HM chest running in Fronis Dungeon

Well I decided this weekend to burn off some alcohol points while working on TH/Wisdom titles so I went to Fronis Dungeon to run chests...

On Friday I didn't notice anything unusual for about an hour of running 75% gold 25% purple or tome and a 35% pick break rate randomly distributed - went through a total of 20 picks.

On Staruday it was about the same - similar drop rate & same 20 picks.

On Sunday I stopped after my 1st 8 picks. ALL 8 were purple & broken picks. This really seems like a statistical anomally. Anyone else experience this penomenon in HM chest runs repeatedly in the same Dungeon (note all different instances)?

How is it that people will notice a run of bad luck while chest running and start a thread about it here, but I've yet to see someone start a thread wondering about their luck with chests when they have a run of saved picks/gold drops?

Retention rates aren't based off a few picks, so there will be periods when one breaks several picks in a row. There will also be periods when one retains several picks in a row. It all balances out with the stated retention rate in the end.

Ghost in the machine doesn't like you

It's variance, you'll find that over the short run it can vary wildly from the percentage it quotes. Can be aggravating but as pointed out it can also work to your advantage sometimes.

why are you using the word "statistical" ? you only ran 48 picks...
Bad luck it happens,but it would probably even out if you kept running keys

I've been monitoring my NM chest running (Locked Chests in EotN) for a little while...

689 picks used, 2229 chests opened, 713 gold items (going from r4/r5 lucky/treasure (63% retain) to r5/r6 in the process (68% retain))

or 3.24 chests per lockpick and a 32% chance of a gold drop.

Has anyone got similar quantities in HM to compare? I'm more interested to see if there's a greater gold drop rate in HM more than anything else.

In many circumstances, a sample size of 30 is all you need to return a useful estimate. More is always better, of course, but not always necessary.

You cannot rule out the null that random chance caused the result with statistics. You need theory to get somewhere. Once you have a theory, you then use statistics as a tool to show that it is overwhelmingly likely that the theory is correct.

Show me several samples of this happening to different players under an identical set of specific conditions and I will be interested in what you have to say. But a single observation or series of observations in search of a theory is not news.

lol...



Welcome to Guild Wars (metaphorically speaking). Nothing in this game is guaranteed (loot-wise). If you get a good streak for retaining then you were just fortunate to get a good streak. If you get a bad streak with lots of purples, no chests or breaking all your picks then like I said, "welcome to Guild Wars". Nothing is definate or absolute.


Have fun.

I thought this string of luck with HM dungeon chest drops is the same for everyone. What, you'd think that if the runs would yield better loot in weapons more frequently that we wouldn't have such high prices for these very items that only spawns from specific HM dungeon chests?

I've run over 15,000 picks in fronis' lair (yes, 15,000) about half were in hm and half in nm. (and yes, I've maxed lucky, unlucky, treasure, and wisdom from it) Regardless of what anyone says, I've noticed that sometimes I will have runs of incredible good luck (18 gold drops in a row one time) and I will have runs of terrible bad luck. I would say more bad luck than good. I was trying to find a correlation between having favor and getting better drops, and it did seem to trend that way but I think 15000 is too small a sample size to say for sure. I did notice that after a certain amount of opening chests in the same dungeon (somewhere between 60 and 90) I would start getting almost all purples. I assumed it was some sort of anti farming code, but it usually meant I was done running for the day.

why dont you actually get a meaningful set of data and perform the appropriate statistical measurements before spouting on about how it must be an anomaly due to 8 broken lockpicks.. el oh RED ENGINE GORED ENGINE GORED ENGINE GORED ENGINE GOing el

Even with 25% of purple droprates you can end up never, ever seeing gold item from chest. Chance is only that ... chance. Nothing guarantees you that every fourth item will be purple. or every other pick will be broken.

There is this amusing story about propability: professor gave students task of producing 200 random coin flips: students could choose if they would really flip coin or whether they will fake results. Professor wuld then look over their results and figure out every person who faked results.

Why? people who were taked with comming up with random flips never produced more than 5 heads or tails in row (because it did not seem 'right' to them), people who just flipped dice on the end ended up having them (suprisingly often).

Lesson? Random is not what you think it is...

I agree and dont forget its stated a human cannot be absolutely random as were are limited and can be predicted ( we are drawn by logic ) but a computer can be random as its just numbers rather than a human thinking " i cant have 3 numbers together it doesnt look random ".

I'm going to bookmark this and quote you on the next 10 threads about Lockpicks breaking, golds not dropping and getting 10 Fungal Wallows in a row from presents.

TYVM for posting your EXPERIENCE (not just arm-waving postualtes & bashing the OP using inference based on general knowledge of statistics).

Since I still have about 5300 more to max TH & Wisdom, I didn't want to follow a "bad" path.

Just out of curiosity, did you wait any period of time before going back to that same dungeon for running (a day or two for example)?

Google "Martingale System". This foolproof way to earn big money playing Roulette is the main reason Casinos are still in business and still taking money from suckers that don't understand basic probability. Your professor was a very wise man.

Indeed, I tracked my progress in normal mode as well, running chests out of Boreal Station.

3,564 chests
1,162 gold items
32.6% of drops were golds

My original hypothesis was that I *always* got crappy (req 12 & 13) gold items on weekends, especially during double gold drop weekend events. Would chest running in the morning, when there are fewer players online change the results? So I started tracking morning results versus evenings/weekends.

The stats initially confirmed my suspicion, though as the sample grew, it started to average out. But even after almost 1,200 gold items, there are some interesting differences worth noting. (If the chance is truly random, and the sample is large enough, all should be 20%...)

Morning Results:
Req 9 - 19.2% of all gold drops from morning runs
Req 10 - 20.2%
Req 11 - 22.9%
Req 12 - 18.9%
Req 13 - 18.8%

Evening/Weekend Results:
Req 9 - 18.5% of all gold drops from evening/weekend runs
Req 10 - 20.0%
Req 11 - 18.9%
Req 12 - 20.8%
Req 13 - 21.8%

The 3% difference for req 11 and req 13 is worth highlighting, though as I said before, the numbers are slowly converging toward 20% as the sample grows.

Another concept that I was testing was "initial success rate" to see if there was any effect on the longer term success rate by looking at the the first ten attempts. I normally ran 20 to 40 chests per session. If I get 9 broken picks out of 10 attempts, should I stop for the day? The answer turned out (of course) to be no. The results of the first ten attempts had no bearing on the success of the next ten.

Plotting the "success to attempt" ratio (six unbroken picks in ten attempts) looked like a bell curve, centered around my retention percentage. Sometimes I'd retain nine of ten, sometimes two of ten. In no way did it ever correlate with my gold/purple ratio, nor the subsequent lockpick retention probability.

About the only useful fact that I gained out of the whole exercise is that my actual retention rate always hovered around 3% higher than my expected rate. When it was supposed to be 57%, I averaged 60% retention. When It got up to 68%, I actually retained 71% of my lockpicks. Now I don't know if this is unique to the chests just outside Boreal Station, or all EotN chests, or all chests. But it's interesting...!

mathscraft ftl

shit happens, run 100 picks, if there all purple and broken then report

useless topics

a null hypothesis of 35% pick retention with 8 failures from 8 trials (binomial) gives a P-value of .0517, which is not less than .05, but its close. we cannot reject the null but further investigation should continue.

i find it interesting that you stopped at 8 trials and failures, because it supports the theory that humans have an innate "it's ok to be wrong 1 in 20 times." This concept of 95% probability is "close enough" seems to be ingrained in the human brain.

You must have been running the same time as me. I went through 11 chests without breaking a pick. I almost saved that one out. Broke on the 12th chest. Cumulative probability? 0.0000096549157373046875. Not bad eh?

any numbers you posted are statistically insignificant

Just curious, is the assortment of items that drop in Fronis' any different than Boreal Station? Boreal seems a bit faster (plus the odd chance at a second chest), but I'm sure getting tired of all the same crap dropping.

all the law of large numbers says is that the distribution of means of identical random variables converges to a normal distribution as the number of samples taken goes to infinity. Saying that because of this numbers in this thread are insignificant is wrong.