Monster math still off: Five L1s demolish two trolls

In a previous packet, we pitted five level 1 characters (dwarven fighter, elven archer, human wizard, human cleric, halfling rogue) against one troll. They killed it easily. Repeatedly.

This packet, we tried this again. The elven archer was replaced with a monk. Otherwise same lineup. We tried one troll, which died quickly (3 rounds), and then we tried two trolls, which died almost as quickly: 4 rounds.

In testing, the main issues seemed to be:
- Low monster HP. The first troll was down by round 2, finished off by a Burning Hands from the wizard; the second troll fell at the beginning of round 4.
- Low monster AC. The PCs are almost guaranteed to hit an AC11 monster.
- Low-ish monster to-hit. +6 seems respectable, but is it for an L5 monster? PCs at that level would have between +8  and +9 to hit (+4/5 from stat, likely +1 from item, +3 from level). If the troll lasted longer, that might not have been an issue. It had a hard time touching the monk or dwarf (both 18 AC), and when it died and its companion went after the "squishies", it missed the rogue (AC 14) twice, hit the wizard (AC 11) once, and then it died.

No PC died. Not even the wizard, who took one hit before the remaining troll went down. Because it was all over in 4 rounds, the trolls didn't really get a chance to swipe at people a lot. The second troll did save against Command, improbably, but fell prey to the monk's Stunning Blow.

Somewhere deep in the monster math, there's a disconnect. A party of L1s stumbling into a troll lair and disturbing two trolls should have exactly two likely outcomes: Run like hell (double-move) and get away from those trolls, hoping they don't catch up - or TPK. That, instead, the party delivers a resounding beat-down, is bizarre.

The troll's re-gen was lowered to 5 because in play testing at Wizards, 10 for a pack of trolls was too much, so we are told. While that may hold for a pack of trolls - we haven't tested that - I can say that the 5 regen on two trolls barely registered. The monk does 2d6+10 damage each turn (Flurry of Blows), add to that the dwarven fighter with 1d10+4, the rogue with 1d6+4 (sling), the wizard with his ray of frost 1d10 and the cleric with his lance of faith 2d6, and individual trolls die very, very quickly.

"If they hit!", I hear you cry. Well, PCs have between +5 and +6 to hit, so with an AC 11, you do the math. They'll hit about 70% of the time, so that's 70% of that damage going through, on average.

 
Overall, I'll agree.
In the latest packet, to-hit bonus from level was lowered for players - a lvl1 Fighter now gets a +1 from his class, instead of the massive +3 they got in last packet. The monk appears to be a prime target for a nerfbat swing anyway, but otherwise monsters are indeed squishy. AC11 for a troll is below pathetic.
I find the troll having a hard time hitting the fighter as being logical. After all, a heavily-armored fighter is supposed to be a hard target.

Bear in mind that while a troll is listed as level 5, its xp value rates it as something less of a tough target for a group of four 3rd-level PCs (check the table in the DM Guidelines document), so it is possible for a group of five 1st-level PCs to defeat a single troll. Two trolls, however, should have ripped right through the group.

While checking out the DM Guidelines document, I noticed that in the version found in the latest packet, it mentions that an encounter might be more difficult than indicated if it includes a large number of monsters. In contrast, the version in the earlier playtest packet mentioned a "single monster limit." Apparently, a single monster (or an encounter with fewer monsters than the number of PCs) must be of a level way higher than the PCs' to pose a serious threat any more... Fascinating.
D&D almost always featured epic battles of PCs against extremely powerful solo monsters (hence the inclusion of the word "Dragons" in the game's name). A possible switch in focus to monster groups as the apex of encounter difficulty is going to be a lot more fundamental that it seems...
For giggles, we re-tried with tweaked trolls. "Now longer-lasting! The most durable troll yet!" And yep, when the trolls don't fold faster than a Chinese laundry, their ability to hit often enough tells, and PCs start dieing around round 5 or so. How that is accomplished can be subject to tweaks. Higher AC or higher HP, or a combination of the two.

So what Wizards did with this playpacket is:
- Boosted the trolls to-hit a bit, which was needed
- Nerfed their staying-power by dropping AC from 14 to 11 while keeping HP the same as before

Taking the XP and encounter building guidelines into account, a level 5 party would find 3 trolls an average encounter, and 5 trolls a tough one.

So is what we're seeing here a side-effect of bounded accuracy? Is it that a level-appropriate tough encounter of 5 trolls needs them to go down this fast, but because of bounded accuracy, that means that a single troll, or two, no longer poses a challenge even to a level 1 group?

Is this by design? An unavoidable side effect of design?
 
lok, I think you're on to something. Because of bounded accuracy, single monsters indeed don't pose much of a challenge any more. That said, the math's still off. I bet you that 3 trolls are not a "normal" encounter for these PCs at level 5, but instead a "cake walk". In a normal encounter, I expect that the PCs will be somewhat pressed. In a tough encounter (5 trolls), I expect them to be hard-pressed and use up all of their resources, and need to be working as a team to avoid PC death.

This is indeed a fundamental change. I am so used to the idea of "two trolls will just mop the floor with a party of foolhardy low levels" that I am having a hard time wrapping my head around the idea that, maybe, trolls aren't that tough unless they are encountered en masse.

I still think AC 11 is pathetically low, and gut feeling, I'd give them 50% more HP on top of an AC14 to make them feel like a troll.
 
I think it's bounded accuracy in conjunction with increased PC damage (especially at higher levels), that makes single monsters not to pose a threat. Not only is a monster's attack not guaranteed to hit even a 1st-level PC, but the damage output of a party now can generally down a monster of even higher level within a single round.

Besides the effort required of the DM now (you'd have to manage more monsters in every fight, if you want fights to be interesting), we have to consider how monster "ecology" changes. In the old days, a single troll roaming the woods struck fear in villagers and adventurers alike, because it was indeed a formidable opponent to low-level PCs. That also meant that said troll could afford to live alone in a cave in the woods - few things posed a threat to its existence. But if lone trolls are no longer a threat, does it mean that from now on trolls will adopt a tribal lifestyle? And what about monsters of lower intelligence, that are unlikely to live in groups, like manticores? Are they suddenly considered endangered species, since it is now far easier to hunt them down and kill them?
WoTc is zeroing in on numbers, but as everyone has pointed out, they are not there yet.   I agree with the lot of you.   AC 11 for a troll seems too low, and if PCs are going to dish out so much damage, monsters will need to have more hit points.   Personally, I think PC damage output should be cut back some how.   I don't like how they get combat expertise and monsters don't (but if monsters had combat expertise it would be too unwieldy and mega dangerous). 

I'm thinking that combats that pit equal number of PCs against an equal number of equally leveled foes the combat could run about 3-4 rounds long.  (Average Encounter)

An easier encounter could last 1-2 rounds.

A difficult encounter could run about 5-12 rounds...depending.  

If they can work out these numbers, D&DNext combats will give everyone what they want...quick combats for story driven parts of the adventure, average encounters to deplete PC resources, and longer more difficult combats to let players play more tactically.   I like the mixed bag approach.           

A Brave Knight of WTF

 

Rhenny's Blog:  http://community.wizards.com/user/1497701/blog

 

 

Trolls are a really unfair test.

Trolls have an incredibly powerful ability that is easily overcome if you are prepared. That makes it tough to balance them. Trolls in 2nd edition were the same way where they awarded several times as much experience as monsters of the same level and power, provided you could overcome the regeneration. And that is fine, it is a reward to the party for being prepared. But trolls should never be used as the litmus test for how strong monsters are at that level.
Trolls have an incredibly powerful ability that is easily overcome if you are prepared. That makes it tough to balance them.



Yeah, I was curious about this, since I didn't see it mentioned in the OP: how much fire and/or acid damage did the party have?  Do you see similar results with monsters of comparable XP value but without a powerful mechanic like that?
I get that bounded accuracy is meant to keep some of the numbers down, but it's pretty obvious they're still fine tuning it.  They seem to be doing pretty well on the player side, but it could use some adjusting on the monster side.  Dragons, trolls, everything could use a small boost to AC and HP, just to help them equal out to their Challenge Rating (which it still is, whatever they actually call it).  If a Troll is supposed to be a lvl 5 encounter, it should give a challege to a party of lvl 1 PCs.  Lvl 1 PCs have a total attack of +4, generally, with a total damage ranging from 1d10 (5.5) for wizards to 1d8 + 3 + 1d6 (11) for martial classes.  Realistically, they should be able to hit it about 50% of the time, and deal serious damage to it only on crits.  For a lvl 1 encounter, they should be able to hit the same 50%, maybe a little better, but they don't need to crit to make it mad, or even kill it out right.


You know, I've gotten annoyed with writing all that to make my point, so I'm just going to simplify it.  Equal level encounters should more about tactics, on both sides, than raw power.  Encounters like a troll to lvl 1 PCs, the troll should come close to mowing over them via sheer power.  So give the troll fast healing 10/fire and acid, an AC of 14 or 15, and raise their HP a little, and they'll feel as powerful as trolls should.
Up until this point I had my reservations about increasing monster hit points, because it meant more HD, but I just realized a monster's HD have thus far had zero game use (A good night's sleep works miracles...). No spell, ability, or effect depends on a monster's HD, and no combat value or character aspect (feats, skills, atk bonus) gets calculated off them. So apparently, increasing monster hit points is simple and easy.

OTOH, with monster hit points at their current values, combat feels a lot more realistic. One or two solid blows can down a minotaur now, which makes sense. A well-placed hit should have an effect on any kind of target. Glancing blows would be another matter, of course.

I think what monsters need is defense options. Preferably as static as possible (i.e. few actions or dice rolls), but otherwise a monster supposed to be tough should be tougher than the average character.
Perhaps if a monster's tough hide gave it more than AC, and less than full resistance... How about having monsters reduce damage by a die roll? All weapon damage would be affected for sure, and depending on the monster, perhaps some other damage types as well. The idea is not mine, comes from Alternity (why did Wizards discontinue one of the best sci-fi systems ever?), where armors reduced damage by a die roll.

For example, for out troll test subject, let AC and hit points as they are. The troll's tough hide, however, reduces all weapon damage by 1d6. (Assume the troll's hide is only slightly tougher than a cow's hide, the troll mostly relies on its regeneration for survival.) A dragon, otoh, could have a damage reduction of 2d8 or 3d6, and its magic resistance trait could be modified to allow it to apply that DR even against spell damage.
Also, critical hits would ignore DR - a critical is by definition a hit placed in a most vulnerable spot, and I'd say "vulnerable" means the monster does not get to apply DR.
Introduce a general maneuver that lets PCs attempt a weak spot attack (thus ignoring DR even on a normal hit). I'm thinking something along the lines of "Make an attack at -2. Spend a single martial die to halve a monster's DR against your attack, or spend two dice to ignore it. Spend an extra die to negate the attack penalty."

A similar approach might be applicable to PCs wearing heavy armors (e.g. chain 1d4, plate 1d8), but PCs in heavy armor are already difficult enough for monsters to hit as is, so I'm not sure it would really serve a purpose, even though it is a realistic approach.
when running this test did you also take into account that players would not always go directly into battle fully prepared?

hp may already be dropped, spellcasters may have already used up some of their spell/daily resources and the prepared spells will not always be appropriate for fighting trolls
I agree with a lot of the OP's points, except for the one about the trolls' to-hit bonus.  For me, at +6, they're hitting AC 18 much too often.  I'd like to see monster attack bonus numbers dropped by one or two points across the board.

Additionally, while I agree that the trolls' AC is depressingly low, I also think +8/+9 to hit is a bit much for level 5 PCs.  Given that even +1 weapons are powerful under the flatter math they're proposing this time around, I certainly wouldn't allow PCs with single-digit levels to have them.  Also, I probably wouldn't let them start with ability scores so high that they're already rocking 18s in their abilities by 5th level.

If you have to resort to making offensive comments instead of making logical arguments, you deserve to be ignored.

Up until this point I had my reservations about increasing monster hit points, because it meant more HD, but I just realized a monster's HD have thus far had zero game use (A good night's sleep works miracles...). No spell, ability, or effect depends on a monster's HD, and no combat value or character aspect (feats, skills, atk bonus) gets calculated off them. So apparently, increasing monster hit points is simple and easy.

OTOH, with monster hit points at their current values, combat feels a lot more realistic. One or two solid blows can down a minotaur now, which makes sense. A well-placed hit should have an effect on any kind of target. Glancing blows would be another matter, of course.

I think what monsters need is defense options. Preferably as static as possible (i.e. few actions or dice rolls), but otherwise a monster supposed to be tough should be tougher than the average character.
Perhaps if a monster's tough hide gave it more than AC, and less than full resistance... How about having monsters reduce damage by a die roll? All weapon damage would be affected for sure, and depending on the monster, perhaps some other damage types as well. The idea is not mine, comes from Alternity (why did Wizards discontinue one of the best sci-fi systems ever?), where armors reduced damage by a die roll.

For example, for out troll test subject, let AC and hit points as they are. The troll's tough hide, however, reduces all weapon damage by 1d6. (Assume the troll's hide is only slightly tougher than a cow's hide, the troll mostly relies on its regeneration for survival.) A dragon, otoh, could have a damage reduction of 2d8 or 3d6, and its magic resistance trait could be modified to allow it to apply that DR even against spell damage.
Also, critical hits would ignore DR - a critical is by definition a hit placed in a most vulnerable spot, and I'd say "vulnerable" means the monster does not get to apply DR.
Introduce a general maneuver that lets PCs attempt a weak spot attack (thus ignoring DR even on a normal hit). I'm thinking something along the lines of "Make an attack at -2. Spend a single martial die to halve a monster's DR against your attack, or spend two dice to ignore it. Spend an extra die to negate the attack penalty."

A similar approach might be applicable to PCs wearing heavy armors (e.g. chain 1d4, plate 1d8), but PCs in heavy armor are already difficult enough for monsters to hit as is, so I'm not sure it would really serve a purpose, even though it is a realistic approach.



I like what you are saying.   Resistance to weapons, resistance to non-magical weapons, resistance to types of weapon damage and energy sources is the easy way to make a monster tougher, so I'm not sure WoTC will add damage resistance.   Regeneration is in the game, and that effectively boosts hit points for those monsters that have it.  For really powerful monsters immunity could also be part of the design. 

Also, like you noticed, there is no rule that says a DM can't add hit points to a monster.  I think WoTC will most likely keep the core with monsters having lower hp so individual groups can vary the toughness as they see fit.   They can even give guidelines ---  Mook = 1/2 hp....Average has standard....elite has +50% or more...etc. 

After DMing and playing a number of sessions, it is getting much easier for me to alter monsters to vary the threat level vs. the party.   Also, in the last session I DMd I had planned for a group of 4 pcs, but only 2 players could make the session.  Within 1 min, I was able to rescale the adventure and make it work.   If WotC can keep monster use/building and encounter building easy and fluid, this will be one of the big benefits to D&DNext.   Bounded accuracy helps do this.

A Brave Knight of WTF

 

Rhenny's Blog:  http://community.wizards.com/user/1497701/blog

 

 

I think that monster attack bonuses are fine. Monster AC, however, should be expanded to the 13-20 range, with fringe cases existing between 10-12 (commoners in cloth) and 21-23 (demon lords/angel captains/and creatures with magic items). Likewise, monster HP seems a little low across the board. I would like it increased between 125%-200% (depending on the type of creature being stated). 

Two factors I think were involved here. First the party was 25% stronger than an average party. the second factor that was missed here was the fact that this was a test fight. The players knew exactly what they were fighting and were not only prepared but could alpha stike said trolls. They knew there would be no fights after this one, so you simply use everything you have no need to hold anything in reserve. Clinical fights like these are rarely a factor in how a monster measures up. Everything after the first encounter in a day becomes increasingly harder as more and more resources are used up. If you take an average game where you are in the middle of adventuring (not happening upon a single random encounter) you can figure 3-5 encounters in a day before you will likely need to rest. That means mathmatically you would have a 66-80% chance that you encounter the trolls when you are not at full strength. You can't plot every monsters numbers as if its going to be the first and only encounter in a given time period. If the math were calculated that way, by the 2nd to 3rd encounter in a day, characters would be dying off like flies.
Two factors I think were involved here. First the party was 25% stronger than an average party. the second factor that was missed here was the fact that this was a test fight. The players knew exactly what they were fighting and were not only prepared but could alpha stike said trolls. They knew there would be no fights after this one, so you simply use everything you have no need to hold anything in reserve. Clinical fights like these are rarely a factor in how a monster measures up. Everything after the first encounter in a day becomes increasingly harder as more and more resources are used up. If you take an average game where you are in the middle of adventuring (not happening upon a single random encounter) you can figure 3-5 encounters in a day before you will likely need to rest. That means mathmatically you would have a 66-80% chance that you encounter the trolls when you are not at full strength. You can't plot every monsters numbers as if its going to be the first and only encounter in a given time period. If the math were calculated that way, by the 2nd to 3rd encounter in a day, characters would be dying off like flies.



This is a good point.   Also, the way the game is built, if the PCs are surprised or outnumbered the encounter becomes much more dangerous.

A Brave Knight of WTF

 

Rhenny's Blog:  http://community.wizards.com/user/1497701/blog

 

 

It was also VERY lucky that none of them were killed.  With multiattack vs. AC 18, the troll does the following average damage:
Bite - 4.23 damage
Claw 1 - 3.68 damage
Claw 2 - 3.68 damage
Total - 11.59

Given the monks AC of 18, I'm guessing he didn't have a very high Constitution (in fact, I'm curious as to how he has such a high AC at all), and so his HP are going to be around 8 or 9.  Dead in 1 round on average.  Same goes for the rogue and cleric, especially given their lower ACs.  The wizard should be toast, most likely in one hit.  The dwarf fighter could last a bit longer, especially if he is a Hill Dwarf with a high Con.

But I also wonder about the characters: did you use the standard stats or roll?  Because I'm really wondering about that 18 AC for the monk (and the fact that he gets +5 to damage).  The only way to get that is by having a 20 Dex, which means you must be rolling (and got very lucky).  In addition, his Wisdom must be 16.  Assuming he is human, that means he rolled a 17 and a 15, so he got VERY lucky.  When you have characters with stats like that, they can take down harder foes more easily.  Looking back at your first post, I see that the Dwarf has a Strength of 18 (+4 to damage), and the Rogue has a Dex of 18 (+4 to damage). 

So a combination of very good luck against the monster, good luck with their own rolls, and high stats.  While the monsters certainly might need more HP, this is not a fair test.
It was also VERY lucky that none of them were killed.  With multiattack vs. AC 18, the troll does the following average damage:
Bite - 4.23 damage
Claw 1 - 3.68 damage
Claw 2 - 3.68 damage
Total - 11.59

Given the monks AC of 18, I'm guessing he didn't have a very high Constitution (in fact, I'm curious as to how he has such a high AC at all), and so his HP are going to be around 8 or 9.  Dead in 1 round on average.  Same goes for the rogue and cleric, especially given their lower ACs.  The wizard should be toast, most likely in one hit.  The dwarf fighter could last a bit longer, especially if he is a Hill Dwarf with a high Con.

But I also wonder about the characters: did you use the standard stats or roll?  Because I'm really wondering about that 18 AC for the monk (and the fact that he gets +5 to damage).  The only way to get that is by having a 20 Dex, which means you must be rolling (and got very lucky).  In addition, his Wisdom must be 16.  Assuming he is human, that means he rolled a 17 and a 15, so he got VERY lucky.  When you have characters with stats like that, they can take down harder foes more easily.  Looking back at your first post, I see that the Dwarf has a Strength of 18 (+4 to damage), and the Rogue has a Dex of 18 (+4 to damage). 

So a combination of very good luck against the monster, good luck with their own rolls, and high stats.  While the monsters certainly might need more HP, this is not a fair test.



Another good point.  Overpowered PCs will make it much easier. 

A Brave Knight of WTF

 

Rhenny's Blog:  http://community.wizards.com/user/1497701/blog

 

 

the second factor that was missed here was the fact that this was a test fight. The players knew exactly what they were fighting and were not only prepared but could alpha stike said trolls. They knew there would be no fights after this one, so you simply use everything you have no need to hold anything in reserve. Clinical fights like these are rarely a factor in how a monster measures up. Everything after the first encounter in a day becomes increasingly harder as more and more resources are used up. If you take an average game where you are in the middle of adventuring (not happening upon a single random encounter) you can figure 3-5 encounters in a day before you will likely need to rest. That means mathmatically you would have a 66-80% chance that you encounter the trolls when you are not at full strength. You can't plot every monsters numbers as if its going to be the first and only encounter in a given time period. If the math were calculated that way, by the 2nd to 3rd encounter in a day, characters would be dying off like flies.



so... what i said

Yeah, I was curious about this, since I didn't see it mentioned in the OP: how much fire and/or acid damage did the party have?  Do you see similar results with monsters of comparable XP value but without a powerful mechanic like that?



Just the burning hands from the wizard, and some torches, which they didn't use. We haven't tested other monsters yet. I'll be happy to do so and see what happens.
 

hp may already be dropped, spellcasters may have already used up some of their spell/daily resources and the prepared spells will not always be appropriate for fighting trolls



That is a good point. No, that was not taken into account. Let's take them one by one:

- Daily spells. Command, cure light and burning hands were used. Without cure light, we'd have had a death, most likely. Burning hands was useful, but they would have brought them down without. Good point, though.
- Prepared spells. At Level 1, the selection is so tiny that the Cleric just prepares Command and Cure Light every day, and the Wizard prepares Thunderwave and Burning Hands. There may be a Shield in there somewhere, too. They don't really choose what to prepare each day, yet.
- HP dropped. That could happen after a few encounters. Mostly, the party uses healing kits to heal up after fights, and will long-rest when they are out of healing. That said, yes, the party could encounter trolls when they are low on HP and have no healing spells left, and that would make it a lot harder for them.

Is that what it comes down to, though? A fully rested party of L1s should make short work of two trolls (*), and only depleting them of resources first will make this a challenge?

If so, I will have to make a major, major shift in thinking. I'm just not used to that. Two trolls vs L1 party = TPK, unless the party really prepared with things like digging pits and making the trolls fall into them, or other uses of terrain (rock fall, what have you), or some other really clever thing that showed they prepared for this fight. Prepared more than waking up that morning, that is. That's what I'm used to, and if that's no longer what happens with monsters in 5e, then I am not saying it is the end of the world: It would however be so very different as to be nearly unrecognizable to me as "classic D&D". 

And maybe that's the point. First I thought the math was off; now I'm thinking maybe I just don't understand the math. My main focus in this thread now is to find out which it is.

(*) I hear you on "test on more than trolls". We shall.

 

Additionally, while I agree that the trolls' AC is depressingly low, I also think +8/+9 to hit is a bit much for level 5 PCs.  Given that even +1 weapons are powerful under the flatter math they're proposing this time around, I certainly wouldn't allow PCs with single-digit levels to have them.  Also, I probably wouldn't let them start with ability scores so high that they're already rocking 18s in their abilities by 5th level.



Agreed that +8/9 to hit is a bit high.

As for stats, some of these guys run around with 18 at l1, some with 20. It's a group of people that like to optimize, and role-play at the same time. I don't want to take that away from them. It's part of what makes the game fun for them, and I can throw a bit more at them and expect them to handle it. (*)

And I've thought about whether it's the optimizing that caused this outcome, but I don't really think so. Say they were less optimized. Maybe the monk would do 2d6+6 or 2d6+8 each round then, and maybe he'd have an AC of 17 or 16, and the fighter's 18. Maybe the fighter and rogue damage would be 1 point less each round. Honestly, the way the fight felt, I don't think it would have made a difference.

I really encourage others to test "non-balanced" encounters as well, pitting low level PCs against higher level monsters, and reporting the outcome. I think it'll be helpful for everyone to get a feel for how encounter building works now.
 
(*) But to think of two trolls as "a bit more" is so very odd to me. The reason we did this test is that the party rips through equivalent-level fights with nary a scratch. I wanted to see where the limits are.
 

But I also wonder about the characters: did you use the standard stats or roll?

So a combination of very good luck against the monster, good luck with their own rolls, and high stats.  While the monsters certainly might need more HP, this is not a fair test.



These are rolled stats. We're using a system where everyone rolls dice, and then players get to use a common dice pool to make up stats, so everyone has the same aggregate stat value.

You're right on HP, the monk has 9 HP. So he got lucky. He's a wood elf, so there's your plus to stats, which gives him a 20 Dex (after combining die rolls to start that stat with 18).

You are right that this makes the characters tougher than they would otherwise be. So, maybe you are right: Two trolls against an optimized party of l1s is not fair to the monsters. I'd LOVE to see others run such tests and report their results, with less optimized characters. When you run this against your group, what's the outcome?

The reason these are tests is that I'm trying to find the boundaries, I'm trying to find out what I have to do. We get to play rarely with a full group (schedules haven't worked out), so when I only have one or two players, we test, to get a feel.

Maybe what I need to do is up monster HP by 50% across the board, and up AC selectively. That's what I'm trying to figure out right now: What do I need to do to monsters and encounters; how much of that is due to PC optimization; how much of that is due to monsters just being a bit weak; and how much of that is due to me not having grasped the "run more monsters, not tougher ones" concept.
 

You know, I've gotten annoyed with writing all that to make my point, so I'm just going to simplify it.  Equal level encounters should more about tactics, on both sides, than raw power.  Encounters like a troll to lvl 1 PCs, the troll should come close to mowing over them via sheer power.  So give the troll fast healing 10/fire and acid, an AC of 14 or 15, and raise their HP a little, and they'll feel as powerful as trolls should.



I find myself nodding my head to all of that. Thanks for your input!

Your system for creating stats is broken. It leads to overpowered characters. It is fine if that is how you run your home-games, but it is not how the game assumes that it should be run. There should be no way to get a 20 in a stat until level 8. There should be no way to start with an 18 in a stat unless you are a human. If you are not a human, your best stat ought to be a 17. Your group didn't optimize, they just broke the rules. The manner in which they broke the rules gave them attack bonuses of characters quite a bit higher than an 1st level character is expected to have. It sounds like their AC also benefited from such an advantage. In a bounded accuracy game, artificially boosting stats will have a drastic effect on a PCs powerlevel.

You should try the scene again. This time, though, force everyone to use the standard stat arrays from the book. If a group of 5 level 1 characters using standard stat arrays can kill 2 trolls without any difficulty, then we might have a problem.


Your system for creating stats is broken. It leads to overpowered characters. It is fine if that is how you run your home-games, but it is not how the game assumes that it should be run. There should be no way to get a 20 in a stat until level 8.


The game allows for 20 stats right out of the box, first level. The only thing his method does is increase the odds of it happening.

Creating a character states that the 'normal' way to get you stats is to roll them, meaning that you have a range of 3-18 before your race and class adjustments which can be up to +2 for everyone except the human that can be +3. 18+2=20 in my book. Since this is the case, the game SHOULD be able to handle those stats at 1st. 


Your system for creating stats is broken. It leads to overpowered characters.

Your group didn't optimize, they just broke the rules.

You should try the scene again. This time, though, force everyone to use the standard stat arrays from the book. If a group of 5 level 1 characters using standard stat arrays can kill 2 trolls without any difficulty, then we might have a problem.




I can drop their stats to standard point-buy, and we'll try that again. However, I disagree that they broke the rules. The rules state that stats can be rolled, and that's exactly what happened. With an 18 starting stat, +1 Dex from wood elf and +1 Dex from Monk, you end up with 20. 

This may be one lesson from these tests: In 5e, rolled stats are overpowered and to be avoided because of bounded accuracy. If that's so, it's important for me, and hopefully others, to realize. I am used to rolled stats meaning that a party can do more, can handle more, but not that it whacks the entire game out of balance. If it leads to complete imbalance, I want to figure this out now.

We are approaching this very much as a play test. Using stat generation mechanics that we've used in previous editions, as long as they are allowed by RAW, is part of that play test. 

Since this is the case, the game SHOULD be able to handle those stats at 1st. 

Arguably. Maybe. Maybe the design intent is that rolled stats lead to "cakewalk". And if so, that's okay: There are groups that want that "I sneeze at it and it dies" experience.




Your system for creating stats is broken. It leads to overpowered characters. It is fine if that is how you run your home-games, but it is not how the game assumes that it should be run. There should be no way to get a 20 in a stat until level 8.


The game allows for 20 stats right out of the box, first level. The only thing his method does is increase the odds of it happening.

Creating a character states that the 'normal' way to get you stats is to roll them, meaning that you have a range of 3-18 before your race and class adjustments which can be up to +2 for everyone except the human that can be +3. 18+2=20 in my book. Since this is the case, the game SHOULD be able to handle those stats at 1st. 





No, it really doesn't. I mean, it allows for rolling of stats. But its system of rolling stats will only give you a 1.62% chance of rolling an 18 in any given stat. That means that your chance of having a single 18 in one of your 6 stats is only 9%. The numbers it gives you as a standard array are your statistical norm. Meanwhile, anyone who is outside of that statistical norm will be under or overpowered. That is what randomly generating characters does. It leads to under and overpowered characters playing alongside each other. There is no way to avoid that. Meanwhile, if the DM does not force the group to play their character, be it under or overpowered, then the group will end up playing a game in which they are outside of the expected powerlevel. Even worse, random chance could cause that anyway. But there is still no way to avoid that, other than making stats virtually meaningless, which would be an even worse problem! That is why I hate randomly generating stats, and will never use that system, nor play alongside anyone who does. Meanwhile, the system used by the original poster breaks the statistical probabilities that the game assumes resulting in a group full of people who regularly have stats that any given player should only have 9% of the time. 


If you want to see if the game works given a group completely composed of the statistical norm, you need to play with a group made up of characters that all use the standard array or point buy. After that, if you choose to play with random stat generation, you should accept that imbalance will happen. Moreover, if you allow generation by the manner the original poster described, you should accept that your group will be drastically more powerful than their actual level suggests. Of course, if you are drastically more powerful for those reasons, you might be able to kick the butt of creatures that are higher level than you. That, though, doesn't seem like a problem to me--you allowed for the characters to be more powerful than they ought to be.

 


Your system for creating stats is broken. It leads to overpowered characters.

Your group didn't optimize, they just broke the rules.

You should try the scene again. This time, though, force everyone to use the standard stat arrays from the book. If a group of 5 level 1 characters using standard stat arrays can kill 2 trolls without any difficulty, then we might have a problem.




I can drop their stats to standard point-buy, and we'll try that again. However, I disagree that they broke the rules. The rules state that stats can be rolled, and that's exactly what happened. With an 18 starting stat, +1 Dex from wood elf and +1 Dex from Monk, you end up with 20. 

This may be one lesson from these tests: In 5e, rolled stats are overpowered and to be avoided because of bounded accuracy. If that's so, it's important for me, and hopefully others, to realize. I am used to rolled stats meaning that a party can do more, can handle more, but not that it whacks the entire game out of balance. If it leads to complete imbalance, I want to figure this out now.

We are approaching this very much as a play test. Using stat generation mechanics that we've used in previous editions, as long as they are allowed by RAW, is part of that play test. 

Since this is the case, the game SHOULD be able to handle those stats at 1st. 

Arguably. Maybe. Maybe the design intent is that rolled stats lead to "cakewalk". And if so, that's okay: There are groups that want that "I sneeze at it and it dies" experience.







This is the thing: you didn't use the rolling system described in the playtest packet. If you did, then each character rolls 4d6 keeping the highest 3 for each stat. Then you assign the numbers to your six stats. That is not the system you described above. The system in the book gives you a 9% of having a single 18 in any given stat. And, statistically speaking, you will end up with numbers that average around the standard array granted in the book. Even if one character is overpowered, it will statistically be balanced by another character who is underpowered. As no pool can be created that is larger than 6 numbers (because characters don’t share rolls as you allowed), the probability of a character having a major weakness in some area is far higher. And yes, sometimes you can have streaks of luck between a group that leads to an overpowered group that turns the game into a cakewalk (unless you throw them up against much harder than usual challenges). That is what rolling stats does. It has done that in every edition. Bounded accuracy will make it a little more obvious, though.


In any case, this is why I have always hated the “roll for stats” system. I never want to play in a game that doesn’t use point buy/standard array. I never have, since 3e. In fact, I actively avoid RPGs that make you randomly generate stats these days. The only reason I am OK with D&DN’s rolling system is because it also provides alternative systems that match my needs. 


Your system for creating stats is broken. It leads to overpowered characters. It is fine if that is how you run your home-games, but it is not how the game assumes that it should be run. There should be no way to get a 20 in a stat until level 8.


The game allows for 20 stats right out of the box, first level. The only thing his method does is increase the odds of it happening.

Creating a character states that the 'normal' way to get you stats is to roll them, meaning that you have a range of 3-18 before your race and class adjustments which can be up to +2 for everyone except the human that can be +3. 18+2=20 in my book. Since this is the case, the game SHOULD be able to handle those stats at 1st. 





No, it really doesn't. I mean, it allows for rolling of stats. But its system of rolling stats will only give you a 1.62% chance of rolling an 18 in any given stat. That means that your chance of having a single 18 in one of your 6 stats is only 9%.


See, that right there is my point. 9% isn't 0%, yet you state "No, it really doesn't." or "There should be no way to get a 20 in a stat until level 8.". Around 1/10th of the time it DOES. Now add into that the fact that 17's (and 16's for humans) will get you to that 20 before 8th, and that 9% is looking a bit bigger now isn't it? It's not THAT out of the box and it should be something the game can handle.

And yet, statistically speaking, with a group of 4 players, there is only a 31% chance that one of them will have an 18 in one stat. Statistically speaking, the group will not have anyone with a roll of an 18. And, the game can handle one person rolling an 18. What the game can't handle is multiple people rolling 18s, because that is a statistical anomaly. The math of the game is designed to make such situations very unlikely. When such situations happen anyway... well, statistical anomalies will result in under/overpowered groups. Choosing to roll invites the possibility of playing with a statistical anomaly. If, like me, you don’t like that, you shouldn’t allow groups to roll for stats…


And yet, statistically speaking, with a group of 4 players, there is only a 31% chance that one of them will have an 18 in one stat. Statistically speaking, the group will not have anyone with a roll of an 18. And, the game can handle one person rolling an 18. What the game can't handle is multiple people rolling 18s, because that is a statistical anomaly. The math of the game is designed to make such situations very unlikely. When such situations happen anyway... well, statistical anomalies will result in under/overpowered groups. Choosing to roll invites the possibility of playing with a statistical anomaly. If, like me, you don’t like that, you shouldn’t allow groups to roll for stats…


AH... Did you read what the stats for his group was? Only the monk had a 20 and the other two had 18's AFTER adjustments. SO only one person rolled an 18 and the others rolled a 16 or 15(human) since they did +4 damage. So AGAIN, how is 1/3 of the time something the game shouldn't be able handle? You are acting like 31% is SO small that it should be ignored as a corner case. 31% isn't a statistical anomaly.
Well, we have a lively discussion going, and thanks for that! I think the discussion about how badly rolled stats may or may not break 5e is very worthwhile having. That's important information for DMs and players alike.

> You didn't use the rolling system described in the playtest packet. If you did, then each character rolls 4d6 keeping the highest 3 for each stat. 

You are right, I didn't use that system. Instead, the system I used with five players (*) is:

Every player rolls 5 dice
The lowest 7 from that dice pool are chucked
Players can then combine the remaining dice into stats as they see fit, as long as every stat gets exactly 3 die results

Overpowered for 5e? It turns out, yes, it is. That system is from 3.5/4e, where it was used without breaking the game. I didn't invent it, I grabbed it from another DM who described it in forums.

So far we have: Rolling stats in this manner, which not only creates rolled stats but creates very favorable rolled stats, is a seriously bad idea in 5e.

I dropped all PCs to standard 27-point buy, not allowing any stat to be higher than 15 after point-buy and before racial/class adjustment. And I re-tested, with one troll, then with two trolls.

One troll: Went down in 3 rounds. Cleric exhausted his dailies, wizard used one. No PC deaths. Troll got to hit the monk once, and then was commanded. Monk used his Ki.

Two trolls: Went down in 5 rounds. Cleric exhausted his dailies, wizard exhausted his dailies. Both fighter and monk went down and were brought back up to 1 HP by cure minor (not in the same round ). Monk used his Ki. Group had to shuffle around a bit, and it was touch-and-go. This could have been a TPK had the dice fallen differently.

So that makes a lone troll a tough encounter for a group of five l1s, and two trolls a very tough encounter. Is that what it looks like?

By the XP values and encounter building guidelines, this was way above their collective heads. And maybe that's where the "1 or 2 monsters need to be really WAY above the PCs to be very scary" guideline comes in. More monsters, not tougher ones.

Well, this has been really, really helpful to me, and I am curious to see the ongoing discussion. To recap what I've learned:

- Don't let your players roll stats
- Certainly don't let them use a rolling system that's designed to give them an edge
- One high-level monster is not that scary. Testing against one high-level monster does not give you a good yardstick for group performance. Test against mobs of monsters of their own level, instead.


How do you guys feel about the fact that higher-level mobs can still be taken out by l1s, if the higher-level mobs are isolated or, at most, in pairs? (A well-prepared group of l1s with a bunch of acid and fire would have had an easier time taking out two trolls, for sure). This could be a feature rather than a bug. It could fit into a campaign, where PCs have to take out a pair of trolls terrorizing the nearby village, and actually have a chance of doing so if they prepare well.

Thoughts?


(*) For four players, roll 6 each, chuck the lowest 6. For six players, roll 4 each, chuck the lowest 6.

Every player rolls 5 dice
The lowest 7 from that dice pool are chucked
Players can then combine the remaining dice into stats as they see fit, as long as every stat gets exactly 3 die results 

LOL. Well that IS quite good.


- Don't let your players roll stats
- Certainly don't let them use a rolling system that's designed to give them an edge
- One high-level monster is not that scary. Testing against one high-level monster does not give you a good yardstick for group performance. Test against mobs of monsters of their own level, instead.

Well, I have a big issue with this. If the game allows rolling, it SHOULD be able to use characters rolled up with said method. So at least rolling method given in the rules should work or somethings wrong.

As far as alternate rolling systems, those always come with risks.

As far as non-scary monsters, I think a better way to 'test' them is to have the players pick out spells and equipment THEN pull out the monster. Then you'd see how a normal party would handle the monster. A well prepared party should always be able to take on something tougher than they could wandering into it.   


And yet, statistically speaking, with a group of 4 players, there is only a 31% chance that one of them will have an 18 in one stat. Statistically speaking, the group will not have anyone with a roll of an 18. And, the game can handle one person rolling an 18. What the game can't handle is multiple people rolling 18s, because that is a statistical anomaly. The math of the game is designed to make such situations very unlikely. When such situations happen anyway... well, statistical anomalies will result in under/overpowered groups. Choosing to roll invites the possibility of playing with a statistical anomaly. If, like me, you don’t like that, you shouldn’t allow groups to roll for stats…


AH... Did you read what the stats for his group was? Only the monk had a 20 and the other two had 18's AFTER adjustments. SO only one person rolled an 18 and the others rolled a 16 or 15(human) since they did +4 damage. So AGAIN, how is 1/3 of the time something the game shouldn't be able handle? You are acting like 31% is SO small that it should be ignored as a corner case. 31% isn't a statistical anomaly.




It is not the individual stats that are a problem, but the way the stats are grouped. There are too many good stats. The results are still a statistical anomaly if one uses the RAW. Of course, he didn't. Read the system he used. He describes it. That system will statistically result in much better stats on average! It is no surprise that his first run-through resulted in such a cake-walk. The second time, if you notice, things were not nearly so easy. The second time he used the standard array. So, I was right...

In any case, my opinion is that monster HP is a little low right now in general. Things are closeto where they should be. But, monster's should have a little more staying power than they currently do. 125%-200% of the current HP totals should about do it. Meanwhile, monster AC is also a bit low right now. The range should be extended to an 8 point spread between 13-20 instead of 13-18. Peasants wearing cloth can still drop below that (hitting the 10-12 range). Meanwhile, demon lords/angel captains and monsters using top end magic items should be able to hit up to the 21-23 range.


Well, I have a big issue with this. If the game allows rolling, it SHOULD be able to use characters rolled up with said method. So at least rolling method given in the rules should work or somethings wrong.

As far as alternate rolling systems, those always come with risks.

As far as non-scary monsters, I think a better way to 'test' them is to have the players pick out spells and equipment THEN pull out the monster. Then you'd see how a normal party would handle the monster. A well prepared party should always be able to take on something tougher than they could wandering into it.   



The rolling method given in the rules may still work, but I think you'll find there will be corner-cases where it breaks the game either way. Either towards the PCs steamrolling everything (several players had good rolls) or the PCs being steamrolled (several players had bad rolls). And I think the  effect stats have is far larger in 5e than it was in previous editions.

You are right that alternate rolling systems come with risks. In a way, this was deliberate. I didn't know what would happen, but I did want to test corner cases. Trying to break something and then seeing why it broke is what product testing is all about in my mind. I didn't expect things to break this way, and it took the collective wisdom of the forum to point out why things broke the way they did.
 
As for non-scary monsters: That's how this was tested. The PCs are equipped the way they went into a dungeon full of kobolds (which are really not scary at all to them) - Wormwrithings in Blingdenstone, to be exact. And because those kobolds posed absolutely zero challenge - not even a small one - I wanted to see what would pose a challenge. The only preparation this group had was a good night's sleep. And then I threw some trolls at them. For testing, outside the campaign we're trying to run, just to see what would happen.
 
 If they wandered into the trolls after exhausting their spells, then yes, Gygax be with them.
 
 And I hear you on "they should be able to take out two trolls if they sleep well, have a hearty breakfast, and come prepared." Actually, prepared, they'd probably polish them off, because they'd range them (trolls wouldn't be on them in round 1), and maybe they'd even try to kite them a bit, and there'd be more fire and acid than just the wizard's standard-prepared burning hands and a hastily-tossed torch. And maybe that's the way it should be now.

 
Cyber-Dave: Yes, I think that's reasonable on monster HP. To really test that, it'll have to be at-level. I'm hampered with that because schedules haven't allowed us to do more than these rather isolated tests in a while. Once I get these guys together again, I am going to see how they fare against at-level with standard HP, and at-level with boosted HP, in an actual play setting, not an isolated fight.

The second time, if you notice, things were not nearly so easy. The second time he used the standard array. So, I was right.

You were indeed. And in hindsight it makes perfect sense. Admittedly, I had not thought stats would make that huge of a difference, but in this system, they do. And now that I know that, it seems obvious.
 
I'm still struggling to wrap my head around the idea that "not nearly so easy" is the expected outcome of l1s against two trolls. There's a previous-edition mindset in me somewhere that will take a crowbar to dislodge. I am coming to think more and more that this is entirely by design: A well-rested group of level 1s should have a chance to take out two trolls, a well-rested and fully prepared group of level 1s should be all but guaranteed to take out two trolls and have a chance at taking out three.

 It's a whole new world.
 

And yet, statistically speaking, with a group of 4 players, there is only a 31% chance that one of them will have an 18 in one stat. Statistically speaking, the group will not have anyone with a roll of an 18. And, the game can handle one person rolling an 18. What the game can't handle is multiple people rolling 18s, because that is a statistical anomaly. The math of the game is designed to make such situations very unlikely. When such situations happen anyway... well, statistical anomalies will result in under/overpowered groups. Choosing to roll invites the possibility of playing with a statistical anomaly. If, like me, you don’t like that, you shouldn’t allow groups to roll for stats…


AH... Did you read what the stats for his group was? Only the monk had a 20 and the other two had 18's AFTER adjustments. SO only one person rolled an 18 and the others rolled a 16 or 15(human) since they did +4 damage. So AGAIN, how is 1/3 of the time something the game shouldn't be able handle? You are acting like 31% is SO small that it should be ignored as a corner case. 31% isn't a statistical anomaly.




It is not the individual stats that are a problem, but the way the stats are grouped. There are too many good stats. The results are still a statistical anomaly if one uses the RAW. Of course, he didn't. Read the system he used. He describes it. That system will statistically result in much better stats on average! It is no surprise that his first run-through resulted in such a cake-walk. The second time, if you notice, things were not nearly so easy. The second time he used the standard array. So, I was right...


Even if you have the standard array, that only drops the monks attack/AC by one. The other characters, playing humans, can get 18's in their main stats so I don't see how this proved much if anything. They just had way less in their 'drop' stats. Also, it's hard to ANYONE is right without knowing how things went behind the scenes. The first time could have had more crits by players and the latter might have had more monster crits. One game might have just better die rolling in general for monsters or players. 1 run really isn't statisticly significant.

EDIT: To put it simply, I'm having a hard time thinking that a +1 stat modifier turned a hard fight into a cakewalk.

@ Gumba: Yup. In this system, a +1 to 3 difference makes a very big deal. 

@ elecgraystone: What you can roll is far less important than what your statistical grouping of rolls will look like. The problem was not one individual stat, but the groups of stats his players had. Together, it resulted in a big power difference. And, I can tell you given his description of the rolling system he used that would be the statistical norm. 

See, its not that one person had a +1. Its that across the board there were a lot of areas that were +1 to +3 points higher than they were likely to be. Together, given the bounded system, that adds up quickly.  

@ Gumba: The flip side to a group of 1st level characters having a much easier time against a small number of trolls is that a group of trolls will be a much larger threat to higher level PCs than they ever were before... at least if things are working right.