PDA

View Full Version : Here are the concrete probabilities for each Lockbox tier



Yokai
04-11-2013, 09:20 AM
TLDR: As of the date shown in the chart below, the best tier to buy appears to be T4 if you want oranges, and T3 if you want purples. This analysis uses the validated collected data from Noupoi's excellent community data collection project here: http://forums.defiance.com/showthread.php?20963-The-Lockbox-Data-Collection-Project

UPDATE: I won't bother with a second screenshot, but here are the same probabilities for a purchase of THIRTEEN T4 lockboxes. (As opposed to the metrics below, which are based on THREE T4 lockboxes.) At the current 30% discount for buying $50 USD of bits, that is the number of T4 lockboxes you can open for spending $50 cash. If you buy 13 T4s and open them all in a row, your probability of getting AT LEAST ONE orange is 95%, AT LEAST TWO oranges is 80%, and AT LEAST THREE oranges is 57%. In real world terms, that means if you spend $50 USD on a lockbox spree, you have a 20% chance of scoring AT MOST only one orange, and you have a 43% chance of scoring AT MOST only two oranges. Think about it.

POST DLC1 Update (23-Aug): I've seen a lot of threads since the release of DLC1 claiming that the drop rates for T4 are either better or worse than before. This is incorrect. Dahanese said on 12-Aug that "I have a reply from the devs! We should change that to 'will now also drop new charged weapons' as that's the change.". So early release notes for the DLC implied that drops rates for T4 would be better, but in fact that turned out to be wrong and the only difference was that the new charge weaps were added to the T4 loot tables. Similarly, adding new weapons to loot tables does not change drop probability for purples or oranges or blues, etc. That's not how loot tables work. Point of all this is that not much has changed since the last time I updated the chart below, so these probabilities are still accurate.

http://i.imgur.com/4KzrZFG.png

----- Keep reading if you care to see the details behind the TLDR -----

First, you should look at Noupoi's excellent community project here, and look at his spreadsheet. Ignore the "cost per item" data, because the percentages you see there are telling you a different measure than the actual cumulative probability of scoring oranges or purples. Just focus on the "drop rate" numbers there. http://forums.defiance.com/showthread.php?20963-The-Lockbox-Data-Collection-Project. Also, be sure to look ONLY at the "Validated" worksheet tab. The chart above is based only on the validated data collected so far.

Now, with those baseline drop rates, we're going to ask the question: "What is my cumulative probability of getting AT LEAST 1 of each color if I spend 192 keys?" For example, you can spend those 192 keys on 24 T2 lockboxes and open all of them. What is your chance of scoring AT LEAST 1 orange from those 48 "item rolls"? What is your chance of scoring AT LEAST 2 or AT LEAST 3 purples from those 48 item rolls? What if you decide to buy eight T3 lockboxes instead with those 192 keys? What is your chance of AT LEAST 1 orange from those 24 item rolls (with slightly higher drop rates than the T2 boxes)? And so on.

This type of probability question is identical to questions like "If I flip a coin 20 times, what is the probability of getting AT LEAST 8 "heads" as a result? Or of getting EXACTLY 12 tails as a result?" And so on. Or with dice rolls; "If I roll this 6-sided die 10 times, what is the probability of getting at least two 6s?" And so on.

This type of probability is easily calculated using a notion called "cumulative probability", which is based on a "binomial distribution". Look it up on Wikipedia. If that makes your head hurt, know that there's an easier way! You can simply plug three simple numbers into an online binomial distribution calculator such as at http://stattrek.com/online-calculator/binomial.aspx . Such calculators will typically require inputs called "trials" and "successes". The trial is the number of flips or rolls. In our case of using 192 keys, the number of trials is 48 for T2, 24 for T3, and 12 for T4. The success probability for each trial is, in our case, the "drop rate" for a given color from each type of lock box. The "Number of successes" is, in our case, the number of oranges or purples you're looking for.

So, for example, if you want to know the cumulative probability for AT LEAST 1 orange from spending 192 keys on eight T3 lockboxes, the three numbers you'd plug into the calculator are:

"Probability of success on a single trial": .05
"Number of trials": 24
"Number of successes": 1

The result would be 71%, which is your odds of actually scoring at least 1 orange from those 192 keys spent this way.

------ A Note About Sample Size and Accuracy -------

Edit: I'm adding this section after seeing several spurious comments challenging accuracy or sampling techniques, hopefully to stave off further well-intentioned but incorrect arguments.

In this type of problem, it is a well-established, universal technique to use sampling techniques to make reasonable predictions about a total population. This revolves around concepts of normal distribution, confidence levels, confidence intervals, and standard error. Without going into all the statistical jargon, the short TLDR is very very simple:

Any sample of 20 data points or more can yield strongly significant results. Any sample of more than 377 data points is usually waaaaaayyyyyy more than necessary to yield high confidence results. Specifically:

A sample size of 30 data points yields results that we can be 95% confident will match the actual population results all but 17.8% of the time (if we could measure every single member of the population accurately) if we tested 100 such different samples.

A sample size of 100 data points yields results that we can be 95% confident will match the actual population results all but 9.8% of the time (if we could measure every single member of the population accurately) if we tested 100 such different samples.

A sample size of 377 data points yields results that we can be 95% confident will match the actual population results all but 5% of the time (if we could measure every single member of the population accurately) if we tested 100 such different samples.

And so on.

Noupoi's sample size is now over 2000 data points. That equates to 95% confidence that the collected data already matches the actual population (aka the "real" drop rates put in by Trion) all but 2% of the time if we were to wipe out Noupoi's results right now and start over with 2000+ data points. And then wipe and re-do another 100 times.

In other words, the observed results we have RIGHT NOW are "close enough for rock and roll". You can trust them. At least until Trion makes a stealth change to the actual drop rates.

Finally, because the base drop rates we're working with have such a VERY HIGH confidence to be accurate, the calculated cumulative probabilities are effectively bullet proof. If you argue otherwise, you're simply wasting everyone's time. These numbers will remain bullet proof until the day that Trion decides to tweak the underlying drop rates (whether they bother to tell us that fact or not.)

Now for a final caveat: The above rules of thumb about sample size are true only if you have truly random sampling in your sampling methodology. In the case of Noupoi's collection technique, it is not truly random. So yes, the confidence level cannot truly claim to be a 95% confidence level within 2% error of the actual population. But regardless, the sheer number of data point still makes the confidence very high overall.

Exes
04-11-2013, 09:40 AM
The sample size is way too tiny.

Kurze
04-11-2013, 09:47 AM
You are working with invalid data

Noupoi's project, while commendable, is utterly useless because it doesnt factor in the amount of sample size

for example, if 800 people uploads their results with tier 3 boxes but just 22 do likewise with tier 4, the results for tier 4 are annecdotical at best and we shouldnt base any calculations on them.

also the data is skewed in many other ways.

for example, people are more likely to stop and upload their results when they get something good, like a purple or orange, but they wont bother if they got nothing but green/blue rubbish in a few boxes in a row.

Xynnvincible
04-11-2013, 10:00 AM
You are working with invalid data

Noupoi's project, while commendable, is utterly useless because it doesnt factor in the amount of sample size

for example, if 800 people uploads their results with tier 3 boxes but just 22 do likewise with tier 4, the results for tier 4 are annecdotical at best and we shouldnt base any calculations on them.

also the data is skewed in many other ways.

for example, people are more likely to stop and upload their results when they get something good, like a purple or orange, but they wont bother if they got nothing but green/blue rubbish in a few boxes in a row.

The data is not invalid. You are questioning whether the sample is a representative sample (large enough sample size, and whether it's skewed by people posting only good results. You assume that people only post when they get good results, which may or may not be the case.

fishboy11
04-11-2013, 10:02 AM
too bad it doesnt seem to work like the stats say, i have bought soooooo many lockboxes of each kind and still have not seen an orange and only 4 purples

Kerry
04-11-2013, 10:07 AM
13 Oranges from Tier 2 lock boxes to date.

Receiving at LEAST one orange item a day, including shields, grenades, and guns.

Nice post though and excellent effort, regardless of the validity.

Zhaocore
04-11-2013, 10:08 AM
My very first t3 lockbox was an orange. =/

EmoPyroNecrosis
04-11-2013, 10:12 AM
If the data is accurate, then it's best to get a T4, why? because I can acquire a grand total of 75 keycodes in less then 2 hours of gameplay due to the massive number of events Defiance spawns, seriously, just stop when someone screams help, score 3 keycodes 5 from a major arkfall.. not to mention the money from killing things, and loots. less then 2 hours another T4, also drop rates change when in a group, I've been in a 4 person group, and will see purples dropped from Hulkers 9 times out of 10, friend of mine in group with me got an Orange Guided Rocket Launcher from a HellBug Matron during Mother of all Hellbugs. So push comes to shove group up, only downfall is baddies are a tad more healthy when in a group.

tabularasa
04-11-2013, 10:13 AM
Seems alright, have bought 12 or so T2 boxes and have gotten 3 orange. Hardly hear of other people getting anything good from t4 boxes.

The problem here is the "higher tier MUST mean its better" thinking. This is the rule among every single scenario in the real world. Higher tiers gain better benefits. This used to be the case in the Alpha. Then from complaints and a move to a new system of keycodes, it was changed so that each box gives a different chance. Its counterproductive thinking that makes most people believe that spending 64 keys on a T4 box must equal something really good. But from what I have seen in beta and alpha, T2 is the way to go.

The problem is that this falls on deaf ears as people refuse to believe that a Higher tier is basically a waste of money. Its like saying that Lamborghini is not faster than that Yugo. Its a way of thinking that has many people questioning the fairness.

Clanmates are always surprised when I get an Orange from a T2, when they had just spent 64 keycodes on a T4 and got a few blues and greens.

Common sense is the least common sense.

EmoPyroNecrosis
04-11-2013, 10:17 AM
Seems alright, have bought 12 or so T2 boxes and have gotten 3 orange. Hardly hear of other people getting anything good from t4 boxes.

The problem here is the "higher tier MUST mean its better" thinking. This is the rule among every single scenario in the real world. Higher tiers gain better benefits. This used to be the case in the Alpha. Then from complaints and a move to a new system of keycodes, it was changed so that each box gives a different chance. Its counterproductive thinking that makes most people believe that spending 64 keys on a T4 box must equal something really good. But from what I have seen in beta and alpha, T2 is the way to go.

The problem is that this falls on deaf ears as people refuse to believe that a Higher tier is basically a waste of money. Its like saying that Lamborghini is not faster than that Yugo. Its a way of thinking that has many people questioning the fairness.

Clanmates are always surprised when I get an Orange from a T2, when they had just spent 64 keycodes on a T4 and got a few blues and greens.

Common sense is the least common sense.

Not to rain on your parade but.... every time I get a T4 I get at least one-3 purples, on the rare occasion I get one purple I also get one orange. >_> Although I rarilly care about rarity, so far all the "Legendary" weapons I've used were rather ****ty, had a shotgun that shot one pellet, had a magnam with 3 shots, best one was a 3 shot burst semi auto sniper, issue is sniper reserve ammo is low, ended up selling it due to that. then their was the crappy blast shield... day I find a Legendary Health restoration shield, I'll shat my pants. Regardless, I find purples are more useful, which is sad.... I still use my EGO Rating 106 Purple Nano Fragger @.@ and I'm almost Ego Rating 700

jetah
04-11-2013, 10:19 AM
If the data is accurate, then it's best to get a T4, why? because I can acquire a grand total of 75 keycodes in less then 2 hours of gameplay due to the massive number of events Defiance spawns, seriously, just stop when someone screams help, score 3 keycodes 5 from a major arkfall.. not to mention the money from killing things, and loots. less then 2 hours another T4, also drop rates change when in a group, I've been in a 4 person group, and will see purples dropped from Hulkers 9 times out of 10, friend of mine in group with me got an Orange Guided Rocket Launcher from a HellBug Matron during Mother of all Hellbugs. So push comes to shove group up, only downfall is baddies are a tad more healthy when in a group.

The OP is stating that if you purchase 1 T4 lockbox you have gain 4 items. however you could have purchased 9 T2 lockboxes with a total of 18 items. you'll have a better chance of getting orange/purple with 18 chances then with 4.

I'll say I'm entering my T4 boxes in the spreadsheet. Hoping that more people do the same so the T4 can get higher entries.

another thing to note is we dont know if the drop rates are different between the platforms (PC, Xbox, PS3).

Hope499
04-11-2013, 10:26 AM
I have bought like 10 T4 boxes with nothing but the odd crappy purple and blue and greens. I will be switching to T2 boxes, Thanks :)

Arch Fiend Folio
04-11-2013, 10:34 AM
The sample size is way too tiny.

I agree here... I have opened 10 of them and gotten 5 Orange ... so... either I am really lucky or your math is off... there are others in my guild that have similar findings... and others have yet to see one


granted it took me till my 4th box to get one... but on a good streak since

LoVato
04-11-2013, 10:35 AM
some guys or gals have all the luck! i have not scored a legendary weapon (orange) with any tier boxes since i started playing alpha/beta2-3. i guesstimate that my playing time is over 100 hours since inception. i grabbed the loot booster just to see if it makes any difference, right, 90% of the time it's either white shields or white nades. WTF!? so is it even worth wasting cash on boosters (with the exception of XP booster; that helps!)?

nevertheless, interesting post. you also typed 4,765 characters with spaces.

DeMoted
04-11-2013, 10:37 AM
Random sorry orange from T4.

EmoPyroNecrosis
04-11-2013, 10:37 AM
The OP is stating that if you purchase 1 T4 lockbox you have gain 4 items. however you could have purchased 9 T2 lockboxes with a total of 18 items. you'll have a better chance of getting orange/purple with 18 chances then with 4.

I'll say I'm entering my T4 boxes in the spreadsheet. Hoping that more people do the same so the T4 can get higher entries.

another thing to note is we dont know if the drop rates are different between the platforms (PC, Xbox, PS3).

Here's the issue, anytime I get T2's I get 2 greens, I never even get a Blue.. T3 I got an orange from once. and most of my Keycodes are spent on Syn Mod Caches. and like I said I fill my Keycodes every hour and a half, and I am an MMO Vet, so I don't mind Grinding, I find it more profitable anyways. :D

fishboy11
04-11-2013, 10:39 AM
I agree here... I have opened 10 of them and gotten 5 Orange ... so... either I am really lucky or your math is off... there are others in my guild that have similar findings... and others have yet to see one


granted it took me till my 4th box to get one... but on a good streak since

you are just really lucky, because the majority of people have not even seen an orange yet, i have seen 1 purple from a tier 4 thats it, the rest is greens and blues.

also did you buy them with bits or keycodes?

EmoPyroNecrosis
04-11-2013, 10:40 AM
some guys or gals have all the luck! i have not scored a legendary weapon (orange) with any tier boxes since i started playing alpha/beta2-3. i guesstimate that my playing time is over 100 hours since inception. i grabbed the loot booster just to see if it makes any difference, right, 90% of the time it's either white shields or white nades. WTF!? so is it even worth wasting cash on boosters (with the exception of XP booster; that helps!)?

nevertheless, interesting post. you also typed 4,765 characters with spaces.

lol.. I have yet to buy anything for the game, note. Power Rating can not exceed Ego Rating, Power Rating increases every 100 Power Experience.. I use Pursuits to level up. I think I normally leveled up once, otherwise I finish a Pursuit, my power rating goes up, then my Ego goes from 2 bars full, to zero due to a level up XD

Radzwe
04-11-2013, 10:44 AM
The math is valid. The starting percentages may not be exact, but they are within a reasonable range. He processed the starting percentages correctly. For fun, someone could vary the starting values and compare.

I appreciate OP's math.

DeMoted
04-11-2013, 10:55 AM
This is about as concrete at flubber.

Yokai
04-11-2013, 10:58 AM
The sample size in Noupoi's collection is large enough to be fairly close to the mark. Of course, that's assuming people are accurately reporting results and not inputting falsified results. Here's a bit of statistical theory most people aren't familiar with. Assuming you have a completely random, non-biased sampling technique, you need a sample population of only 30 -- yes, only 30 -- data points to yield statistically significant results with a reasonable degree of certainty. Of course, the non-biased and completely random conditions are hard to achieve in real-world sampling efforts (such as Noupoi's effort), so larger sample populations can help to improve certainty.

Point being, Noupoi's results include thousands of data points now, so we're well beyond that benchmark of 30. Are the numbers perfect? No. But there's enough data to guess at the most likely drop rates for each tier with a large degree of certainty. Even if the actual drop rate is off by one percentage point compared to Noupoi's results at this point in time, such a difference will not affect the fundamental truth that right now, T2 is your best bang for the buck.

Now, it should be understood by all reading this that Trion can of course go change the base drop rates at any time in a server side patch, without ever saying a word. What's true right now might not be true next week. I think the community would be well served to periodically wipe the collection data and start over to do a regular "pulse check" on the observed drop rates.

EmoPyroNecrosis
04-11-2013, 11:06 AM
only issue with the OP, is he's using an experiment without a controlled group. Nor did the experiment use pre-set variables in the experimental group. for the whole thing to be 100% valid as a Hypothesis Experiment, He needs a control group, and an experimental, with pre-set variables within the experimental group, control group purchases ONE type, while experimental CYCLE purchases lock boxes. This would show the best result. Another words, control and experimental both consist of 3 people. Each person catalogs results of 15 purchases, this would mean the control group buys 15, T2, T3, and T4 Lock Boxes. but never buys any different. IE Tom, Susan, and Aoi are control, Tom buys 15 T2's Susan buy's 15 T3's and Aoi buy's 15 T4's. Then the experimental cycles. George, Nancy and Conner of the experimental group buy 5 of each in a preset patern so nether buy the same Tier at the same time ever, so Nancy gets a T2, then T3, then T4, and repeats, George starts at T3, then gets a T4, and finally a T2 and repeats like Nancy. and Conner starts at T4, then gets a T2, T3, and like Nancy and George, Conner repeats, if someone wishes to test if different consoles have altered drop rates the test can be repeated on each console. this would show the best drop rate percentages. :D Best part, it could have a Pie Graph made, and boy to I love PIE 8D

Yokai
04-11-2013, 11:12 AM
only issue with the OP, is he's using an experiment without a controlled group. Nor did the experiment use pre-set variables in the experimental group. for the whole thing to be 100% valid as a Hypothesis Experiment, He needs a control group, and an experimental, with pre-set variables within the experimental group, control group purchases ONE type, while experimental CYCLE purchases lock boxes. This would show the best result. Another words, control and experimental both consist of 3 people. Each person catalogs results of 15 purchases, this would mean the control group buys 15, T2, T3, and T4 Lock Boxes. but never buys any different. IE Tom, Susan, and Aoi are control, Tom buys 15 T2's Susan buy's 15 T3's and Aoi buy's 15 T4's. Then the experimental cycles. George, Nancy and Conner of the experimental group buy 5 of each in a preset patern so nether buy the same Tier at the same time ever, so Nancy gets a T2, then T3, then T4, and repeats, George starts at T3, then gets a T4, and finally a T2 and repeats like Nancy. and Conner starts at T4, then gets a T2, T3, and like Nancy and George, Conner repeats, if someone wishes to test if different consoles have altered drop rates the test can be repeated on each console. this would show the best drop rate percentages. :D Best part, it could have a Pie Graph made, and boy to I love PIE 8D

You are so far off base it's not funny. Seriously, I know you probably mean well, but you are simply, fundamentally wrong. Look, I've worked with probability issues on and off for 30 years. Literally -work- with questions like these, but fundamentally harder. Currently, I work with pricing science software. You can very accurately make high confidence predictions and answer tough analytic questions with some relatively small sets of data. You just need to apply the right theory to the problem and know how to construct and interpret the answers.

This binomial stuff is probability 101 stuff you learn in the first week of any decent college level statistics course or market research course, etc.

Yokai
04-11-2013, 05:16 PM
Adjusted the computed probabilities tables to reflect a pure apples-to-apples comparison of three T4 boxes, eight T3 boxes, or 24 T2 boxes. These numbers all factor evenly into the lowest common denominator of 192 keys. The probabilities yielded from these specific values of 12 T4 trials versus 24 T3 trials versus 48 T2 trials give us accurate probabilities that are true regardless of how many or how few boxes we actually purchase over time.

Thanks to [b]Stickasylym[\b] for dogging me to base the numbers off this 192-key baseline. :)

loops73
04-11-2013, 05:24 PM
I have opened dozens of lockboxes of all tiers and have never got an orange drop. my only orange ever was at an arkfall,and it was a crappy pistol.

Ikagawa
04-11-2013, 05:25 PM
Probability is screwing me. I don't get anything useful! *sob* ;)

Xionsmith
04-11-2013, 06:51 PM
I get more greens out of T4 boxes then I do out of T2 boxes. Something isn't right about that. T4 should have 0% chance to get a green.

JonDav
04-11-2013, 07:10 PM
I woudn't mind the warped dropped rates if progression was actually adaquate. The fact that orange weapons are so mediocre in general makes this glaring flaw even worse. I got an orange rifle that was worse than a white one I got 100 ego levels earlier. Same with a orange shotgun I got. Being that these weapons are this poor, they should have a decent drop rate so people can at least think they're progressing.....

Ikagawa
04-11-2013, 08:13 PM
I woudn't mind the warped dropped rates if progression was actually adaquate. The fact that orange weapons are so mediocre in general makes this glaring flaw even worse. I got an orange rifle that was worse than a white one I got 100 ego levels earlier. Same with a orange shotgun I got. Being that these weapons are this poor, they should have a decent drop rate so people can at least think they're progressing.....

Look at this guy, not understanding the weapon system.

Archetype
04-12-2013, 07:26 AM
Haha, what a great thread. I am actually taking Probability and Statistics in University right now and just finished a test a few weeks about about Binomial Distributions. It is absolutely uncanny to see it in real practice! I absolutely love it. Ignoring that, great work on the data analysis.

Dunnomatic
04-12-2013, 08:18 AM
Been doing this all along. Not as good luck as the guy with 5 (WTF!) but got an orange and tons of blues and purples. The math makes sense. The individual percentage per box is quite a bit higher but the economics of it make tier 2 the best option, you simply as he stated get significantly more cjances for your random drop. Each box has a lower chance then a 3 or 4 but when you can a ton more it makes sense to increase how many random chances you get.

Yokai
04-12-2013, 09:47 AM
Updated the OP with a better, cleaner TLDR summary, including a screenshot from the spreadsheet I use to calculate the AT LEAST chances per 192 keys spent. I will be updating this screenshot every few days (give or take) based on the latest collection data.

Also note that I'm now using actual collected percentages to 2 decimal places, directly from Noupoi's spreadsheet at the time that I update my own spreadsheet. Previously, I was rounding up to 0 decimal places to keep the narrative simple. Now that I'm just plugging numbers into a spreadsheet and taking a screenshot, it's easier to work with more exact percentages.

Finally, note that in this update, T3 has pulled ahead as the best teir to purchase if your goal is scoring purples and oranges.

tyguild
04-12-2013, 09:53 AM
Tiny sample size and there have been official posts in regards to T3 orange drops - which report significantly less than your post.

Yokai
04-12-2013, 09:56 AM
Tiny sample size and there have been official posts in regards to T3 orange drops - which report significantly less than your post.

Read below the TLDR for a comment specifically about sample size. Also, quote your sources. Also, be sure you understand cumulative probability before making a statement like "report significantly less than your post".

l DefaulT l
04-12-2013, 10:03 AM
Just got 2 oranges in a Tier 4. Finally gave in after my tier 3 strategy and only being able to get one out of 25ish boxes. Not saying its worth it buut makes me double think it when i got 2 on my first buy since. Does the loot boost have to do with more rare items or just more in general...that could be a factor too i guess.

Kuldred
04-12-2013, 10:13 AM
The data is not invalid. You are questioning whether the sample is a representative sample (large enough sample size, and whether it's skewed by people posting only good results. You assume that people only post when they get good results, which may or may not be the case.

Thank you for clearly explaining why the data is in fact, invalid. Because the data was collected through voluntary survey submissions, the data cannot then be verifed. You have to take thier word for it. If data cannot be verified, it is anecdotal at best.

Any college level Statistics professor will confirm that data collected through survey cannot result in actionable conclusions. It provides a 'Hmm, that's interesting' set of information, nothing more.

As for sample size, I believe the OP stated that they have over 2000 submissions? Assuming uniform representation for all 4 tiers, then the sample size is adequate.

Kuldred
04-12-2013, 10:17 AM
The sample size in Noupoi's collection is large enough to be fairly close to the mark. Of course, that's assuming people are accurately reporting results and not inputting falsified results. Here's a bit of statistical theory most people aren't familiar with. Assuming you have a completely random, non-biased sampling technique, you need a sample population of only 30 -- yes, only 30 -- data points to yield statistically significant results with a reasonable degree of certainty. Of course, the non-biased and completely random conditions are hard to achieve in real-world sampling efforts (such as Noupoi's effort), so larger sample populations can help to improve certainty.

Point being, Noupoi's results include thousands of data points now, so we're well beyond that benchmark of 30. Are the numbers perfect? No. But there's enough data to guess at the most likely drop rates for each tier with a large degree of certainty. Even if the actual drop rate is off by one percentage point compared to Noupoi's results at this point in time, such a difference will not affect the fundamental truth that right now, T2 is your best bang for the buck.

Now, it should be understood by all reading this that Trion can of course go change the base drop rates at any time in a server side patch, without ever saying a word. What's true right now might not be true next week. I think the community would be well served to periodically wipe the collection data and start over to do a regular "pulse check" on the observed drop rates.



"Guess at the most likely" being the best interpretation of this data set.

Yokai
04-12-2013, 11:22 AM
Any college level Statistics professor will confirm that data collected through survey cannot result in actionable conclusions. It provides a 'Hmm, that's interesting' set of information, nothing more.


If this were a true fact, then not much would get done in the world at all. Many scientific, medical, economic, political, sociological endeavors and decisions are in fact made on "data collected through survey".

You should have stayed focused on the only -real- weakness of the current data set, which is the non-random nature of it's collection. Still, that fact alone doesn't guarantee a large margin of error.

Regardless of nits like this, it's all we have. If Trion won't reveal the actual drop rates themselves, our only recourse is voluntary submission of observed results. If the majority of the responders are -honest-, then we can in fact rely largely on the results.

And here's another aspect to consider: sample bias is a far larger problem when surveying opinions and other very subjective data. Sample bias is a much smaller issue when surveying factual numeric datapoints as simple as "when you opened the box, what number of these 4 colors did you receive?" The only potential for bias here is:

Lulzers trying to grief the results by posting stuff like "4 orange items from one T4 box"

People clicking the wrong dot, as in clicking 2 when they meant to click 1.

IMO those are both fairly uncommon scenarios, so I tend to trust the collected data more than I distrust it.

Hope499
04-12-2013, 11:25 AM
GOt my first legendary last night. Been buying T4 the entire time since release. bought 5 T2's for the first time and got it. I shall be sticking to T2 from now on.

Yokai
04-13-2013, 07:00 AM
Saturday morning update (13-Apr). The lockbox collection data team has been doing some cleanup of the submitted results to more carefully weed out possible troll submissions. Annnnnddddd, it now looks like T4 has indeed risen to the top of the pile in terms of best cumulative probability for oranges. However, T3 still seems to be best for purple drops. See the OP on page 1 of this thread for the latest chart.

Setari
04-13-2013, 08:23 PM
Guys, of course with larger sample sizes we are going to have a more accurate reading. This is a very very rough estimate at best. This is common sense. If you want to contribute to his findings, then help him, don't bash it.

Kronn
04-14-2013, 03:02 PM
I tried this and every other combo and I have not received 1 legendary in the game. I have done countless co-op missions, arkfalls, side missions and spent over 500 keys for lock boxes and have received NOTHING. I do have good luck with greens, blues, and a rare purple. People saying that have received an orange a day are either liars (probable) or extremely lucky.

Shadow Viper
04-14-2013, 03:43 PM
Might want to look up the definition of concrete before using it ;-)

Yokai
04-15-2013, 07:16 AM
Chart in the OP has been updated. The trend has not changed after two more days of data collection.

Yokai
04-18-2013, 05:50 AM
Chart in the OP has been updated again. No change in trend since the last update. T4 is still best for oranges, and T3 is still best for purples.

Lyokira
04-18-2013, 05:58 AM
I have opened dozens of lockboxes of all tiers and have never got an orange drop. my only orange ever was at an arkfall,and it was a crappy pistol.

You're lucky then, you're the first person I've heard of getting a legendary from an arkfall. :O

Yokai
04-18-2013, 07:27 AM
You're lucky then, you're the first person I've heard of getting a legendary from an arkfall. :O

Never seen an orange from an arkfall yet, but I've gotten several purples. From what I've personally seen, it seems like if you're in the top 10 at the end of the arkfall, you have a chance of getting a purple. Or maybe it's the top 5? Or maybe it's totally random? Not enough data points to be sure.

Slay
05-01-2013, 12:09 PM
Oh yeah, forgot to mention I put 10 oranges on T4 on accident. Data's probably wrong for T4 lol.

Yogizilla
05-16-2013, 02:29 PM
The data sampling may not be massive but this chart simply reinforces that Tier 2 and 3 are better deals. Trion should change it so that the minimal rarity is bumped up at each tier; otherwise, I just cant justify spending 3x more on Tier 4 for a marginally better chance of a decent dice roll. lol

There is other data out there and plenty of personal trials to support the findings in this community project, too. Hopefully the elitist grinders will hush up now. I mean, if you wanna speed run on arkfalls and what-not all day to farm keycodes, go for it, but to me Tier 4 seems like a waste of resources!

Personally, I have had better luck with Tier 2 and 3, and its more chances to get good gear. I rather get more for less because at least then you can max out weapon classes, break down to resources, or sell/trade what you don't want. Not to mention orange items can suck sometimes. I have gotten some sweet purples and even blues that pwn.

Well, to each their own.. Thanks for sharing this. I sent to my clan's mailing list and will post on our forums too. I think this will put many at ease. 8)

Yogizilla
05-16-2013, 02:32 PM
Never seen an orange from an arkfall yet, but I've gotten several purples. From what I've personally seen, it seems like if you're in the top 10 at the end of the arkfall, you have a chance of getting a purple. Or maybe it's the top 5? Or maybe it's totally random? Not enough data points to be sure.

I have a clanmate that spends almost all his time doing arkfall and hulker speed runs. At the end of the day, he usually has at least five legendaries just from drops. He does speed runs so he does not sit around for the full battles. It works for him.. He is always trying to get us to join him on the grinding. =oP

Cynical Jester
05-20-2013, 07:47 AM
Great information!

I think some people that are not receiving oranges are looking at it wrong and I also wanted to add a couple thoughts... Any of this could be wrong so feel free to correct it / ignore it.

I believe people are thinking about themselves individually to much when they should be thinking about it more like a casino. Someone next to you could hit two jackpots in a row (Very unlikely) and you could hit 0 your entire life. Really it depends on how much you play and the machine.

All machines are fixed, the question no one will ever know other than the owner of the machine is how fixed. Nothing in the digital world is really random. Lets move away from the mindset of a big casino real quick that may actually own their own games and go more with the standard of leasing machines (Especially popular amongst states that allow those "sweepstakes" games).

A machine is guaranteed to make (they profit, you do not) a certain amount of money over a certain amount of time. The people that lease the machines set the percentages. If a machine ends up paying out way to much it's considered broken and needs to be adjusted/fixed. The people that lease the machines to the places gain a certain amount of profit of whatever goes into the machine, no matter what the places may have had to pay out. The simple side: Not everyone can be a winner.

A question I would have would be this:

Is the loot determined when you buy the box or open the box? is each piece rolled individually or is it all selected at the same time?

Then lets talk about "weights". A good example is how Arkfalls are known to give shields and grenades more often than not. Those two items could be "heavier" than the rest of the loot table increasing the likely hood of receiving those two items. For example (and I'm not sure if this is true but a theory): If you buy a LockBox from the bit store the chance to receive an orange item may be higher than if you bought it with regular resources due to the "weight" of the oranges in the bit store being increased.

An example in C# from, http://programmers.stackexchange.com/questions/150616/return-random-list-item-by-its-weight:

Here's an algorithm (in C#) that can select random weighted element from any sequence, only iterating through it once:


public static T Random<T>(this IEnumerable<T> enumerable, Func<T, int> weightFunc)
{
int totalWeight = 0; // this stores sum of weights of all elements before current
T selected = default(T); // currently selected element
foreach (var data in enumerable)
{
int weight = weightFunc(data); // weight of current element
int r = Random.Next(totalWeight + weight); // random value
if (r >= totalWeight) // probability of this is weight/(totalWeight+weight)
selected = data; // it is the probability of discarding last selected element and selecting current one instead
totalWeight += weight; // increase weight sum
}

return selected; // when iterations end, selected is some element of sequence.
}

I honestly believe to get the true percentages (other than looking at TRIONs' data or an addon that everyone could use that tracked the data automatically) would to approximately know the entire population of the game and then assign a certain amount of people to one person using the OPs method.

Those selected people that counts for everyone else would then have to start a new game and track every LootBox precisely every time they opened one. Over time the data would become more accurate. But any changes to the weights could throw off that accuracy quite a bit if it wasn't re-started every so often (Especially depending on how much time had gone by to set those percentages in stone)....

Whether or not t3 is better or not is relative to the player.. Do you want a better chance at the orange or do you want a better chance at just a generally decent gun of any color. That's to say t4s aren't weighted to have more 'popular' guns / stats.. No one knows really... Would love to know their loot break down.

TLDR; You are either an unlucky soul like me and the world hates you... If/When you do get an orange it's going to be something completely useless you didn't want anyway. Like a rebound lobber or the infector that just slows people...

or

You're just one of the lucky ones / gamble a lot.

Really I just PvP to much.. Takes me 64 games to get one lockbox... You barely get any xp for it.. ahh well such is the life of PvPer.. Still rockin with my white and green weapons! (All the blues / purples I seem to get have bad stats or just the wrong gun).

Edit: I also wonder if the weights are not increased as your ego goes up or vice versa.

Yokai
05-28-2013, 06:30 AM
Chart in post #1 updated to reflect the latest probabilities. It's been 6 weeks since my last update because the collection data looked pretty stable back then in mid April. 6 weeks later, the validated drop rates are stilll essentially the same as back then. Nearly all of the reported drop rates seem to be moving around within a 0.30 to 1 percent variance at most.

By now, with all the collected data since launch, these probabilities are definitely rock solid. At least until Trion decides to tweak the lockbox drop rates. Currently there is no indication that they've done so yet.

Keep contributing to the collection project!

Chidy1776
06-09-2013, 04:12 PM
Thank you so much for this thread! Even if the probobilities arent 100% accurate its nice to have at peast some kind of statistic.

DIRK DIGGLER69
06-09-2013, 04:48 PM
Lol I've now gone from 2900-5000 EGO without getting an orange. I only open tier 4. Maybe I just have horrible luck.

c4l1d3n
06-09-2013, 04:51 PM
Inaccurate OP...

Wtflag
06-09-2013, 05:18 PM
Question:

How did you get 54% chance to get atleast 1 orange from three T4 boxes when their base chance is 6.2%.

P of NOT getting a single orange is 93.8%^3 = 82.5%

Therefore to get atleast one (or more) oranges = 100% - 82.5% = 17.5%

Qbit
06-09-2013, 05:40 PM
Question:

How did you get 54% chance to get atleast 1 orange from three T4 boxes when their base chance is 6.2%.

P of NOT getting a single orange is 93.8%^3 = 82.5%

Therefore to get atleast one (or more) oranges = 100% - 82.5% = 17.5%

Because it is for 3 T4 lockboxes (192 key codes) and there are 4 items per box. 1-93.8%^12=54%

Wtflag
06-09-2013, 05:57 PM
Because it is for 3 T4 lockboxes (192 key codes) and there are 4 items per box. 1-93.8%^12=54%

Wait.

6.2% is the cumulative odds per box or just 6.2% per roll (of 4)?

That is very important because T3 and below are completely worthless if it is the later. (Which I don't think it is).

Yokai
06-09-2013, 08:01 PM
Inaccurate OP...

Wow! That is the best rebuttal ever! So informative and full of erudite reasoning! Awesome! Thumbs up!

wick220
06-10-2013, 10:02 AM
13 Oranges from Tier 2 lock boxes to date.

Receiving at LEAST one orange item a day, including shields, grenades, and guns.

Nice post though and excellent effort, regardless of the validity.

Sweet Jesus....can I be you for a day? Almost 1750 ego and not a single orange from either a lockbox or a drop. Hell, I wouldn't know they existed if not for the forums. I only JUST got my first synergy mod...

Qbit
06-10-2013, 10:56 PM
Wait.

6.2% is the cumulative odds per box or just 6.2% per roll (of 4)?

That is very important because T3 and below are completely worthless if it is the later. (Which I don't think it is).

Per roll is how I read it.

c4l1d3n
06-10-2013, 11:14 PM
Wow! That is the best rebuttal ever! So informative and full of erudite reasoning! Awesome! Thumbs up!


http://i1342.photobucket.com/albums/o763/5IVE5/Gifs/8104c6cd_zps17518f24.gif

Yokai
07-03-2013, 10:47 AM
It's been 5 weeks since the last update. Drop rates for T2 and T3 lockboxes have remained virtually unchanged, but the validated data collection for T4 shows that purple and orange drop rates have both fallen by more than half a percent since the late May number crunching.

Check the OP in this thread for the latest probabilities, and also a new set of information right above the chart that explains the effective real world cost per orange. (In USD, anyway. ^.^)

Shogo_Yahagi
07-03-2013, 01:06 PM
A lot of bad data is still bad data.

c4l1d3n
07-03-2013, 01:08 PM
A lot of bad data is still bad data.

So much truth it hurts.

Mowazzy
07-03-2013, 01:18 PM
The sample size is way too tiny.

I agree with this. Thank you OP for your efforts. Unfortunately for me your statistics do not hold true for my experience with the boxes.

Indra Echo
07-03-2013, 01:28 PM
I agree with this. Thank you OP for your efforts. Unfortunately for me your statistics do not hold true for my experience with the boxes.


Here's the thing. Buy lottery tickets. It makes no difference how many you buy. Your odds do not fundamentally increase no matter how many you get, even if you buy a million. Each ticket still has the same odds of winning. That means your odds do not get better the more you buy.

Your neighbor who bought one ticket has virtually the same odds of winning as each of your individual tickets has.

It's all anecdotal evidence anyway as far as these lockboxes are concerned. There's a probability of error and other unknown factors (how are they programmed to randomize the weapons and what is their assigned degree of rarity).

For instance, a purple assassin syphon SAW with 2 great bonus rolls and 4 open mod slots is an extremely rare (maybe non-existent) item. Compare that to an orange no synergy, no nano, no bonus, TACC Autopistol with no mod slots open. Assess the rarity and yet some will chase the orange forever. That skews any data you can compile since color alone is not determinant of overall rarity and may not be used as much as we think by the game to dole out items.

Cavadus
07-03-2013, 01:29 PM
ITT: A lot of people humorously assume their feelings are a much better indicator of lockbox drop probabilities than actual data.

Indra Echo
07-03-2013, 01:35 PM
ITT: A lot of people humorously assume their feelings are a much better indicator of lockbox drop probabilities than actual data.

This and exactly my point in my edit above. People look at the color, see orange, and go "rare! Yippee!" I have some blues that I tend to think would be far more rare and desirable than some of the oranges people think are rare.

I have no problem with real data at all but I do think data can't totally account for the lockbox rates because people can mis-remember what they did and there's not a large enough sampling across the board to be able to accurately assess it.

For instance, the other day people were pulling oranges out of a lockbox like crazy. It started with one guy getting 4 and then others were getting 1 or 2 (at most). Several people did that. I went there and didn't get one. I got crap. I generally get crap (but that's not even true-I've gotten oranges before). People are actually a lot more likely to remember the bad stuff they got or the truly amazing ones-they don't remember much of the in-betweens.

People tend to forget that for loot boost weekends Trion says something like it increases the chances of getting rare items-not it increases the chances of getting orange ones.

Yokai
07-03-2013, 01:42 PM
ITT: A lot of people humorously assume their feelings are a much better indicator of lockbox drop probabilities than actual data.

Yup. I can explain it to them, but I can't understand it for them.

Shogo_Yahagi
07-03-2013, 02:01 PM
ITT: A lot of people humorously assume their feelings are a much better indicator of lockbox drop probabilities than actual data.

A lot of people humorously assume that posters know nothing about mathematics, statistics, or data collection methodologies.

Burntpowder
07-03-2013, 03:29 PM
A lot of people humorously assume that posters know nothing about mathematics, statistics, or data collection methodologies.

Well based on the sampling of replies, I'd say that it's a fair assertion.

Personally, RNG is winning with my success rates, or lack thereof. I seem to pull mostly greens, but I imagine there is a fair speckling of blues and purples in there that didn't make the cut. Fact is, people remember the bad and not the good. When you feel you've been slighted, you remember it.

Yokai
07-03-2013, 03:31 PM
To the naysayers who love to post vague, unsupported counterpoints like "bad data is still bad data" and the like, I'll ask one simple question: how many of you have kept records of all of your lockbox results and even bothered to run numbers on the results and check them against the probability math and the validated numbers in the data collection project?

Ya. That's what I thought.

Here: I ate my own dogfood about 2 weeks ago. I kept meticulous results of 28 T4 openings in a row. Observed results fall well within outcomes predicted by the validated data collection (and the binomial distribution behind the probability crunching in the OP).

http://forums.defiance.com/showthread.php?95943-Are-T4-lockboxes-broken-or-heavily-ego-dependent&p=888486&viewfull=1#post888486

Instead of meaningless vague proclamations and assertions, how about real counterpoint with real reasoning and numbers to back it up?

Burntpowder
07-03-2013, 03:36 PM
Here: I ate my own dogfood about 2 weeks ago. I kept meticulous results of 28 T4 openings in a row. Observed results fall well within outcomes predicted by the validated data collection (and the binomial distribution behind the probability crunching in the OP).



... did it make you gassy?

Btw, I do appreciate all the number crunching that has been done, selflessly. Thank you.

TigrisMorte
07-03-2013, 04:43 PM
... did it make you gassy?

Btw, I do appreciate all the number crunching that has been done, selflessly. Thank you.

Nah, TBH most dog food is better cuts of meat than McD burgers so...

Cephious
07-04-2013, 08:37 AM
big waste of time for such a small sample size. after hitting t4 boxes from ego 1500-3800 and not a single orange. i saved salvage and bought a bunch of teir 3s once I graduated to EGO 4000...and I got 3 legendaries in about 20-30 minutes.

Yokai
07-04-2013, 11:33 AM
big waste of time for such a small sample size. after hitting t4 boxes from ego 1500-3800 and not a single orange. i saved salvage and bought a bunch of teir 3s once I graduated to EGO 4000...and I got 3 legendaries in about 20-30 minutes. I know you mean well, but this is simple perception bias instead of based on actual math or an understanding of sample size principles and methodology. As explained above, I've done an empiric validation with a 100% random sample size of ~30 to verify whether or not the Noupoi -validated- collection data is likely close to the mark or skewed. Result: the Noupoi collection data is not significantly skewed.

Read up on sampling techniques and sampling sizes and binomial distribution and probability, then offer an assertion based on numbers and logic and test methodology rather than emotive perception.

I guess I should also point out that I work for a company that deals in pricing science, which involves all manner of segmenting and clustering and forecasting and optimization based on _very_ small sets of data. Which makes a lot of companies a lot of money, because despite what you might take away as convenient rules of thumb from a university statistics course, you can make incredibly accurate forecasts with very spotty data. For example, if you make literally two or three purchases on amazon, they have already calculated highly accurate "product affinity" metrics for you.

Indra Echo
07-04-2013, 12:15 PM
Yup. I can explain it to them, but I can't understand it for them.

I think it's great work you are doing here, using mathematical probabilities and all, and I have no argument with you. I can't presume to even understand all that you went through to arrive at those statistics, so hats off to you.

I do still believe that there are some unknowns that will insert error into all this, and I don't think what you are saying is that people can actually use this as some basis to say they'll buy 100 of a certain type of lockbox in order to increase the probability of 6 (or whatever) of those having orange weapons.

And I don't think you are saying that this even assesses the possibility or probability of a rare or ultra-rare item getting pulled from a box. It only addresses the data you've compiled on what color type of weapon/item is more likely to be found in a given lockbox. I'm saying this because people in the game's chat do constantly quote these statistics as if they measure the likelihood of a rare or ultra rare find. And I don't think that color is the only thing and sometimes not even the main thing that determines rarity.

There is a certain degree of error in collecting such data, especially since even though you have done everything scientific you could do in this and are to be commended. People are giving you anecdotal evidence of what they remember receiving or what they will say they got. That's not even because they mean to do something bad-it's just sometimes what they think happened. To be truly accurate you'd have to be able to track what a sampling of players (random players, not those that just wish to contribute) do actually get and not what volunteers say they got.

For instance, there are people that assert and want to prove that the only good boxes are tier 4s, but I frequently hear people getting oranges from tier 2s (I myself have mostly gotten oranges from tier 2s). Others swear by tier 3s. So without knowing it people might volunteer to give you info based upon their bias. And people do this sometimes with good intent and sometimes with more selfish intent-they want to be right. Nearly one third of all discussions I've heard about which is the better tier box to get ends with an argument. This inserts error into the discussion.

I don't want you to think I'm criticizing you-I'm not. You are doing a great job and I have no doubt most of your contributors are as well. But this will never overcome the argument I hear all the time-"I only buy tier 4s. Never got an orange out of anything else." Or tier 2, tier 3, whatever. Well, that right there is the problem. If I only ever buy tier 2 boxes and I get oranges from some then I'd say tier 2s are better than tier 3s or 4s, but I could be very wrong.

You have to have a sampling where an equal number of each tier is being bought or it skews the results-you must ignore the fact that a given box has more items in it. In order to say that out of 10 tier 4 boxes, 10 tier 3s, and 10 tier 2s a player is somewhat more likely to get oranges, you have to buy an equal number of each type of box. It's not about the item count, but the box count. Otherwise, to me it seems the sampling of higher tier boxes is just too small in comparison to tier 2s.

In order to give an accurate comparison of the boxes and the likelihood of what each box contains you actually have to ignore the number of keycodes spent and buy the exact same number of boxes. Only then can you attempt to statistically predict the possibility of A box dropping A certain color. I may have misunderstood some of your data and what you base predictions on but to me it looked like you were comparing the keycode cost. I think that's erroneous.

Indra Echo
07-04-2013, 12:26 PM
I know you mean well, but this is simple perception bias instead of based on actual math or an understanding of sample size principles and methodology. As explained above, I've done an empiric validation with a 100% random sample size of ~30 to verify whether or not the Noupoi -validated- collection data is likely close to the mark or skewed. Result: the Noupoi collection data is not significantly skewed.

Read up on sampling techniques and sampling sizes and binomial distribution and probability, then offer an assertion based on numbers and logic and test methodology rather than emotive perception.

I guess I should also point out that I work for a company that deals in pricing science, which involves all manner of segmenting and clustering and forecasting and optimization based on _very_ small sets of data. Which makes a lot of companies a lot of money, because despite what you might take away as convenient rules of thumb from a university statistics course, you can make incredibly accurate forecasts with very spotty data. For example, if you make literally two or three purchases on amazon, they have already calculated highly accurate "product affinity" metrics for you.

This is off topic. I know it but since you brought it up. People are not just statistics.


I'm sorry. Not to take away from your education and work experience and all but I'm actually sort of sick of just the thing you apparently do. In fact, based upon 2 purchases from Amazon what ultimately happens is that Amazon then steers its customers to have certain things thrown in front of their faces that they may not have ever wanted but are now given impetus to buy. That's what it's used for-a company's convenience and not in order to actually help the customer. As a savvy consumer I then must go beyond and search for exactly what I want and will be the most happy with ultimately-what I had decided to buy myself, and not what they decided would garner them the most money in the short-term.

This is what is wrong with consumerism today and how companies use it. There is no desire to maintain long-term customer bases based upon actually fulfilling the customer's needs and exceeding expectations. The desire is short-term exorbitant profit margins based upon leading customers around with some carrot tied to a stick. Read what I'm saying-the company's interpretation of what I may buy based upon their metrics in no way can ascertain what I will ultimately be happy that I did buy.

I've worked in retail as well as many other places involving Customer Service where you actually talk to people to understand more nuance than what raw data can show. I even worked at one time in the Funeral Services industry (and there's a real corollary here about impulse buying). There's a big difference between serving someone's short-term wants and their long-time needs. The latter can mean oodles of money over a life-time. The former can mean short-term profits with a high probability of returns and even anger. Customers are loyal people. Targeted consumers are mere dollars.

As such I do know what companies want their employees to do in order to pump up the daily stats. I also know that doing so ensures that in a day or two there's a reasonable likelihood that buyer's remorse will take hold and that a customer that might have trusted someone to help them make an informed decision, may never do so again. And some other store might now get their money.

Of course this leads to the other big problem and that's often shareholders that drive the need for targeted sales for short-term profits. Unfortunately, what often happens is that this can all catch up to a company when they realize they don't have a loyal following, they have profits that resulted from targeting impulse buying and they can't keep up with their shareholders' demands for more profit. Then they outsource everything to countries like China or elsewhere where "slave" labor is somewhat legal. Or as some companies are trying to do, they go back to being privately held.

Yokai
07-04-2013, 02:18 PM
This is off topic. I know it but since you brought it up. People are not just statistics.


I'm sorry. Not to take away from your education and work experience and all but I'm actually sort of sick of just the thing you apparently do. In fact, based upon 2 purchases from Amazon what ultimately happens is that Amazon then steers its customers to have certain things thrown in front of their faces that they may not have ever wanted but are now given impetus to buy. That's what it's used for-a company's convenience and not in order to actually help the customer. As a savvy consumer I then must go beyond and search for exactly what I want and will be the most happy with ultimately-what I had decided to buy myself, and not what they decided would garner them the most money in the short-term.

This is what is wrong with consumerism today and how companies use it. There is no desire to maintain long-term customer bases based upon actually fulfilling the customer's needs and exceeding expectations. The desire is short-term exorbitant profit margins based upon leading customers around with some carrot tied to a stick. Read what I'm saying-the company's interpretation of what I may buy based upon their metrics in no way can ascertain what I will ultimately be happy that I did buy.

I've worked in retail as well as many other places involving Customer Service where you actually talk to people to understand more nuance than what raw data can show. I even worked at one time in the Funeral Services industry (and there's a real corollary here about impulse buying). There's a big difference between serving someone's short-term wants and their long-time needs. The latter can mean oodles of money over a life-time. The former can mean short-term profits with a high probability of returns and even anger. Customers are loyal people. Targeted consumers are mere dollars.

As such I do know what companies want their employees to do in order to pump up the daily stats. I also know that doing so ensures that in a day or two there's a reasonable likelihood that buyer's remorse will take hold and that a customer that might have trusted someone to help them make an informed decision, may never do so again. And some other store might now get their money.

Of course this leads to the other big problem and that's often shareholders that drive the need for targeted sales for short-term profits. Unfortunately, what often happens is that this can all catch up to a company when they realize they don't have a loyal following, they have profits that resulted from targeting impulse buying and they can't keep up with their shareholders' demands for more profit. Then they outsource everything to countries like China or elsewhere where "slave" labor is somewhat legal. Or as some companies are trying to do, they go back to being privately held.


All very valid points. Unfortunately, just like government ability to spy on its citizens and the citizens of other countries in extremely efficient ways now, due to technology, none of this will change. The only self-defense for the individual consumer is awareness and actively working against the methods used.

Which brings me to my point about your post. :)

I'm attempting to use --and educate the playerbase about-- techniques like these to HELP this game's consumers. By showing why and how T4 is best if you want an orange, and that the Store prices for bits and lockboxes are too high.

Mladenov
07-04-2013, 02:24 PM
Ok stop arguig with each other!!! Im gonna slap you ALL!!!
Ok, i opened like 20 tier 3 lockboxes and i havent seen worth using item inside!
Just one legendary shield, which reduce damage from my own explosions (fun heh)!
After that i opened tier 2 lockbox with my last resources and guess what! Legendary Vot Fragger with siphon and Cannoneer synergy, mag bonus, and 2 more! BAM gour theory about lockboxes drop rates!

Shogo_Yahagi
07-04-2013, 05:02 PM
To the naysayers who love to post vague, unsupported counterpoints like "bad data is still bad data" and the like, I'll ask one simple question: how many of you have kept records of all of your lockbox results and even bothered to run numbers on the results and check them against the probability math and the validated numbers in the data collection project?

Well, Yokai, I think there's a pretty good chance that I was doing statistics before you were doing arithmetic and I can tell you without the slightest trace of doubt that you're working with a bad dataset because your collection methodology doesn't ensure a random sample of the data. You keep saying "validated data" like it means something other than the fact that a stranger on the internet entered numbers that were the right total of items for the number of items in the lockboxes they said they were entering. I could say that I opened 100 T4 lockboxes and got 400 oranges, and your data validation metric would say "yup, sounds good to me." So yeah, I'm going to stand by my statement that a lot of bad data is still bad data.

Sure, it's possible that everybody saying there's a problem with your numbers is waaaaaaaaaaaay off in the bad tail of the random number distribution. It's not likely, however, given that we know your data is not a representative sample to start with, and your validated data isn't validated in any real sense of the word.

Have I tracked my numbers? No. However, I do know exactly how many oranges I've received. I've racked up almost 3000 EGO between 2 characters and dumped all of my keycodes, including a lot from play time when emergencies still gave keycodes and a fair number of keys purchased with arkfall salvage, into T4s (except for 6 T3s that I picked up recently on the not-so-off chance that Sledge was wrong when he said that T4s had a better drop rate for oranges), and I have a grand total of 1 orange item as a result. Despite that, sure, why not. I'll start tracking my results now. If I miraculously start averaging 1 in 20 oranges, I'll be sure to let you know, but I wouldn't hold my breath if I were you.

Wtflag
07-04-2013, 05:36 PM
The last eight T4s produced only greens and blues.

The chances of that happening based on the provide probabilities of 16.66% chance of getting a purple or orange per item is 0.3%.

Either it is extreme bad luck, or they have adjusted the chances.

Yokai
07-04-2013, 07:30 PM
The last eight T4s produced only greens and blues.

The chances of that happening based on the provide probabilities of 16.66% chance of getting a purple or orange per item is 0.3%.

Either it is extreme bad luck, or they have adjusted the chances.

In my trial of 28 T4s opened in a row, I definitely had a streak of more than 8 boxes without a purple or orange. Assuming the drop rates haven't been recently nerfed into the ground (unlikely), then, yes, it was just "bad luck" for the purples for you.

Remember, there's still a 2.3% chance of not seeing a single purple at the currently measured drop rates in the validated collection data. Have you ever rolled a critical 100 on a dice roll for table top games? Yep, it happens, despite the low odds.

Remember, the notion of "cumulative probability" (yes, this is inexact but it is adequate layperson conceptual terminology) works like this:

In binomial distribution problems where you're trying to calculate the probability of 7 heads in a row from 20 coin flips: If you do 1000 sets of 20 coin flips each and measure the results of all 1000 sets, you'll see results that are VERY close to the calculated probability. But on any one of those 1000 sets? You could get any kind of result that is no where near the calculated probability. This is why I use the word "cumulative", even though some people conversant with statistics and probability will hate that imprecise term. (Note that wikipedia will do a redirect from "cumulative probability" to "binomial distribution".)

As for getting no oranges from those 8 T4s, that's well within predicted outcome. There, you had a 15.6% chance of not seeing an orange.

As I tried to show further up in this thread, with my own results, if you were able to open 30 T4 lockboxes in a row, you'd end up with actual data that would be within a 15% margin of error at a 90% confidence level. This is close enough for rock and roll to match against the probabilities predicted by the Noupoi collection project (the validated entries only). If the observed results from the ~30 experiment seem to fall within the predicted probabilities of the collection project data, then you've gotten the closest possible validation of the overall accuracy of the collection project as we'll ever be able to get, as a player community doing it the hard way.

And the results did match. Therefore, I assume that the collected data is "close enough for rock and roll", and therefore we have validated for ourselves that Trion's assertion that T4 gives you the best chance at legendaries (yes, Sledgehammer did assert that back just after launch) is true. We can also make some confident observations about the real world monetary cost per orange if you buy your lockboxes only with bits.

Perfect? No. Relatively confident? Confident enough to BET on? Yes.

Maitreakow
07-08-2013, 10:39 AM
/bumpity bump bump since there seems to be discussions about this lately

defiant
07-08-2013, 11:13 AM
I've been spamming t3's since launch and only have one orange fragnova to show for it, I've salvaged everything else lol lucky eh

Yokai
08-23-2013, 01:17 PM
Updated the OP with an observation that nothing significant has changed because of any changes in DLC1.