, how do we deal with Jaws in a mostly foot based list.

Current Armies: Blood Angels, Imperial Guard (40k), Skorne, Retribution (Warmachine), Vampire Counts (Fantasy)

 
   
Made in us
Devious Space Marine dedicated to Tzeentch




Well, not exactly an ideal counter to Jaws. Jaws is 36", right? Put a guy in a Rhino and it seems like he wouldn't have too much trouble out-ranging the SitW while still getting decent shots on Tfexes (or something), and even out on foot in a squad (if the Rhino gets popped) it doesn't seem like the list has enough power to reach out and touch him quickly through a squad of ablative wounds.


Automatically Appended Next Post:
Now if it's 24" it's not nearly as bad. Just can't remember off the top. Thought it was 36.

This message was edited 1 time. Last update was at 2010/02/19 17:55:42


 
   
Made in us
Fixture of Dakka






San Jose, CA

Jaws is 24".

Quis Custodiet Ipsos Custodes? 
   
Made in ca
Decrepit Dakkanaut





Rugrud wrote:
Nurglitch wrote:ED209:

Those red numbers are the expected value, or reliability, of the different options weighted by their potential.

A Carnifex with no close combat weaponry has:
A4+1 charging, for 5 attacks
An expected value of 0.71 attacking a cruising Land Raider.
Therefore: (5)(0.71) = 3.55 overall effectiveness vs LR

If i understand correctly, you multiply the expected number of penetrating hits (already including the number of attacks) by the number of attacks?
Doesn't that double account for the number of attacks?

Yes, it does, and on purpose. That's why I'm saying that the expected value is weighted by the potential number of attacks. Think of the expected value as the expected value given the number of attacks and average luck, and then the potential as the number of attacks and good luck. So the attacks are counted twice because there's 1/3 of the time when the average can be expected, and there's the 1/3 of the time when you should get lucky. The final 1/3 is getting unlucky and as mentioned that is a wash.

This message was edited 1 time. Last update was at 2010/02/19 19:08:58


 
   
Made in us
Scarred Ultramarine Tyrannic War Veteran





Arlington, VA

I should have paid more attention in stats, but that seems almost like fuzzy math.

Check out my blog for bat reps and pics of my Ultramarine Honorguard (Counts as GK) Army!
Howlingmoon wrote:Good on you for finally realizing the scum that is tournament players, Warhammer would really be better off if those mongrels all left to play Warmachine with the rest of the anti-social miscreants.
combatmedic wrote:Im sure the only reason Japan lost WW2 was because the US failed disclose beforehand they had Tactical Nuke special rule.

 
   
Made in us
Devious Space Marine dedicated to Tzeentch




Nurglitch wrote:
Yes, it does, and on purpose. That's why I'm saying that the expected value is weighted by the potential number of attacks.


If you weight your expected value by the number of attacks, you're not generating the data you claim it's generating...namely, you're not generating an accurate picture of the benefits of the various configurations. We're discussing the utility of the extra attacks of the Crushing Claws vs. the other possible ccw configurations. If you weight the numbers by how many attacks are made, and then simply compare them based on which number is highest, you're coming up with an unrealistic view of the benefits of the claws. Obviously, the claws shine if the numbers are being weighted by the number of attacks generated, as that skews the results in favor of having more attacks (in other words, in favor of the claws).

For a fair comparison between the builds, one of two things should happen. Either generate the expectation value for a single attack, given the factors like no rerolls, rerolls on 1's, or all rerolls, and then multiply by the number of attacks. Or else generate the expectation value for the batch of attacks (exactly like you did), but then don't multiply by the number of attacks. By doing both, you're essentially looking at the expectation value for the number of attacks squared, which obviously increases in value per attack much more quickly than when you don't double-count the attacks.

Otherwise, one could just as easily make the argument the other way, against the crushing claws, by weighting the expectation value with the chance to hit with an attack. Weighting the hit chance instead of the number of attacks will cause the numbers to favor rerolls (so Scything Talons) over # of attacks (claws).

   
Made in ca
Decrepit Dakkanaut





sirisaacnuton:

We are discussing the utility of the extra attacks of the Crushing Claws in comparison to the other possible configurations. If we limit our consideration to expected utility, then we're disregarding potentiality, which is rather important in a game where the number of attacks puts a hard and fast limit on the amount of damage a model can do.

Where we only consider the expected value without considering the maximum potential, then we get a skewed notion of the configuration's potential because we can expect 1/3 of the time to have good luck, 1/3 back luck, and 1/3 to roll average, or the expected value.

So basically by weighting the expected value by the potential, we are weighting the expected value by the probability of hitting with an attack.

It's basically:

Expected Value
EV = number of attacks x probability of hits

Potential Value
PV = number of attacks x probability of hits

Total Value
TV = EV x PV

A Carnifex with two pairs of Scything Talons vs Tactical Space Marines:

[((5 x 50%)+(5 x 50% x 50%))][(5)(100%)] = 18.75

A Carnifex with Crushing Claws vs Tactical Space Marines

[((6 x 50%)+(7 x 50%)+(8 x 50%))/3][(8)(100%)] = 28

And yes, as you say, of course the Crushing Claws shine when we consider the weighted expected value: that's why they're an upgrade! They cost 25pts: they should yield an improvement of at least 16% overall.
   
Made in us
Devious Space Marine dedicated to Tzeentch




What I'm not on the same page as is the multiplication part of that. The expected value is just that...given the number of attacks and the odds of each one hitting, it gives the number of expected hits.

And the potential value, as you defined it, basically seems to be the maximum number of possible hits, in other words if every attack were to hit.

So in your first example there, you have an expectation value of 3.75 hits. I'm with you. You have a maximum potential value of 5 hits. I'm with you.

But then once you multiply those, you have a "total value" of 18.75 hits squared (that is, hits^2). This, to me, is not a meaningful value.

In the two comparisons you have there, the 2xtalons build has a higher expected value than the crushing claws. Which means in an infinite sample size, the talon build outperforms the crushing claws build. Sure, the claws can certainly kill more Marines than is possible for the talons to do, thanks to the higher attack number, but despite that chance the most probable outcome of the combat favors the talons. I just simply don't see what new information is gleaned when you multiply by the max value. I can't see what your "total value" part really reflects.

Maybe it can be illustrated with a simpler example. The expectation value on a d6 is 3.5. The maximum possible value is a 6. If I multiply the two, I get a value of 21. If I do the same procedure with a d20, I get an EV of 10.5, a max of 20, and the product of the two is 210.

But what do those numbers, 21 and 210, tell me about the statistics or probabilities of those dice? I understand what information is conveyed by the mean, max or min values, variances and standard deviations, etc etc. But not by multiplying the mean and the max value.
   
Made in us
Longtime Dakkanaut






Los Angeles, CA

kaiservonhugal wrote:Shep - do you think this list would be improved:

tervigon with catalyst, onslaught, crushing claws, adrenal glands and toxin sacs
3x hive guard
3x hive guard
3x hive guard
15x termagants with devourers
16x termagants with devourers
tervigon with catalyst, onslaught, crushing claws, adrenal glands and toxin sacs
tervigon with catalyst, onslaught, crushing claws, adrenal glands and toxin sacs
tyrannofex with rupture cannon and cluster spines
tyrannofex with rupture cannon and cluster spines

Down grade the devourers to deathspitters - saving 124 points and add Gargoyles. Combined - the gargoyles and gaunts can screen the MC's for a turn. Reducing gunline HW hits by half for a turn....


I almost ran 2x10 termagants and some gargoyles, but i have no experience with gargoyles, and haven't really thought of a "mission" for them in my games. Thats on me, not the gargoyles... but if gargoyles and termagants in front of my tervigons and t-fexes give me an 'argument free' 4+ cover save, well then hot damn, that would be for sure a list upgrade. it would also send me back to the drawing board really quick because that is pretty big.

Argument free is most vital to me as i play a lot against people i don't know that really want to win. If my list becomes functional only when I have that 4+ cover, and i end up having to argue until I'm blue in the face about 51% obscured... then that just won't do.

And I think you meant fleshborers, not deathspitters right?

Thanks for that tech! i will be glueing a box of gargoyles together tonight, taking pics and setting up a poll this weekend.

Please check out my current project blog

Feel free to PM me to talk about your list ideas....

The Sprue Posse Gaming Club 
   
Made in ca
Decrepit Dakkanaut





sirisaacnuton:

I think the problem is that you're taking the most probable outcome, the set of outcomes described by the expected value calculation, and generalizing it to mean "all outcomes". The problem is that the expected value calculation only covers the sort of situations we'll be most likely to encounter, which is good and bad. It's good because it's more reliable than not covering the situations we're most likely to encounter, but it's bad because it's less relaible than covering all of the situations we're going to encounter.

Though I think we may be talking past each other with reference to the expected value. In the terms I'm familiar with, in game theory and decision theory the way I was taught, the expected value is simply the expected utility of any given option intersected with the likelihood of that option paying out. In that case, the idea that a D6 has an expected value is somewhat incomprehensible to me. I mean I'm familiar with the expected value on 1/6, 2/6, etc, or different outcomes upon rolling it but I don't see what the utility of a whole D6 is. Do you mean the highest number that you can roll?

In that case the expected utility of the highest number on 1D6 would be 6, the likelihood of it obtaining would be ~17%, and its maximum potential is equal to the expected utility of the highest number. Meaning that:

Weighted Expected Value of 6 on 1D6
(6 x 17%) x (6 x 100%) = 6

Weighted Expected Value of 20 on 1D20
(20 x 5%) x (20 x 100%) = 20

These numbers tell you that if you were rolling for money and had the option of either rolling 1D6, or rolling 1D20, then you should roll the 1D20. If you got the money on a 6 on 1D6 or a 20 on 1D20, and that was the amount of money you'd get on a successful roll, you'd act more rationally (taking the value of rationality as that of consistency in preferring more dollars to less) to roll the 1D20 despite having a higher likelihood of getting money with the 1D6.

Similarly the number previously told you that if you wanted to maximize the combat potential of a Carnifex (ignoring issues of Initiative and points cost), then you should upgrade one set of Scything Talons to Crushing Claws for the reasons given above.
   
Made in us
Nasty Nob






Gardner, MA

Yes Shep - I mean Fleshborers. Definitely add the Gargoyles and put them up front when screening. Gaunts cant screen them so it doesnt do any good putting guants in front. The combined height of gaunts covering the ground and Gargoyles covering the mid section allows for cover saves. Youll need more than a box though thats the downside - I put together 4 boxes.

Theyre good - real good just dont take troop choices away to pay for their points - they are not a substitute for scoring units.

A man's character is his fate.
 
   
Made in us
Bush? No, Eldar Ranger





Los Angeles

I don't think that analysis is very helpful. I don't have a good sense of what 18.75 vs 28.00 means.

[((5 x 50%)+(5 x 50% x 50%))][(5)(100%)] = 18.75

A Carnifex with Crushing Claws vs Tactical Space Marines

[((6 x 50%)+(7 x 50%)+(8 x 50%))/3][(8)(100%)] = 28

I think this is a better way to consider the worth of crushing claws

the format of this is

# of hits, Prob of event occurring, Prob of at least this many hits happening

For two ST with 5 attacks

0 Hits 0.0976563 100
1 Hit 1.4648438 99.90234375
2 Hits 08.7890625 98.4375
3 Hits 26.3671875 89.6484375
4 Hits 39.5507813 63.28125
5 Hits 23.7304688 23.73046875

For Crushing Claws with 8 attacks



0 Hits 0.390625 100
1 Hit 3.125 99.609375
2 Hits 10.9375 96.484375
3 Hits 21.875 85.546875
4 Hits 27.34375 63.671875
5 Hits 21.875 36.328125
6 Hits 10.9375 14.453125
7 Hits 3.125 3.515625
8 Hits 0.390625 .390625

For Crushing claws with 7 attacks with at least this number of hits probability looks like:

0 100
1 99.21875
2 93.75
3 77.34375
4 50
5 22.65625
6 6.25
7 0.78

This message was edited 5 times. Last update was at 2010/02/19 23:38:39


The Sprue Posse

Armies  
   
Made in us
Longtime Dakkanaut






Los Angeles, CA

kaiservonhugal wrote:Yes Shep - I mean Fleshborers. Definitely add the Gargoyles and put them up front when screening. Gaunts cant screen them so it doesnt do any good putting guants in front. The combined height of gaunts covering the ground and Gargoyles covering the mid section allows for cover saves. Youll need more than a box though thats the downside - I put together 4 boxes.

Theyre good - real good just dont take troop choices away to pay for their points - they are not a substitute for scoring units.


Sweet, I'd just buy a box to get some together in order to see if they cover my scratch-built tervigons. Knowing my spending habits, if they did work I'd probably be in for 8 boxes haha.

And are you saying that just the gargs could possibly give me 51%? Or I'd need the gants for low and the gargs for high? i suppose I'll find that out tonight, If my FLGS has gargs that is

Please check out my current project blog

Feel free to PM me to talk about your list ideas....

The Sprue Posse Gaming Club 
   
Made in us
Nasty Nob






Gardner, MA

The latter - both units work to shield the MC. At least thats how mine work. There is room for taking advantage of modeling but I didnt. I simply put the Gargoyles in front - even one row to expend frontage works. Then the guants then the MC's. I keep the Gargoyles dirt cheap. Theyre the first ones in and get shot to ribbons.

In your last list - I didnt remove any models from it to make room for the Gargoyles - I just downgraded the gaunts weapon to make room for 20 more Gargoyles. You replace 60 S4 AP- shots with 50 S4 AP5 shots and GAIN 20 more wounds, 20 6 to hit auto wounding attacks (40 on the charge) and a way to give cover to your MC's - PRICELESS.

Trygons are still too high but Carnies and T-Fexi are easy to cover. As a side note - you can move the Gargoyles out of the way for shooting the Rupture cannon and then Fleet them back in to cover the T-Fex back up again.

This message was edited 2 times. Last update was at 2010/02/19 22:45:57


A man's character is his fate.
 
   
Made in us
Pyromaniac Hellhound Pilot






Shep said it in a nice way let me say it Kenpachi style . Nurg it is very immature of you/others to believe that your numbers is better then actual game play experience. In the real world, Generals do not care how a weapon perfoms in a laboratory. Crushing claws makes carnifexes better in cc, thank you for using numbers to tell us that more attacks equals better results. There is a simple error in your "intense studies", you are telling Shep and Mahu how good these cc units are when they are using shooty list. They are focusing on durable shooting early on then relying on weight of numbers or monstrous creatures to overwhelm their opponent in cc later. Why wouuld you make suggestions that do not fit their playing style? Do you honestly believe that you can convince someone with numbers that their own personal experience is flawed?

Dont get me wrong your input is valuable, but you can honestly tell who is playtesting and who is not. You have the purpose of the heavy venom cannon down, but if you playtested like Shep and I you would know that we can not take enough of them to matter, we are in the MSU era where stoping one vehicle from shooting is not good enough. Also having carnifexes with 8 attacks on the charge is nice but an opponent that has multiple units with multiple melta or plasma weapons is more then capable of taking down a carnifex in 1 turn(you can figure out the probability of that happening if you like ). When you have short range weapon that can take out the cc carnifex, you focus all your attention on the other things that are immediate threats, making target priority easier for your opponent. Carnifexes have a role and can be good but in this thread where people are obviously discussing a certain playing stlye your comments are not useful at all.

   
Made in us
Pyromaniac Hellhound Pilot






As an observer of this thread, you are using numbers to defend your opinion when others are using actual game play experience.

   
Made in us
Longtime Dakkanaut





Orlando, Florida

Nurglitch hate not with standing, I just got back from three rounds of play testing with my new list.

Here it is again:

Hive Tyrant w/ Lashwhip-Bonesword, Scything Talons, Ancient Enemy, Life Leach, Paroxysm, Regeneration = 215

2 Tyrant Guard = 120

-Elites-

3 Zoanthropes = 180

3 Zoanthropes = 180

3 Hive Guard = 150

-Troops-

10 Termagaunts = 50

10 Termagaunts = 50

Tervigon w/ Catalyst, Onslaught, Toxin, Regeneration, Cluster Spines = 230

Tervigon w/ Catalyst, Onslaught, Toxin, Regeneration, Cluster Spines = 230

-Heavy Support-

Tyrannofex w/ Rupture Cannon, Regeneration, Thorax Swarm w/ Shreddershard Beatles, Cluster Spines = 295

Tyrannofex w/ Rupture Cannon, Regeneration, Thorax Swarm w/ Shreddershard Beatles, Cluster Spines = 295

Total = 1995

Played three games using Adepticon missions versus a Terminator Heavy Space Marine List, a highly competitive IG gunline list, and a pretty tough Terminator-Logan led Space Wolf list (with Jaws). And though I don't have Battle Reports per say, I will say I learned some pretty good lessions.

1. Regeneration is overpriced as people have predicted. I least I know that now.

2. Zoanthropes > Hive Guard. Because the shooting from them got substantially better versus heavy transports, I was more able to fight on my terms more often then relying on my screen to save me. Also, because their save is more reliable, I can afford to be more aggressive with them. Range wasn't too much of an issue because of onslaught. It actually benefited me that they are not immune to instant death because I was able to pull low ap fire away form my monstrous creatures. Almost every game by turn 4 the Zoanthropes may have been dead, but I barely had any wounds on my TMCs.


I didn't win any of my games, mainly due to bad dice rolling on my part, bad target priority on my part, and some incredible clutch rolling on my opponents part (for example, 2 Powerfist Terminators rolling 7 hits against my Hive Tyrant under the effects of Paroxysm, which killed the squad, when statistically the Hive Tyrant should have lived.)

However, all my games where hella close, and usually came down to some critical rolls. Having the Zoanthropes on the table versus Hive Guard were a huge force multiplier for me. And I think is necessary to build a proper "take all comers" meat shield list.

This message was edited 1 time. Last update was at 2010/02/21 05:43:03


Current Armies: Blood Angels, Imperial Guard (40k), Skorne, Retribution (Warmachine), Vampire Counts (Fantasy)

 
   
Made in us
Longtime Dakkanaut






While I think CKO is borderline trolling, Nurglich I have to agree with him that your math is off--though not due to lack of experience. The idea of multipling the EV and PV to form a meaningful number, as you have implimented it, is very wrong.

As for your example of rolling a d6 or a d20, with a 6 or 20 being a payout, I must again disagree that the 20 is the better option. They are identical, as the risk versus reward are identical for both.

IE, roll the d6 and the d20 120 times each. You can 'expect' the 6 to come up 20 times on the 120 rolls, and the 20 will come up 6 times on the 120 rolls. The total payout for both is expected to be 120!

Now, if you are looking at distributions, then the 6 will obviously have more central tendancies, as the data sets will only vary from 1 to 6. This can be both good and bad, depending on what you want. Likewise, the 1 to 20 data sets will have more of a 'burst' appearance, as the data is more spread out, thus increasing the chances for one element of the data set to be inflated by a trend.

So, in the above example, if you must have 'at least' 70 back on the above bets, then by choosing to roll the d6 is a better option as the smaller data set will more likely result in at least 70. However, if you must have 'more than' 130, then you are more likely to get more than 130 with the d20, as you are more likely to burst with the larger data variation. As long as both are still expected to be the same, however, the spread between the two is largely unimportant, though sometimes this data can become very useful.

As this data pertains to a carnifex with 2 st versus 1 st and 1 cc, the expected vaules all points to the fact that 2 scytal are far (FAR) superior when you need to hit faster moving vehicles, while the data all points to crushing claws being ALWAYS better when you hit automaticly. Thus, the question becomes, in relation to claws, do you need to spend points on the fex to hit immobile vehicles better, or is it better to spend less points, while hitting mobile vehicles better? Most people I know realize that nid's have more problems with mobile vehiles than immobile vehicles, thus meaning that crushing claws are not only a waste of points, but REDUCE your ability to damage mobile vehicles.

PS: to CKO, using numbers, as long as they are correct numbers, is by far the better estimation of usefulness. Purely going by experience is a terrible indicator in 40k--the variables are simply too great; you would need to play literally millions of games (representing multiple games with all different list variations and mission variations) to gain enough experience to claim that your data is not flawed by personal perceptions--and then you would need to be able to remember the millions of games in intricate detail in order to sort through the data to pass it on. We can not do this, thus any experience gained will not be able to be passed on as reiable to another, as it amounts to a personal opinion without math to back it up.

Edit-- to Mahu, if you found yourself versus razorback spacewolves, how do you think the zoan's would have fared compared to the hive guard? I see more razorback/rhino lists than termie raider lists, so my playtesting is pretty much exactly counter to yours.

This message was edited 1 time. Last update was at 2010/02/21 05:50:17


 
   
Made in ca
Decrepit Dakkanaut





DevianID:

No, the idea of multiplying the expected value by the potential value to determine a weighted expected value (weighted for potential...) is quite valid. Here's why: suppose we do your exercise of rolling 1D6 120 times and 1D20 120 times. Can we honestly expect 20 rolls of 1D6 to be 6? We can if we remember that 120 rolls means there is a margin of error to that expectation. If that margin of error is not in our favour, what I was calling 'unlucky', it's a wash. But if we get lucky, when the actual results of those 120 rolls means more 6s or more 20s, more 20s are preferable.

Do you see what I'm getting at here? If the dice will reliably adhere to the average, then there is no difference. But when there's deviation from the average, it'll be a wash if they both roll more non-target number, but it'll very much be in the D20 favour if they roll more target numbers.

That's why weighting the expected value by the potential value is a good description of overall value, because it not only accounts for what we can reasonably expect, but what we can unreasonably expect - those far ends of the bell curve that show up and crap on us at unexpected moments.

Accounting for the average is reasonable, but just accounting for the average is unreasonable since, particularly for the case of Warhammer 40k, we're not rolling enough dice for even small deviations to be smoothed away over the long run. When a single lucky roll can make the difference between winning with $6 and winning with $20, the consistent decision, the rational decision is that you go by the expected utility or potential.

Let's take the numbers I ran: I originally ran them against a cruising Land Raider on purpose, because then they would need a 6 to hit, the case that would make reliability the more important than the potential number of hits. Crushing Claws turned out to be only slightly less reliable at hitting on a 6, their worst-case scenario. As you say, they're better at their best-case scenario for the reasons you're illustrated.

This reliability is not factoring in the effectiveness of hitting more times: That's why the reliability of Assault Cannons are comparable to Lascannons: it's not because of Rending, not directly, it's because the volume of fire is 4x as much, multiplying the chances of affecting the vehicle it's directed at.

The reason I think people are so enthusiastic about reliability in 40k is because GW dice tend to show up 1s more, so they're used to rolling abnormally low and that skews their perception of potential value.
   
Made in us
Raging Ravener



Virginia

Shep:

Your list (the one kaiser quoted) looks almost identical to the one I'd been kicking around, but I'm just not sure it puts out enough damage. You might be able to weather the shooting -- MIGHT: Stelek's SW list is just plain ridiculous. But it doesn't seem like you're going to actually outshoot him, and you have no dedicated CC units. Are you thinking gunline CC is weak enough that you can kill them with non-dedicated units, and just outshoot CC-oriented armies? I haven't done any playtesting, mind you, so I defer to you in all matters.

Also, Nurglitch, I have no problem with math, nor with playtesting, but you seem to switch between "Mathhammer rules! Experience sucks!" and "Mathhammer sucks! Experience rules!" as it suits you. Here's what I think would make people take your argument seriously: a) Give us some math that shows your list holding its own against the test lists Shep posted. b) Write a battle report in which you beat any of those lists with yours. Your claims are far enough from the norm that, unless you do one of those things, people are gonna stay skeptical.

But CKO, that (your second-to-last post) was kind of harsh, and not really accurate. I think Nurglitch is suggesting Carnifexes as alternatives to Tyrannofexes within this style of list. Besides, the idea is to find any kind of Tyranid list that can beat a gunline (and operate within the time constraints of a tournament). If Pyrovores, Rippers, Biovores and Raveners worked, I'm sure that everyone would love to hear about it.

Oof, I gotta work on shortening my posts. Not used to the forum format.
   
Made in us
Longtime Dakkanaut






While what you are saying about luck is close, I think my wall o' text hid my bit on distributions.

For example, what you say about a single roll winning you either 6$ or $20 is again flawed, as the context is off.

IE, if you HAVE to get $20, and you are only allowed 1 bet, then betting on the 1-20 is, of course, the ONLY option, as it is the only thing with the potential earnings you need.

BUT, what if you simply need $1? What if ANY positive result is acceptable? Do you want a 1 in 6 chance to have a positive result, or a 1 in 20 chance?

Basicly, what you are describing is a 'win more' situation. If the target is killing a tank, do you want a 1 in 6 chance to kill it, or a 1 in 20 chance to kill it 5 times? While a 1 in 20 chance to kill it 4 times, mathwise might provide more kills, only a single positive result matters in 40k. Longer odds for a staticticly equal payoff mean less reliable results. Less reliable results = less value.

Edit: By the way, a charging fex with 2scytal expects to get 1.52 hits on a vehicle needing 6's to be hit. A charging fex with 1scytal and 1claw, rolling 2 extra attacks for claws, expects to get 1.36 hits on a vehicle needing 6's to be hit. Thus claws on a fex REDUCE the number of hits expected, while COSTING more points. Claws are a bad option for hunting mobile vehicles!

This message was edited 1 time. Last update was at 2010/02/21 06:40:24


 
   
 
Forum Index » 40K Battle Reports
Go to: