|
On October 09 2014 07:43 GrandInquisitor wrote:
Researching that question led me to create this runes quiz:
1) What are the five Quints for which there are no equivalent non-Quints?
2) What are the two non-Quints for which there are no equivalent Quints?
3) What are the seven runes that do not exist in level 1 / level 2 form?
Wiki-fu go!
1) Exp%, Lifesteal, MS, Spellvamp, death reduction %
2) Energy/5/lvl, Energy/lvl
3) Energy, Energy/lvl, Energy/5, Energy/5/lvl, Exp%, Gp10, death reduction %
|
On October 09 2014 07:17 Goumindong wrote:
Crit is an exceptionally bad mechanic in league it absolutely should be removed
Get luckier? It's not hard.
|
|
|
On October 09 2014 07:50 xes wrote:Show nested quote +On October 09 2014 07:43 GrandInquisitor wrote:
Researching that question led me to create this runes quiz:
1) What are the five Quints for which there are no equivalent non-Quints?
2) What are the two non-Quints for which there are no equivalent Quints?
3) What are the seven runes that do not exist in level 1 / level 2 form? Wiki-fu go! 1) Exp%, Lifesteal, MS, Spellvamp, death reduction % 2) Energy/5/lvl, Energy/lvl 3) Energy, Energy/lvl, Energy/5, Energy/5/lvl, Exp%, Gp10, death reduction %
Researching this answer made me realize how good the IP curve on runes are:
T1 runes are amazingly cost efficient for the IP while T3 runes have the most slot efficiency at the worst cost efficiency.
So why can't Riot make items like this if they can price their runes like it.
|
I am looking forward to playing Azir again after his 15 bugfixes tomorrow.
I just checked, it's 15.
|
watching bjerg play vs azir and they were crushing early game, zion went 19-1 on ryze and still cant finish the game cuz azir 1v9 stalls the game, wont be surprised if the bugfixes + buffs will make him perma ban status
|
On October 09 2014 07:54 xes wrote:Show nested quote +On October 09 2014 07:50 xes wrote:On October 09 2014 07:43 GrandInquisitor wrote:
Researching that question led me to create this runes quiz:
1) What are the five Quints for which there are no equivalent non-Quints?
2) What are the two non-Quints for which there are no equivalent Quints?
3) What are the seven runes that do not exist in level 1 / level 2 form? Wiki-fu go! 1) Exp%, Lifesteal, MS, Spellvamp, death reduction % 2) Energy/5/lvl, Energy/lvl 3) Energy, Energy/lvl, Energy/5, Energy/5/lvl, Exp%, Gp10, death reduction % Researching this answer made me realize how good the IP curve on runes are: T1 runes are amazingly cost efficient for the IP while T3 runes have the most slot efficiency at the worst cost efficiency. So why can't Riot make items like this if they can price their runes like it.
Because constricting gold doesn't make Riot money.
|
On October 09 2014 07:54 Terranasaur wrote:
New Sion any good?
Watched nightblue play him on pbe looked useless.
|
On October 09 2014 07:40 GrandInquisitor wrote:Show nested quote +On October 09 2014 07:17 Goumindong wrote:On October 09 2014 07:01 Gahlo wrote:On October 09 2014 06:57 Goumindong wrote:Requizen wrote:
Randomness is not a comeback mechanic. Randomness is a comeback mechanic*, its just not a very good one because it sucks to play with and against. If you're ahead, more randomness always increases the likelihood that you will lose and similarly less randomness increases it. The same is true the other way around. *excepting random mechanics which have selection options. Those are neutral with regards to the strength of the teams but still bad mechanics. Crit is a comeback mechanic confirmed. It is. If there is no crit the probability that the losing teams ADC does more damage than the winning teams ADC is roughly zero. If there is crit, even when the winning ADC has more crit than the losing ADC the probability is non-zero. Crit will turn some fights that the loser should have lost into fights they win. Crit also has the negative aspect of selection in lane (that is, if you get a lucky crit in lane you can choose to all-in/hard zone after this because the likelihood you will lose is exceptionally small. Whomever gets the first crit effectively wins the lane). The 1% crit rune is a strong option for this reason. Suppose that you pick the 1% crit rune and your opponent does not. You're still trading relatively equally, equally enough that its hard to shove you out of lane on their .97 AD advantage. With 1% crit over 10 attacks the probability you will land at least 1 crit is 9.4%. Over 20 attacks its 18.2%. So if you take a 1% crit rune and expect to trade roughly 20 auto attacks in a lane you can expect that your 1% crit rune will flat win you the lane for free about 18.2% of the time.(supposing they don't have a crit rune as well, in which case its 50/50 for first crit) Crit is an exceptionally bad mechanic in league it absolutely should be removed By this logic, why not 2% crit? 3% crit? 9 crit chance marks means that over 20 AA's you have an 83.6% chance of a crit.
Because while there clearly is a maximization problem to be done* fuck if i am going to do it.
*edit: this wasn't worded well. For every level of risk tolerance there will exist a tradeoff between damage/crit that some people are willing to accept. Because things are never certain there is never a clear cut answer to the question. Though we could suggest that you should be risk neutral in league this might not represent the individual players preferences.
Obviously at 1 crit red we can make the argument that you should, because the damage buildup is not large compared to the highish probability you win the lane for free. Basically we don't lose the lane on damage buildup before going back so we see the crit trade off as reasonable. Similarly at 9 crit reds we can suggest that the total damage lost (8.73 per hit!) suggests that by the time we get a crit (mean time to first crit will be roughly 11 attacks) we will expect to have already lost the lane by the time we get a crit.
More generally we can just estimate
.97 damage = .93% crit in reds. So at 20 attacks you can estimate
1 Crit Red
Total Damage Lost over 20 attacks = 19.4
Probability of at least one crit over 20 attacks=17.04%
2 Crit Reds
Total Damage Lost = 38.8
Probability of Crit = 31.3%
3 Crit Reds
Total Damage Lost = 58.2
Probability of Crit = 43.2%
Total damage lost goes up as you add more abilities and more of an ability/attack ratio. The effect of the crit goes up as you add total AD. The value of crit goes down if you get less auto attacks trades off in lane. The effect of crit goes down when you're winning lane (Since the expected damage from the few % crit will almost certainly be lower than the expected total damage lost)
I can maybe see 3 crit reds being worth it but anything over that is going to be really iffy because you're going to start to lose out on appreciable damage very quickly unless you crit.
At 4 crit reds you're losing almost 4 damage/hit which adds up fast even if you have a 53% probability to get at least 1 crit in 20 attacks if you don't get those crits you're down 77 damage. You've already been pushed out of lane for it or worse.
edit: A better way to think about it might be "how many auto attacks do you expect to trade in lane without getting pushed out before you have to back anyway" and then maximizing based on the amount of damage you can be down at that point without risking an all-in. If that number is 5 obviously you want more crit, if that number is 10 then obviously less.
|
United States23745 Posts
On October 09 2014 07:56 Ketara wrote:
I am looking forward to playing Azir again after his 15 bugfixes tomorrow.
I just checked, it's 15.
Plus the addition of being able to use E on soldiers that are moving with Q. That's a pretty big deal, even if it's only used for escapes in lane.
|
pls stop crit discussion, riot remade crit years ago to be much more consistent than the % chances on tiems lead you to believe.
But of course the game doesnt actually explain this anywhere to newer players.
|
On October 09 2014 09:02 LaNague wrote:
pls stop crit discussion, riot remade crit years ago to be much more consistent than the % chances on tiems lead you to believe.
But of course the game doesnt actually explain this anywhere to newer players.
That doesn't mean that it isn't rng.Still is.
|
On October 09 2014 09:02 LaNague wrote:
pls stop crit discussion, riot remade crit years ago to be much more consistent than the % chances on tiems lead you to believe.
But of course the game doesnt actually explain this anywhere to newer players.
That doesn't change the situation we are talking about. Suppose that with 1% Crit you crit precisely every 100 attacks.
If you start precisely at zero you count it and get the crit you want when you want.
If you start a random place then at 1% crit the probability that you will crit once in 20 attacks actually higher than the probability that we would crit by pure RNG (because you have a 1/5 chance of hitting the "within 20 attacks" rather than 17% chance to hit them). Normalization actually makes crit runes stronger in this sense.
Even then though crit normalization is not as impactful as you give it credit for. The probability of two crits in a row is lower than it should be but that doesn't actually change how we have to play around crit. Why? Because its not necessarily the two crits in a row that we are worried about changing the flow of the game*, but the first crit, which is not effected by normalization. Who gets the first crit skews the game because there is no guarantee that anyone else will get an equalizing crit in a timely manner(or that they can even equalize)
The quintessential example of this is bottom lane. Two AD's have 1% crit. One AD crits. The other one cannot continue to engage saying "well crit is normalized so i will get an equalizing crit soon". They have to back off and concede the lane/farm.
*But anyone who has played against someone with only ghostblade can tell you that this is not so uncommon as you think
|
I think the big deal about the whole one crit thing is that losing 1 damage in a 5 auto fight likely only very rarely actually makes a difference in getting a kill or not.
But getting a crit makes a gigantic difference.
|
United States47024 Posts
It's somewhat reasonable at higher crit values where it becomes sufficiently likely for you to get an additional crit after some number of attacks, but early game crit values are so low that it has virtually no impact on early game crit.
Essentially the way crit normalization works is that it shifts the center of the "number of attacks until next crit" distribution to be maximized at 1/(crit %) attacks--where normally the distribution is monotonically decreasing (because the probability of getting a crit in the next attack is N, in the 2nd attack is (1-N)*N, 3rd is (1-N)^2*N, etc.). However, in the case where you have a really really flat distribution for this (with N = 1%), the distribution is already so flat that a change in the distribution doesn't have much impact on your actual gameplay.
On October 09 2014 09:22 Goumindong wrote:
The quintessential example of this is bottom lane. Two AD's have 1% crit. One AD crits. The other one cannot continue to engage saying "well crit is normalized so i will get an equalizing crit soon". They have to back off and concede the lane/farm.
The example you're giving isn't very good because if crit % values were high, they actually could. If you have 50% crit chance, the % chance of crit on next hit rises extremely high after 4 or 5 attacks. It's just that specifically at 1% crit chance, the normalized distribution is still comparatively flat, and even though you're almost guaranteed a crit sometime in the next X attacks, when that X is like 50-75 attacks, that's not practically useful. You can't just say "oh he got a crit now, don't worry I'll get a crit 75 attacks from now" because you'll lose the lane based on the HP differential well before you get 75 more attacks on them.
Normalization has comparatively higher relevance at higher crit percentages, it's just that early game crit values that are only possible via crit runes make for these scenarios where it doesn't have much practical relevance.
|
On October 09 2014 09:29 TheYango wrote:
It's somewhat reasonable at higher crit values where it becomes sufficiently likely for you to get an additional crit after some number of attacks, but early game crit values are so low that it has virtually no impact on early game crit.
Essentially the way crit normalization works is that it shifts the center of the "number of attacks until next crit" distribution to be maximized at 1/(crit %) attacks--where normally the distribution is monotonically decreasing (because the probability of getting a crit in the next attack is N, in the 2nd attack is (1-N)*N, 3rd is (1-N)^2*N, etc.). However, in the case where you have a really really flat distribution for this (with N = 1%), the distribution is already so flat that a change in the distribution doesn't have much impact on your actual gameplay.
I think you mean changing the distribution type. The mean "number of attacks until next crit" is always 1/crit % what we tend to consider the "center" of a distribution.
Mainly crit normalization changes it so that the mode (most common result) is 1/crit % instead of 1. Which as you mention has basically no effect at low crit %'s because two crits in a row is so unlikely anyway.
IIIRC At high crit %'s you tend to crit enough over the course of a small fight that it doesn't matter either. Edit: Though if their algorithm/distribution is memoryless it could be problematic and mean that very high crit %s either fail to crit as much as they should or crit far more
|
I am curious about this crit thing.
I'll put 1 crit red on my rune pages and get back to you guys in a week.
|
United States47024 Posts
On October 09 2014 09:38 Goumindong wrote:
IIIRC At high crit %'s you tend to crit enough over the course of a small fight that it doesn't matter either. Edit: Though if their algorithm/distribution is memoryless it could be problematic and mean that very high crit %s either fail to crit as much as they should or crit far more
IIRC this happens with DotA's PRD, but it has very little practical relevance because the non-linear stacking of crit means that it's never sensible to actually get crit values high enough for this to matter.
With LoL I'm not sure because I don't recall anyone ever disclosing the nature of LoL's PRD.
|
On October 09 2014 09:29 TheYango wrote:Show nested quote +On October 09 2014 09:22 Goumindong wrote:
The quintessential example of this is bottom lane. Two AD's have 1% crit. One AD crits. The other one cannot continue to engage saying "well crit is normalized so i will get an equalizing crit soon". They have to back off and concede the lane/farm.
The example you're giving isn't very good because if crit % values were high, they actually could. If you have 50% crit chance, the % chance of crit on next hit rises extremely high after 4 or 5 attacks.
Not if they're dead or if their tank is dead and they're now being focused while they're still trying to kill the tank.
Teamfights have hysteresis remember. The reason first crit matters a lot in the end game is because the duration with which you have to have that equalizing crit shrinks as damage increases.
|
On October 09 2014 09:02 LaNague wrote:
pls stop crit discussion, riot remade crit years ago to be much more consistent than the % chances on tiems lead you to believe.
But of course the game doesnt actually explain this anywhere to newer players.
Hidden markov chain mechanic doesn't mean crit % isn't accurate. The difference is the output is put out on an much more even basis
|
|
|
|
|
|
EPT
