Post by me2 on Nov 8, 2008 9:46:44 GMT -5
I'd like to create an own thread about drop statistics, because this is a very difficult topic. There are a lot of objects which can drop (cards, jewels, crystals, stones, weapons, onigiry, money) and a a lot of enemies which drop them. And we can only measure and estimate the probabilities, if we can't look into the code. Also it isn't guaranteed that drop rates stay the same if a new update comes out.
What is the wrong way to measure probability?
If you want to help to get some statistics please don't do it like this:
Here is how to do it right:
If you start measuring, first decide which item(s) you are looking for. If you are looking for stuff which is dropped by normal enemies also decide at which stages and screens you are looking for the item(s). And then think in advance about a number of complete runs you want to do. The higher this number, the better. The best would be if it is larger than the drop rate, but it isn't that important. And now do this number of runs and do not stop until you have finished, no matter if you found one item, no item, or 100 items.
After you have done this you can do it again, as often as you want. But try not to let the fact, whether you did find or did not find anything during the last runs, have an impact on this decision. This is especially important if your chosen number of runs is much lower than the estimated drop rate.
If you can't finish a run, for example because you run out of health points or if your game crashes, do not count any item you found during this run into the statistic. The first screens of a stage are different than the later screens, and if you would count only a half run this will falsify the statistic.
Now the really important thing is: Tell us about it, no matter what happened.. It does not matter if you didn't find the item you were looking for at all, if you found hundreds of those items, or if the drop rate is exactly the same as other guys have measured before you. If you decide to not tell us, for what ever reason, this falsifies the statistic. And this is pretty bad.
I think to get at least a rough statistic we should find the item at least for 3 or 4 times, and to get something more reliable at least 10 times.
What we need to know:
Also we need to know how many runs you did, where you did those runs (if it isn't obvious) and which area (screens/stages) one 'run' covers, for which item you were looking for, and how many of those items you found.
If you look for different items which are dropped by the same enemy also have a look if they can be dropped at the same time or if this has never happened. For example we know a Onigiri's Card can never drop at the same time together with a Vampire's Card. But currently we do not know if any of the other bosses can drop a card together with a weapon at the same time or if this is impossible.
Also always write down the current game version and keep the statistics from different game versions separated. We can always combine them later if we think (or assume) the drop rates are still the same. But its impossible to figure out what's the right number if the drop rate has changed with a new game version.
I think we currently only have some measurements for the cards:
I will start a new card statistic for 2.2 soon.
But you can also have a look for all other items ...
/me
What is the wrong way to measure probability?
If you want to help to get some statistics please don't do it like this:
- "Hey, I found Xyz's Card after 5 runs, you can include it into the statistic."
This is pretty bad, because it means, you probably had some luck and want to tell everybody now. And also it means you have stopped searching after you found this card. - "Hey the comments section says Person_1 found the first Xyz's Card after 5 runs, Person_2 found it after 10 runs, Person_3 after 8 runs, and Person_4 after 15 runs. Can we build a statistic on this?"
This is also bad, because this means that we will only include information from people who actually found the Card. We do not know if there are another 4 people who tried the same and who didn't found the card at all or if there are even 100 people who didn't find the card. This would easily lead to a wrong probability. - "Hey the comments section says Person_1 found the first Xyz's Card after 5 runs, Person_2 found it after 10 runs, Person_3 after 8 runs, and Person_4 after 15 runs. Also Person_5 and Person_6 didn't find a card within 30 runs. Can we build a statistic on this?"
This might better but it's still very vague. We do not know if this corresponds to the normal behavior. Usually if you experience some uncommon behavior like finding a Card very quickly or not finding one at all after a lot of runs, you feel more urge to tell others about your good/bad luck, which means we get more the extreme cases into the statistic and might easily compute a wrong probability. - "Isn't such a bad statistic better than no statistic at all?"
Yes it is. But we should mark them and keep them separated.
Here is how to do it right:
If you start measuring, first decide which item(s) you are looking for. If you are looking for stuff which is dropped by normal enemies also decide at which stages and screens you are looking for the item(s). And then think in advance about a number of complete runs you want to do. The higher this number, the better. The best would be if it is larger than the drop rate, but it isn't that important. And now do this number of runs and do not stop until you have finished, no matter if you found one item, no item, or 100 items.
After you have done this you can do it again, as often as you want. But try not to let the fact, whether you did find or did not find anything during the last runs, have an impact on this decision. This is especially important if your chosen number of runs is much lower than the estimated drop rate.
If you can't finish a run, for example because you run out of health points or if your game crashes, do not count any item you found during this run into the statistic. The first screens of a stage are different than the later screens, and if you would count only a half run this will falsify the statistic.
Now the really important thing is: Tell us about it, no matter what happened.. It does not matter if you didn't find the item you were looking for at all, if you found hundreds of those items, or if the drop rate is exactly the same as other guys have measured before you. If you decide to not tell us, for what ever reason, this falsifies the statistic. And this is pretty bad.
I think to get at least a rough statistic we should find the item at least for 3 or 4 times, and to get something more reliable at least 10 times.
What we need to know:
Also we need to know how many runs you did, where you did those runs (if it isn't obvious) and which area (screens/stages) one 'run' covers, for which item you were looking for, and how many of those items you found.
If you look for different items which are dropped by the same enemy also have a look if they can be dropped at the same time or if this has never happened. For example we know a Onigiri's Card can never drop at the same time together with a Vampire's Card. But currently we do not know if any of the other bosses can drop a card together with a weapon at the same time or if this is impossible.
Also always write down the current game version and keep the statistics from different game versions separated. We can always combine them later if we think (or assume) the drop rates are still the same. But its impossible to figure out what's the right number if the drop rate has changed with a new game version.
I think we currently only have some measurements for the cards:
Opening Street Boss: 0 Drop(s) / 101 Run(s)
Grassland 1 Bosses: 8 Drop(s) / 13 Run(s) (with 12 Bosses)
Grassland 2 Boss: 1 Drop(s) / 37 Run(s)
Grassland 3 Boss: 1 Drop(s) / 19 Run(s)
Grassland 4 Bosses: 5 Drop(s) / 17 Run(s) (with 10 Bosses)
Grassland 5 Bosses: 3 Drop(s) / 24 Run(s) (with 3 Bosses)
Grassland 6 Boss: 2 Drop(s) / 17 Run(s)
Grassland 7 Boss: 3 Drop(s) / 38 Run(s)
Castle Boss: 3 Drop(s) / 3 Run(s)
Note:
More than one drops per run is possible for multiple bosses (GL1, GL4, GL5) and are included in the number of drops.
Statistic is based on version 2.1
Grassland 1 Bosses: 8 Drop(s) / 13 Run(s) (with 12 Bosses)
Grassland 2 Boss: 1 Drop(s) / 37 Run(s)
Grassland 3 Boss: 1 Drop(s) / 19 Run(s)
Grassland 4 Bosses: 5 Drop(s) / 17 Run(s) (with 10 Bosses)
Grassland 5 Bosses: 3 Drop(s) / 24 Run(s) (with 3 Bosses)
Grassland 6 Boss: 2 Drop(s) / 17 Run(s)
Grassland 7 Boss: 3 Drop(s) / 38 Run(s)
Castle Boss: 3 Drop(s) / 3 Run(s)
Note:
More than one drops per run is possible for multiple bosses (GL1, GL4, GL5) and are included in the number of drops.
Statistic is based on version 2.1
I will start a new card statistic for 2.2 soon.
But you can also have a look for all other items ...
/me