• These forum archives have been curated by the community bringing two decades of Diablo 2 knowledge to new players. They are purely for reference and cannot be posted in. Use the main PureDiablo forums for all new conversation.

Save time in Arcane.. read this statistic

Shadefallen

Member
Aug 2, 2005
88
0
18
49
Denmark
Save time in Arcane.. read this statistic

So.. I have been running arcane a lot since the new patch. When you get down there there is like 25% chance for you to pick the right path to Summoner.. or is there???

I have made 50 runs now and this is where I found summoner:

Teleport path: 20 times
The path with stairs: 13 times
The path with a lot of levels and stairs: 9 times
The path with lightning sentries: 8 times

So.. fastest way to Summoner should be taking the teleport path in Arcane. I hope that you understand my description of the paths.. I am a dane and english is not my native language.. if not.. ask me :)

Oh and.. from those 50 runs I got 6 keys :)
 

Freezing Rain

Member
May 25, 2005
1,613
0
346
Funny thing for me is it's usually the path i take last, and when i try to think of what path i would take last, it's STILL the path i take last.
 

Shadefallen

Member
Aug 2, 2005
88
0
18
49
Denmark
Kireiray said:
Funny thing for me is it's usually the path i take last, and when i try to think of what path i would take last, it's STILL the path i take last.

Ah yes.. that was how I used to feel too.. try using my statistic.. should save you some time :)
 

Superhal

Member
Jul 5, 2003
2,815
0
466
Visit site
yes, if i am pressed for time, i always do the winding path first then the tp path. it also seems to me that if a char doesn't have teleport, winding path > tp path.
 

Omikron8

Member
Jun 28, 2003
8,171
0
165
That's not how statistics work

it is 25% chance to find the summoner as the # of attempts increases towards infinity

i doubt you are doing an infinite number of summoner runs

eg. in my case i always take the straight platform path and the skinny stair path and for 15 out of 20 runs he was not on either of those paths, this is less than 50%
 

dang79

Member
Feb 18, 2004
1,373
0
120
Omikron8 said:
That's not how statistics work

it is 25% chance to find the summoner as the # of attempts increases towards infinity

i doubt you are doing an infinite number of summoner runs

eg. in my case i always take the straight platform path and the skinny stair path and for 15 out of 20 runs he was not on either of those paths, this is less than 50%

I'm not sure what you're trying to say.

But the probability of finding the summoner on your first path IS 1/4 or 25% for any SINGLE attempt or trial.

As the number of attempts tends toward infinity, then you should find that the number of times you find him on your first path is tending towards 25% x # of attempts.

If he is hypothesising that it is not random, that you can label paths like "teleport path.. etc", then the number of attempts or trials he makes needs to go up if he wants to accept his hypothesis.
 

AnimeCraze

Member
Jul 13, 2005
8,513
0
120
Vancouver
That is, assuming that the probability is equally likely. Of course, we can't tell either way with just 50 runs. The best way to do it is to assume that it is equally likely. Then make a good number of runs (200?) and collect the data. Use hypothesis rejection methods see whether the original assumption is likely to be flawed, and if it is, then the probability is not equally likely. :thumbsup:
 

aicoped

Member
Jul 10, 2005
117
0
33
I agree 50 runs is not statistically significant.

I would suggest however that the odds of getting 20 runs which is 40% is something of an anomoly. I am sure there is someone that can read code that can point out that it maps something like this.

1/4 summoner north
1/4 summoner east
1/4 summoner south
1/4 summoner west

now if the map config for each of those is such that there are more teleports to the east then that may explain it.

I can think of a better example.

Nithalak in hell from the waypoint seems like he could be anywhere to next level, but there are only 3 map configurations and 2 of them are down and left.
 
Jul 29, 2003
1,185
0
346
Visit site
ive done a lot more than 50 runs and plotted not only which "kind" of path was the right one but also the orientation on the dungeon roll. there is no discernable trend.

your best bet is to take the "kind" of path which you believe you can traverse the fastest and hope the summoner is there. it takes only a second or two to figure out which path you are heading down.

the "kinds" are as follows:

'portals'
'zigzag'
'traps'
'platform'

portals is easy to recognise...it has red teleport pads.

zigzag is the one with the winding pattern.

traps is the one with spike traps and i think is the only one with lightning spires in the middle.

platform is basically 'the other one' out of the 4, and always has a medium sized open squarish platform in the beginning.

the orientation is as follows:

platform----------portal

-------waypoint-------

traps-------------zigzag


this order can appear at any degree of rotation (90, 180, 270), but the order is always the same. you will never see traps across from zigzag or platform. you will always see traps across from portal. i defy anyone to produce a picture showing otherwise on automap.

thus, there are only 4 maps possible: the map shown above, and then the same map rotated clockwise at 90 degree intervals. each 'kind' of path can appear at any compass direction: NW, NE, SE, SW. but the order is the same. thus you can conclude, from one path, where the three other paths are, without exception.

to the original poster: you deserve a prize for "most conclusions drawn from least amount of hard work and data collected." in a game where the odds of finding high runes and tyraels might are something like 1/1000000, youve drawn conclusions from a mere 50 runs and zero mpq diving. why dont you go do some hell baal runs and let us know everything about baal drops once you hit 50.

my "statistics" would lead me to believe that the bottom right, or SE, is the best guess, but i dont have enough data.
 

RTB

Member
Jun 21, 2003
5,108
2
120
Dimension of Numbers
There's some truth in this. D2 maps aren't completely random, and one path is more likely to have the Summoner at the end than others. But I don't remember the best order in which to run the paths. Bottom right first sounds right, but don't quote me on that.
 

Dawnmaster

Member
Apr 23, 2004
5,209
30
469
41
Belgium
acceleration turkey said:
the orientation is as follows:

platform----------portal

-------waypoint-------

traps-------------zigzag


this order can appear at any degree of rotation (90, 180, 270), but the order is always the same. you will never see traps across from zigzag or platform. you will always see traps across from portal. i defy anyone to produce a picture showing otherwise on automap.

thus, there are only 4 maps possible: the map shown above, and then the same map rotated clockwise at 90 degree intervals. each 'kind' of path can appear at any compass direction: NW, NE, SE, SW. but the order is the same. thus you can conclude, from one path, where the three other paths are, without exception.

This is interesting, I never realised that. But this is very interesting indeed.
Perhaps, like most other things in this game, the patterns we see are just streaks at that time. I suggest to do 50 runs per day. Note each time what rotation the sanctuary is and where you found the summoner.
If you find a high match each day, but if differs from day to day, you're just in a streak every day.
If you find the same results (more or less) everyday for a week, it is worth investigation imo.

On a personal note: these last few days I have been thinking about a way to increase summoner runs speed :rolleyes:
 

AnimeCraze

Member
Jul 13, 2005
8,513
0
120
Vancouver
Let's see, with 50 runs, supposed average = 12.5. supposed standard deviation = 3.06. The 20 times on portal is 2.45 SD off the average. This is not insignificant, since given a normal curve, we are ~99.5% certain that the true average is NOT 12.5. But still, he might just have some hot luck. :lol:

I finally got around to pull out my statistic textbook for this...... :uhhuh:

Edit: Don't compare hell Baal to this, since the supposed probability for Baal is ridiculously low for each possible item. It's not ~25%. Therefore, you need good enough data to make any conclusive arguement about the set. :lol:

Take math as major in university, it's good for you. :thumbsup:
 

SalivaMonster

Member
Feb 6, 2005
441
0
324
Hong Kong
www.xanga.com
AnimeCraze said:
Take math as major in university, it's good for you. :thumbsup:

True.

AnimeCraze said:
Let's see, with 50 runs, supposed average = 12.5. supposed standard deviation = 3.06. The 20 times on portal is 2.45 SD off the average. This is not insignificant, since given a normal curve, we are ~99.5% certain that the true average is NOT 12.5. But still, he might just have some hot luck. :lol:

You seem to assume a normal distribution while in fact this is a classic bookwork example of chi-squared test of a tetrahedral dice.

With degree of freedom 3, the 5% chi-squared number is 7.82 (you can look it up in a table, check your statistics textbook), meaning that if the observed chi-squared number is greater than 7.82, there is only a 5% probability that the deviation can be explained by chance.

value--expected--observed--(exp-obs)^2/exp
Tele---12.5-------20--------4.5
stairs--12.5-------13--------0.02
levels--12.5-------9---------0.98
sentries-12.5------8---------1.62
Total: 7.12

To make it simple, it can be concluded that there is a 95% chance the observed deviation is a result of random fluctuation.

Hope you don't have a statistics exam coming up....:lol:
 

AnimeCraze

Member
Jul 13, 2005
8,513
0
120
Vancouver
There is no reason not to assume normal. I am assuming that it is a success or a failure, whether portal path = success. p = 0.25 and q = 0.75. Now, n = 50 so Central Limit Theorem applies. Well, maybe I have not taken advanced stat yet, but I think this should still be a valid arguement.

Edit: Now, you said that the test stat value is 7.12. Doesn't it mean that there is a 90% chance that it is NOT a random error? (ie. With 90% confidence, we can reject the null hypothesis) It is above the reject region of the 90% afterall. ie. It is likely it's just not a random error.

Edit 2: I think you confused at something in logic. If it is not 5% explainable by chance, it doesn't mean that 95% likely that it is random error. It means that it is not conclusive enough to say that it is only 5% chance of random error. It might be 10%, which is what we have here.
...... Too used to finding flaws in mathematical arguement......, I need it for the Putam. :lol:

Edit 3: So used to it that I have to edit the post multiple times...... And hurray for whoever that says math is good.
 

mainaman

Member
Jun 25, 2005
2,294
0
165
virginia
ill tell just that guys statistically the u have 25% to find summoner form the first try.now as the initial post say 50 runs IS REALLY NOT ENOUGH to build statistics and draw conclusions-u need more like 1000 runs too see that ull get
~250 times u found summoner at the end af each path.
So as nobody is gonna count 1000 runs to check that out just wish yourselves good luck and hope ull find him at the end of the first path u take.