Hello guys. Once the new challenges started I decided to track my scores and the number of plays and coins received.

I only did it by difficulty, and I have an estimation of the probability of getting a coin after playing one level in a difficulty. I attach results.

It is clear that the probability of getting coins is not the same. I would like to extend this analysis to each level now, instead of difficulty. I want to do this because as you can see (way more plays in rare levels) I have played way more level 3 rare, and I believe the probability of getting a coin is weighted by the amount of flags of each level. That would help to explain the discrepancy in results!

So this is the summary of the project: I record all levels that I play and the amount of coins received from them. Then if we get a big enough sample we can compute an estimation of probability of getting coins, and see if there is some optimal level (considering flag usage). I will need some help to get more data though so everyone is welcome to help!!

I created a google form Event coin dropping form. The form is made to introduce levels separately (so if you decide to make 10 runs in level 3 rare, you do it and then put the data for that level).

For guys more serious about this and that will provide more data, please download this excel sheet Excel file to download and provide data easily that you should fill and then send to me (lineID: ierazo or email: erazoignacio@gmail.com). An example of how to fill it is here:

Well the main goal is to start it for this Guardians event in January. As always, all data is welcome and please only submit your data. It must be non-biased, so do not only track the coins receivedâ€¦

It seems like a lot of work, so it is acceptable if you only want to give data for some levels or the ones you can track (as an example, you autofarm during the day so you cannot take note of the levels, but if you play yourself at night, you can submit data from those late night levels only!).

Thanks for the attention. I hope to have data at the end of next week so I can give a first look at results!

**PRELIMINARY RESULTS:**

Unfortunaly, less than a handful of people have helped this project with data, so we still have too little in most levels, however as I always use all my flags during this period in challenge events, I still feel like we can get some insights of what we have so far. I will detail as follows:

- Rare Level 3: Over 2503 runs, 48 chests have been received. This screams that the probability of getting a chest in this level is around 1.91%. My guess is that the chance that getting a chest is 2%.

All the other levels have sample sizes below 40, and besides that, because of the way I play these challenges, results are HEAVILY biased. I usually play a level until I get a â€śgood enoughâ€ť score, which usually means that I play until I get a chest in that level and I rush it with items. That being said, because sample sizes are so small, %'s are amplified because I just run each level until I get the first chest. This would be non issue is the sample was big enough, but unfortunately it is not. But by adding all levels over a difficulty (rare, epic, legendary) we may be able to get insights:

- Rare: With total sample size of 2772, the average amount of chests per flag is 0.025, so 2.5%
- Epic: With total sample size of 218, the average amount of chests per flag is 0.0318, so 3.18%
- Legendary: With total sample size of 273, the average amount of chests per flag is 0.0192, so 1.92%

After this summary and the explanation about the bias produced because of the sampling method and the small sample sizes, I believe that the real % chance of getting a chest is around 2% per flag, so it should not matter which level you play or difficulty, the chances of getting a chest depend on the number of flags of the level.

Now, It is clear that one may have more than one chest in a level, so the actual real chance is given by each monster, however, the lack of data does not allow to go deeper in that kind of analysis. These preliminary results suggest though that the chance per monster is constant and adjusted linearly by flag of each level.

If someone is interested in this please let me know, and it would be great if more people was willing to contribute, however I know it is time consuming to track this (as it takes me time to do it).