The Tavern

How many skills should I be having on when specialising, what is the maximum amount of skills I should have on at any time and what indicators should I use to know when to drop a skill, put one on or to stop specialising

Sent from my iPhone using Tapatalk

CypherZel wrote:How many skills should I be having on when specialising

1

CypherZel wrote:what is the maximum amount of skills I should have on at any time

1

CypherZel wrote:and what indicators should I use to know when to drop a skill, put one on or to stop specialising

you should train the skill that currently gives you the most benefit for its cost, generally you will find that you want to switch skills at whole number skill levels since whole numbers are breakpoints for all skills, with the notable exception that you want to switch off of dodging/armour/shields right after gaining a point of AC/EV/SH instead (since there is no fractional AC/EV/SH)

So throughout the entire game, just train a skill until it reaches a whole number value, and usual the one that will benefit me the most at this time as well as cost the least? This pretty much destroys every thin I thought about skilling

Sent from my iPhone using Tapatalk

You should train one skill at a time because at any moment in the game, there is one skill that will benefit you more than any other skill. So by training multiple skills at once, you get less benefit for your xp.

CypherZel wrote:So throughout the entire game, just train a skill until it reaches a whole number value, and usual the one that will benefit me the most at this time as well as cost the least? This pretty much destroys every thin I thought about skilling

Make sure you don't misunderstand; the training comes in small chunks. So don't go thinking that Duvessa means your MiFi should train weapon skill up to 12 before you train anything else.

e.g. newly created SpEn might train stealth from 4.4 to 5.0, then spellcasting from 2.6 to 3.0, then Ice from 0.0 to 2.0 then Hexes from 3.6 to 5.0 then Dodging until EV is 20

Note that you can easily win without playing this optimally.

I'd rather play optimally though as I'm really trying to learn the best strategies

Sent from my iPhone using Tapatalk

Trust me, you'd rather wouldn't. There's a huge gap in Crawl between playing "well enough" and playing "optimally".

Could you elaborate?

Sent from my iPhone using Tapatalk

Optimal play means using a lot of strategies during every moment of the game, and that is incredibly tedious for a game which lasts around 3/5 hours and which actually doesn't require that for the largest amount.
For example, you should lure every single enemy - no matter if dangerous or not - to the nearest staircase. You should manage carefully the noise made by your actions, or ideally by your enemies. You should run away every time you suffer a swingy big hit in a combat - I mean, you should reset a lot more fights than you actually do in a normal game. You shouldn't use autoexplore, to minimize the number of new titles appearing for each step, or to minimize the risk of stepping in a bad position in a nasty vault or situation. You should micromanage your skills. And so on.

This is what most (good) player) do in early game, if they want to maximize the chance of surviving, because early on the char is most weak in comparison of monsters and has less way to deal with bad or unlucky situations. After the very first floors this isn't true anymore, because consumables start to exist, as well gods' powers, while the char has usually gained enough exp\equipment to not suck anymore.
Later, both of those are over-abundant and char is generally much stronger than average monster, so one could just mindlessly tab for most of the game, without worrying of optimal play during most situations or about min-max the stat and still win with a very good margin of success - up 100% if the very few dangerous situations are identified and deal with in time.

And if you don't care about winrate, being optimal in the early game isn't even worth it. You save more time by using autoexplore and not worrying too much about dying during D1-3.

Ah thanks, I just want to be able to win every once and awhile

Sent from my iPhone using Tapatalk

CypherZel wrote:Ah thanks, I just want to be able to win every once and awhile

Sent from my iPhone using Tapatalk

Winning once in a while is mostly about learning to not make terrible mistakes. You can rack up wins with pretty poor average play because every few games you'll get killer gear and the rng will forgive you a couple of misplays. To win consistently you'll want to have a good understanding of tactics, and a solid general knowledge of consumables, god powers, spells etc. so you can make good use of dungeon finds. Skilling isn't really that important because, so long as you don't absolutely cripple yourself, good tactics and consumable use will get you through more often than not.

Optimal play is something else. It's like you're not even playing the game well anymore, but instead just using patience to avoid situations where you have to play well. Moreover, the massive extra effort burden of optimal play over good and careful play will probably only up your win rate by a few percent. At the moment I believe the best win rates among players who play a variety of characters (sometimes chosen at random) is around 70%. Optimal play isn't going to get you much more than that because sometimes the early dungeon just kills you.

Hands wrote:At the moment I believe the best win rates among players who play a variety of characters (sometimes chosen at random) is around 70%. Optimal play isn't going to get you much more than that because sometimes the early dungeon just kills you.

Code:: <amalloy> !streak zxc <Sequell> zxc has 33 consecutive wins (HOBe, DsMo, HaHu, NaIE, FoAK, MfVM, GrFE, TrWn, MiSk, DDGl, CeWr, VSAs, DrWz, HuFi, OgAM, SpEn, KoAr, HECj, VpNe, DEEE, TeSu, FeTm, DgAE, GhCK, OpCj, MuAs, MiWn, TrSk, DDEE, MfAK, GrGl, DrFE, TrAM), and can keep going! <amalloy> !streak :elliptic <Sequell> :elliptic has 43 consecutive wins (TeCj, KoEE, VSEE, NaNe, DDFi, KoCj, OpMo, CeCK, DgEE, MiCK, HEAs, VSFi, CeAE, DrAr, OgNe, NaBe, HuAE, SpAM, OpFE, DDAs, FoSk, SpIE, VSSk, GrTm, HOSk, HOGl, MiAK, HuAK, OgHu, DDAr, FoFE, MiWr, DrVM, SpWz, TrVM, GrBe, TrAE, FeCj, MfAs, GhWr, VpIE, DDSu, TrAM) and has won their last 11 games (HESu, GhVM, GhAr, GhHu, GhSk, GhCj, GhCK, GhAK, GrEE, CeEE, DrEE).

If the chance of winning with optimal play were 70%, the probability of an optimal player getting a 43-game win streak would be 2.18381438e-7: a one in five million occurrence. elliptic has played only 613 games, so it is very unlikely that he would luck into such a streak. zxc's streak is of course not quite as unlikely, but still well outside the realm of probability. I think these streaks are strong evidence that you can do a lot better than 70% if you play with extreme caution.

We need an alphacrawl!

I have the most success with melee training weapon for the first few levels or skill level 6-8 as long as it's with scale or ring mail equpied. Seems to make it much easier.

I tend to leave my primary skill at 40% or 50% of exp. Then I train other skills.

amalloy wrote:If the chance of winning with optimal play were 70%, the probability of an optimal player getting a 43-game win streak would be 2.18381438e-7: a one in five million occurrence. elliptic has played only 613 games, so it is very unlikely that he would luck into such a streak. zxc's streak is of course not quite as unlikely, but still well outside the realm of probability. I think these streaks are strong evidence that you can do a lot better than 70% if you play with extreme caution.

I don't think this is the correct way to look at that probability. This is the probability that you could 'call the streak in advance' so to speak. That is, that, assuming a .7 win rate, a specific player would win 43 games in a row starting from their next game. These big streaks are the peaks across all online play. What we're interested in is the probability that some player could at some point in time have recorded a streak like that. This would be low, but higher than .7^43.

You are probably right about the optimal play win rate being significantly higher than 0.7 though. With strong combos play it's likely to be very close to 100%. My intended point was that for any given combo, optimal play only gives you a small extra chance of winning over very good play.

The optimal play winrate is about 100%, very slightly less.

edit: taking into account the number of times 43 games have been played by a single player in a row online, you still only knock off a few orders of magnitude. It is still overwhelmingly improbable that you would see streaks of this length if optimal play really resulted in a winrate as low as 70%.

goodcoolguy wrote:100%

Don't forget to take into account the situation where you fall down a shaft trap inside the entrance vault and land on top of a D:4 priest who instantly smites you for more than you have max hp.

I mean that isn't exactly very common but it CAN happen.

genericpseudonym wrote:
goodcoolguy wrote:100%

Don't forget to take into account the situation where you fall down a shaft trap inside the entrance vault and land on top of a D:4 priest who instantly smites you for more than you have max hp.

I mean that isn't exactly very common but it CAN happen.

he literally said "very slightly less" as the next three words

goodcoolguy wrote:The optimal play winrate is about 100%, very slightly less.

edit: taking into account the number of times 43 games have been played by a single player in a row online, you still only knock off a few orders of magnitude. It is still overwhelmingly improbable that you would see streaks of this length if optimal play really resulted in a winrate as low as 70%.

With strong combos the win rate would be up around 100%. I think the number for bad combos would be significantly lower. Maybe not as low as 70%. I've already backed away from that claim.

It would be very interesting to see the winrate for MuCK among the best players...

I don't think there are many combos so bad the optimal play winrate goes down much. In the statement I gave above, I'm assuming random combos, i.e. an unweighted average across all combos.

Sprucery wrote:It would be very interesting to see the winrate for MuCK among the best players...

It does not depend on player skill much IMHO. It is decided by Xom.

Edit. Assuming optimal player always does 100% correct actions we still will have rather low win rate due to Xom.
I remember how I spectated a FoCK who got polymorphed into Ice Beast with Orb of Fire in view soon after I joined as the only spectator

Oh FoCK indeed

Sandman25 wrote:It does not depend on player skill much IMHO. It is decided by Xom.

I think it is not true. The faster you are the less chances Xom has to kill you - and you need to be really skillful to reach a low turncount safely.

sanka wrote:I think it is not true. The faster you are the less chances Xom has to kill you - and you need to be really skillful to reach a low turncount safely.

It depends on what you mean as faster here.
Do you mean it is a good idea to use manual exploration? I agree, it is a good idea with other gods also.
Do you mean it is a good idea to explore below full HP? I don't think so.
Do you mean it is a good idea to avoid unnecessary traveling? I agree, it is a good idea with other gods (at least which have piety) also.
Do you mean it is a good idea to avoid Crypt/Elf? It depends, those areas are similar to Abyss in danger and can be useful.
Do you mean it is a good idea to avoid exploring the whole floor and dive instead? I am not sure, it depends. If you are powerful enough for Abyss and have enough teleportation sources, it can be a good idea to explore everything and find that +23 CPA on the ground. If you are weak for Abyss, you can dive to Orc 2 shops to buy +5 CPA in a shop.

Also it's interesting that there was an old thread here where player was asking for advices with MuCK, he/she died every time. He/she got advice to scum Xom i.e. spend much time pressing 5 in Ice Cave and alike and after following that advice he/she won in the very next attempt. Xom granted some cool items like rods and wands as far as I remember.
So I am not sure being fast is good with MuCK.

I am quite sure.

I think you miss that if you are playing safely your turn count will still depend on skill very strongly. Better players take much less damage -> less resting. They judge much better where it is not worth to fight - note that it is not the same as whether the fight is winnable - for example I do not think that its always worths to clear the gnoll castle with MuCK, because it takes too much time.

In my experience with Xom scumming it does not have a better chance for winning. You could get very strong gifts just by playing, and then you are also improving, so making XOm's sticky flame, banishment etc. much less deadly.

I guess we are talking about different things. FWIW I didn't scum my MuCK, I just played normally.

The Tavern

Advice on skill training

Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training

Re: Advice on skill training