Merge High And Deep Elf


Ask fellow adventurers how to stay alive in the deep, dark, dangerous dungeon below, or share your own accumulated wisdom.

Slime Squisher

Posts: 387

Joined: Monday, 15th August 2011, 16:31

Location: Frankfurt

Post Friday, 14th October 2016, 15:58

Re: Merge High And Deep Elf

I'm not "people" and I tried to give reasons why IMO Te's restrictions don't matter as much as it would seem at first sight. I don't know -- was my argument too silly too even try to refute it?

Te get their direct EV bonus only at xlvl 5, but they have an indirect one from xlvl 1 on, since their aux attacks put their damage output way ahead of the killing curve and they can afford to raise dodging much more, much earlier than other species.
"... while we / Unburden'd crawl toward death." -- King Lear I,1
User avatar

Abyss Ambulator

Posts: 1194

Joined: Friday, 18th April 2014, 01:41

Post Friday, 14th October 2016, 15:59

Re: Merge High And Deep Elf

Early game accuracy is very bad. Health and direct ac/ev bonuses are more important than auxiliary attacks.
remove food

Slime Squisher

Posts: 387

Joined: Monday, 15th August 2011, 16:31

Location: Frankfurt

Post Friday, 14th October 2016, 19:08

Re: Merge High And Deep Elf

Well, if you absolutely insist ... who am I to argue?

For the record: Aux attacks don't depend on the weapon attack connecting. Which means that with every hit a Te gets two more chances to do damage. Which means that not only has a Te a higher average damage output, that avg. damage is also more evenly distributed.

But whatever. I don't give a fuck about this, anymore.
"... while we / Unburden'd crawl toward death." -- King Lear I,1

Ziggurat Zagger

Posts: 4432

Joined: Friday, 8th May 2015, 17:51

Post Friday, 14th October 2016, 19:11

Re: Merge High And Deep Elf

No need to get angry, let me run some fsims for HuGl, KoGl, GrGl and TeGl when I have a chance ;)
This is an interesting topic to discuss IMHO.
Underestimated: cleaving, Deep Elf, Formicid, Vehumet, EV
Overestimated: AC, GDS
Twin account of Sandman25

Ziggurat Zagger

Posts: 8786

Joined: Sunday, 5th May 2013, 08:25

Post Friday, 14th October 2016, 21:26

Re: Merge High And Deep Elf

Utis wrote:For the record: Aux attacks don't depend on the weapon attack connecting.
They depend on the aux attack connecting.

Ziggurat Zagger

Posts: 4432

Joined: Friday, 8th May 2015, 17:51

Post Friday, 14th October 2016, 22:17

Re: Merge High And Deep Elf

XL 1 with flail vs jackal

TeGl:
  Code:
           AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
Attacking:      5,5 |     17 |      53% |   2,9 |   124  |  0,81 |      2,3
Defending:      0,9 |      3 |      39% |   0,3 |    71  |  1,41 |      0,5


HuGl:
  Code:
           AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
Attacking:      4,3 |     12 |      47% |   2,1 |   125  |  0,80 |      1,7
Defending:      0,7 |      3 |      31% |   0,2 |    71  |  1,41 |      0,3


KoGl:
  Code:
           AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
Attacking:      4,1 |     11 |      51% |   2,1 |   125  |  0,80 |      1,7
Defending:      0,7 |      3 |      28% |   0,2 |    71  |  1,41 |      0,3


Conclusion: Hu is better than Ko/Te which are about equal to each other.

Spoiler: show
Because damage-wise there is no significant difference (though 2.3/0.5 still is less than 1.7/0.3 but the latter can be 1.7/0.39 in which case the former is greater) and Hu has 100% HP while Te/Ko have 80% HP
Underestimated: cleaving, Deep Elf, Formicid, Vehumet, EV
Overestimated: AC, GDS
Twin account of Sandman25

For this message the author VeryAngryFelid has received thanks:
nago
User avatar

Vaults Vanquisher

Posts: 510

Joined: Friday, 1st July 2016, 22:32

Location: Aachen, Germany

Post Friday, 14th October 2016, 22:52

Re: Merge High And Deep Elf

If this is how you measure which race is "better", naga must probably be one of the best.

Ziggurat Zagger

Posts: 4432

Joined: Friday, 8th May 2015, 17:51

Post Friday, 14th October 2016, 23:15

Re: Merge High And Deep Elf

If this is how you read threads... ;)
Underestimated: cleaving, Deep Elf, Formicid, Vehumet, EV
Overestimated: AC, GDS
Twin account of Sandman25

Slime Squisher

Posts: 387

Joined: Monday, 15th August 2011, 16:31

Location: Frankfurt

Post Saturday, 15th October 2016, 07:11

Re: Merge High And Deep Elf

VeryAngryFelid wrote:No need to get angry, let me run some fsims for HuGl, KoGl, GrGl and TeGl when I have a chance ;)
This is an interesting topic to discuss IMHO.


Well, okay. My apologies. This kind of strategic considerations is what keeps me interested in crawl. (Winning by itself started to loose its appeal after the second time and I can't get myself to care about other meta-game goals like win percent, streaking, speed runs or stunt runs). I won't mind being proven wrong, but I'd like to be taken serious as I did do my homework on this. FWIW, here's an (edited, translated) excerpt from my notes from back then when I was experimenting with melee Tengu:

All chars were using a flail. When I did this, riposte didn't exist. I'd be curious whether long blades change the picture. On paper, riposte should have the effect of lending an offensive aspect to training dodging. I have no idea whether that actually matters.

All fsim runs were done with fsim_rounds = 50000. I chose hound as fast early game monster that is dangerous, but not so dangerous that you would normally just avoid it. Hu level faster than Te, so the skill lvl for Hu was taken from an OpGl at xlvl 4 and 5, respectively. (Op level at the same speed as Te, IIRC, but have the same weapon skill as Hu -- also IIRC.) The relevant baseline for comparision is the ratio between "average turns until monster is killed" and "average turns until player is killed".

Edit: The human was assumed to train only his weapon skill. Te50 trained 50% weapon skill and 50% dodging. Te66 trainend 33% weapon skill and 66% dodging.

Edit2: I don't remember whether I took the level increase in attributes into account. I think, I didn't. FWIW.

  Code:
* xl 1 vs. jackal

| Char | HP | M&F | Ddg |
|------+----+-----+-----|
| HuGl | 16 |   3 |   3 |
| TeGl | 12 | 3.2 | 3.2 |

** HuGl

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      4,4 |     12 |      48% |   2,1 |   125  |  0,80 |      1,7
_Defending:      0,7 |      3 |      43% |   0,3 |    71  |  1,41 |      0,4

** TeGl

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      5,6 |     18 |      51% |   2,9 |   124  |  0,81 |      2,3
_Defending:      0,8 |      3 |      39% |   0,3 |    71  |  1,41 |      0,5

** Comparision

Assuming the jackal has 8HP (worst case). Assuming the worst
case rather than the average case slightly favours the Human.

| Char | t_kill | t_killed | remainder | ratio  |
|------+--------+----------+-----------+--------|
| HuGl | ~ 4,71 |       40 | ~ 35.29   | ~ 0.12 |
| TeGl | ~ 3.48 |       24 | ~ 20.52   | ~ 0.14 |

t_kill: number of turns until foe is killed (= Monster HP / AvEffDam attacking)
t_killed: number of turns until char is killed (= char HP / AvEffDam defending)
remainder: t_killed - t_kill (i.e.: How many more turns
           could the character survive if a second foe of the same type
           turns up immediately after the battle.)
ratio: t_kill / t_killed (lower is better)

* xl 4 vs. hound

Skills etc.

| Char | HP | M&F | Ddg |
|------+----+-----+-----|
| Hu   | 33 | 4.7 |   3 |
| Te50 | 26 | 4.3 | 4.3 |
| Te66 | 26 | 4.0 | 4.7 |

** HuGl

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      4,5 |     13 |      46% |   2,1 |   116  |  0,86 |      1,8
_Defending:      1,9 |      6 |      49% |   0,9 |    66  |  1,52 |      1,4

** TeGl50

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      5,8 |     18 |      47% |   2,8 |   119  |  0,84 |      2,3
_Defending:      2,2 |      6 |      46% |   1,0 |    66  |  1,52 |      1,5

** TeGl66

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      5,7 |     19 |      47% |   2,7 |   120  |  0,83 |      2,3
_Defending:      2,2 |      6 |      41% |   0,9 |    66  |  1,52 |      1,4

** Comparision

Assuming the hound has 23 HP (worst case). Assuming the worst
case rather than the average case slightly favours the Human.


| Char | t_kill  | t_killed | remainder | ratio  |
|------+---------+----------+-----------+--------|
| Hu   | ~ 12.78 | ~ 23,57  | ~ 10.79   | ~ 0.54 |
| Te50 | 10      | ~ 17.33  | ~ 7.33    | ~ 0.58 |
| Te66 | 10      | ~ 18.57  | ~ 8.57    | ~ 0.53 |


* xl 5 vs. hound

| Char | HP | M&F | Ddg |
|------+----+-----+-----|
| Hu   | 38 | 5.5 |   3 |
| Te50 | 30 | 4.8 | 4.8 |
| Te66 | 30 | 4.2 | 5.3 |

In the fsim, Te were flying.

** HuGl

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      4,6 |     13 |      46% |   2,2 |   113  |  0,89 |      1,9
_Defending:      1,8 |      6 |      49% |   0,9 |    66  |  1,52 |      1,4

** TeGl50

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      5,7 |     18 |      48% |   2,8 |   116  |  0,86 |      2,4
_Defending:      2,2 |      6 |      35% |   0,8 |    66  |  1,52 |      1,2

** TeGl66

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      5,8 |     18 |      47% |   2,8 |   119  |  0,84 |      2,3
_Defending:      2,2 |      6 |      31% |   0,7 |    66  |  1,52 |      1,0


** Comparision

Assuming the hound has 23 HP (worst case). Assuming the worst
case rather than the average case slightly favours the Human.

| Char | t_kill  | t_killed | remainder | ratio  |
|------+---------+----------+-----------+--------|
| Hu   | ~ 12.11 | ~ 27.14  | ~ 15.04   | ~ 0.45 |
| Te50 | ~ 9.58  | 25       | ~ 15.42   | ~ 0.38 |
| Te66 | 10      | 30       | 20        | ~ 0.33 |



Originally, I stated that a Tengu is slightly better at xlvl 1, then quickly becomes slightly worse, and then becomes slightly better again at xlvl5. The numbers support the last two parts of my statement, but not that the Tengu would be better at xlvl 1. This latter assessment was based on my intuition from game play experience. Either my intuition or my calculation must be wrong. Maybe it's because at xlvl 1 surviving at all matters more than how you fare on average? A xlvl 1 Tengu is more vulnerable to spikes in damage intake, but he also has a slightly better accuracy with damage spikes on his own which maybe reduces the likelihood of receiving bad damage rolls enough to more than offset his vulnerability. I don't know.

Fsim seems to take aux attack into account in AvHitDam (attacking). So, looking back at the numbers, I also have to qualify another part of my statement: Tengu do have a better accuracy, but the difference is not as large as I thought it would be. Still ...
"... while we / Unburden'd crawl toward death." -- King Lear I,1

Ziggurat Zagger

Posts: 4432

Joined: Friday, 8th May 2015, 17:51

Post Saturday, 15th October 2016, 13:00

Re: Merge High And Deep Elf

That's quite a lot of tests you did, nice. Sorry, I believe you have just proved that Te is inferior to Hu before XL 5. Nobody is going to disagree that Te is better than Hu at XL 5, being able to flee from 90% monsters alone is better than 20% HP.
In these tests we see
Assuming the jackal has 8HP (worst case). Assuming the worst
case rather than the average case slightly favours the Human.

  Code:
| Char | t_kill | t_killed | remainder | ratio  |
|------+--------+----------+-----------+--------|
| HuGl | ~ 4,71 |       40 | ~ 35.29   | ~ 0.12 |
| TeGl | ~ 3.48 |       24 | ~ 20.52   | ~ 0.14 |


40 turns to be killed by a single jackal means we can almost guarantee that we are fighting one jackal at a time, we have lots of time to decide what to do, try different consumables, run into corridor etc.
Also let's not forget that we have lots of very dangerous monsters who have ranged attack. Orc Priest, Orc Wizard, Centaur, all those are much more dangerous to Te than to Hu (speaking about XL 1-4 here, of course).
Underestimated: cleaving, Deep Elf, Formicid, Vehumet, EV
Overestimated: AC, GDS
Twin account of Sandman25

Slime Squisher

Posts: 387

Joined: Monday, 15th August 2011, 16:31

Location: Frankfurt

Post Saturday, 15th October 2016, 14:22

Re: Merge High And Deep Elf

Yeah, well, I noticed that myself. The thing is, humans are pretty good, so "slightly worse than a human until xlvl 5" makes still a decent melee char. Also, whether that makes them "slightly better" or "slightly worse" overall depends on context: If you are into streaking, they are worse. If not, then getting to xlvl 5 takes, like, 3 minutes of your lifetime and if you prefer having an edge from Temple to, I don't know, somewhere in the midgame, then they are better.

(As an aside: The numbers just denote tendencies. From the numbers alone one would have to deduce that dungeon level 1 is easier than dungeon level 3. We all know that's not the case. The reason is that the HP pool is smaller, the accuracy lower combined with higher damage relative to HP, and there are less consumables. For these reasons, spikes have a more devastating impact. You can't really rely on having ~40 turns adjacent to a jackal. But I'm not going to defend my believe that Te have an edge at xl 1. I might very well be wrong about this point and I have no argument that I, myself, find solid. Maybe try to get a couple of HuFi and TeFi (with 33-66) to lvl 2, so everybody can see for him- or herself. -- Gladiator seems more natural, but I formed this opinion while experimenting with TeFi, so maybe AC and SH factor in, I don't know.)
"... while we / Unburden'd crawl toward death." -- King Lear I,1

For this message the author Utis has received thanks:
VeryAngryFelid

Ziggurat Zagger

Posts: 4432

Joined: Friday, 8th May 2015, 17:51

Post Saturday, 15th October 2016, 15:20

Re: Merge High And Deep Elf

Utis wrote:Maybe try to get a couple of HuFi and TeFi (with 33-66) to lvl 2, so everybody can see for him- or herself. -- Gladiator seems more natural, but I formed this opinion while experimenting with TeFi, so maybe AC and SH factor in, I don't know.)


A few years ago I created a thread asking for advice about TeAE, I kept dying too often, both early game and in Lair. So it is easy for me to believe that Te is a challenging species unless you are extremely careful and patient. It's a bit similar to Felid in this regard probably.
If a race is worse than Hu, I treat it as weak race. If I want to have easy late game, I prefer to choose Na or Ds, it is more fun and more powerful IMHO.
Underestimated: cleaving, Deep Elf, Formicid, Vehumet, EV
Overestimated: AC, GDS
Twin account of Sandman25

For this message the author VeryAngryFelid has received thanks:
Rast

Slime Squisher

Posts: 387

Joined: Monday, 15th August 2011, 16:31

Location: Frankfurt

Post Sunday, 16th October 2016, 08:18

Re: Merge High And Deep Elf

It's not at all like a Felid. Seriously, would you please just start a TeFi? Set its training to 33% weapon skill and 66% dodging at turn 1. This is important. Keep it there until xl 5 or 6 and don't upgrade to chain mail or plate mail. Otherwise just play normally. And please play a couple of early game HuFi before that to reset your expectations: There's no point to this exercise if you played an early game Troll before.

I mean the "slightly" part of "slightly worse" or "slightly better" very literally: Even between HuGl and TeGl at xl 1, the difference in the ratio between "turns until monster is killed" and "turns until player is killed" is 0.02. This number corresponds directly to the average loss of HP after a fight. So, we're talking about 2% of 12 and 16 respectively. That's probably too small to notice. Yes, the Tengu is more vulnerable to spikes, so with the melee Te, you once in a while have to start pillar dancing in situations where the Hu could affort to make one more hit. But spikes also work in favour of the Tengu, like when both weapon and aux attack connect. So in many situations where the Hu would have to start pillar dancing, the Te just one-shotted his hobgoblin enemy. It's more or less even steven.

And I just checked in fsim: If it's HuFi vs TeFi, the TeFi does indeed look numerically better. So AC and SH did play a role. But the numbers involved are even smaller. So, whatever.

It's not at all like a Ds, either. I mean, if it's a matter of taste, I can respect that. E.g. I don't like trolls, I don't like spriggans, but that dislike makes neither of them weaker than they are in my book. Te loose some of their advantages in the late game/late midgame in the sense that their relative impact becomes much smaller while the impact of their drawbacks doesn't diminish as much. On the other hand you typically reach xlvl 5 on dungeon level 3. That's still the early game.
"... while we / Unburden'd crawl toward death." -- King Lear I,1

Ziggurat Zagger

Posts: 4432

Joined: Friday, 8th May 2015, 17:51

Post Sunday, 16th October 2016, 12:34

Re: Merge High And Deep Elf

Utis wrote:I mean the "slightly" part of "slightly worse" or "slightly better" very literally: Even between HuGl and TeGl at xl 1, the difference in the ratio between "turns until monster is killed" and "turns until player is killed" is 0.02. This number corresponds directly to the average loss of HP after a fight.


That's misleading. It's 0.12 and 0.14, the difference is better described as 16% instead of 0.02, it is not too far from 20% HP difference.

I am not sure why you recommend playing Fi, did you mean Gl? Fi relies on SH mostly and decreases EV via shield penalty.
Underestimated: cleaving, Deep Elf, Formicid, Vehumet, EV
Overestimated: AC, GDS
Twin account of Sandman25

Slime Squisher

Posts: 387

Joined: Monday, 15th August 2011, 16:31

Location: Frankfurt

Post Sunday, 16th October 2016, 15:54

Re: Merge High And Deep Elf

Not sure if I'm misunderstanding you. The numbers mean: after fighting a jackal a xl 1 TeGl has lost on average 14% of its HP, a HuGl 12%. Not accounting for spikes, of course.

Mostly, I'm recommending TeFi, because that's what was experimenting with when I formed my opinion on Te-strategy. So, this is where I'm most confident in my evaluation. I don't think I played many TeGl.

Here are the TeFi vs. HuFi against jackal results that I mentioned earlier. Again xl 1 and with flails. This time the Tengu is ahead:

  Code:
* Fighter xlvl 1 vs. jackal

| Char | HP |
|------+----|
| HuFi | 18 |
| TeFi | 14 |


** HuFi

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      4,5 |     13 |      43% |   2,0 |   138  |  0,72 |      1,4
_Defending:      0,5 |      3 |      46% |   0,2 |    71  |  1,41 |      0,3


** TeFi

            AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam
 Attacking:      5,5 |     19 |      47% |   2,6 |   138  |  0,73 |      1,9
_Defending:      0,5 |      3 |      45% |   0,2 |    71  |  1,41 |      0,3

** Comparision

Assuming the jackal has 8HP (worst case). Assuming the worst
case rather than the average case slightly favours the Human.


| Char | t_kill | t_killed | ratio   |
|------+--------+----------+---------|
| Hu   | ~ 5.71 | 60       | ~ 0.095 |
| Te   | ~ 4.21 | ~ 46,67  | ~ 0.090 |



Upon reflexion, this is not really surprising. In the low levels 1 AC > 1 EV, which is (at least part of) the reason that previously the xl 1 HuGl was slightly better than the TeGl, since the former starts with a helmet and has 1 AC more. As you can see, fighters already start with better defense but with worse offense. This favours the Tengu, since its aux attacks do not depend on weapon skill. The higher HP from fighting skill also helps a little.

Gameplay-wise, both AC and SH soften the Tengu's vulnerability to receiving damage spikes while not hindering his ability to dish them out.

Still, after thinking about it, my impression that at xl 1 TeFi is better than HuFi was probably just clustering illusion. On average fights are over slightly more quickly than with a Human and you are notably one-shotting d1 enemies more often. That makes it easy to gain the impression that you're doing better when overall you're just doing just as well.

It's still important to raise dodging, though, and not armour. The strategy I'm recommending here is to use the Tengu's head start in damage output to raise defense and thus offset his HP penalty. This means you're falling behind in weapon skill (though not in damage output). The strategy is then to use the 20% EV boost you get at xl 5 to train offense again. Not necessarily immediately at xl 5, but around that time or soon after. This should go fast at this point. One of the more subtle advantages is that this means that you have invested little in weapon skill, which might mean that it's more feasible to switch to a different weapon category if you find something that provides a clear advantage. Or you can branch into magic if you find spells that have a good synergy with melee, like conjure flame or sticky flame. For a TeFi that can make very good sense even well before Lair.
"... while we / Unburden'd crawl toward death." -- King Lear I,1

Ziggurat Zagger

Posts: 4432

Joined: Friday, 8th May 2015, 17:51

Post Sunday, 16th October 2016, 16:11

Re: Merge High And Deep Elf

Utis wrote:Not sure if I'm misunderstanding you. The numbers mean: after fighting a jackal a xl 1 TeGl has lost on average 14% of its HP, a HuGl 12%.


If we look at it this way, we should ignore jackals completely I think.
Fi is the best background defense-wise, of course attack becomes more important with it because of diminishing returns on defense.
Underestimated: cleaving, Deep Elf, Formicid, Vehumet, EV
Overestimated: AC, GDS
Twin account of Sandman25
User avatar

Pandemonium Purger

Posts: 1386

Joined: Sunday, 5th April 2015, 22:37

Post Monday, 17th October 2016, 00:01

Re: Merge High And Deep Elf

If high elves are really to similar to deep elves, please just buff their HP apt by+1 or +2 instead of removing them. This should make them different enough from other species and has the upside that you can play a super powerful elf.
http://crawl.akrasiac.org/scoring/playe ... speon.html. I started playing in 0.16.1
I achieved greatplayer in less than a year.
Remove food

Slime Squisher

Posts: 387

Joined: Monday, 15th August 2011, 16:31

Location: Frankfurt

Post Tuesday, 18th October 2016, 07:25

Re: Merge High And Deep Elf

VeryAngryFelid wrote:
Utis wrote:Not sure if I'm misunderstanding you. The numbers mean: after fighting a jackal a xl 1 TeGl has lost on average 14% of its HP, a HuGl 12%.


If we look at it this way, we should ignore jackals completely I think.


It just means that Gl is friggin' good at xl 1.
"... while we / Unburden'd crawl toward death." -- King Lear I,1
Previous

Return to Dungeon Crawling Advice

Who is online

Users browsing this forum: No registered users and 116 guests

Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group.
Designed by ST Software for PTF.