Forum Discussion

Thelebka's avatar
8 years ago

Retreat Mechanics

Can someone refer me to a link, or could a developer please provide specifics, on the raid retreat mechanics. In game messaging states that any character has a 40% chance to retreat during a raid. With an N=200 (none with Zylo) the actual statistic has been on the order of 15%. While a sample size of 200 has a reasonable margin of error, I have received similar statistics from many others. Perhaps the 40% stated retreat chance is only a base chance modified by other factors. I believe it would be helpful to the community if this was explained in great detail. Thanks in advance.
  • when i'm trying to escape a team because i want to die without doing anymore damage, i usually am able to escape more than 1 character. Granted, i don't really pay attention because dying is just as good as escaping for me personally, but it seems closer to 40% than 15% in my experience.
  • "Man_Uel;c-1195873" wrote:
    I have a word for you: stochasticity


    I'm sorry but the data cannot be explained away with a simple attribution to RNG. A look at the variance and the resulting standard deviation does not support a 40% baseline. As the N of these data sets approaches 1000 margin of error will be in the +/- 5% range. It would be highly improbable that multiple data sets are all outliers. Assuming the trend in the data continues, there are two logical explanations:

    1) The retreat mechanics are functioning as intended returning a real value of 15% (or whatever the end result shows) per attempt. If so the in game messaging is in error and should be treated as a bug and corrected.

    2) The retreat mechanics are not functioning as intended, and there is a technical bug, rather than messaging bug.
  • "leef;c-1195911" wrote:
    when i'm trying to escape a team because i want to die without doing anymore damage, i usually am able to escape more than 1 character. Granted, i don't really pay attention because dying is just as good as escaping for me personally, but it seems closer to 40% than 15% in my experience.


    Leef what is the N of your current data set and did any of the trials involve Zylo? Thanks.
  • Don't forget that if your escape attempt fails and you use retreat, you will reenter the battle with the same RNG seed. I you don't change anything on your attack run, you will get the same result on the escape attempt.
    So, not all data is valid - the number of "fail" may easily be overestimated.
  • "Ambassador;c-1197376" wrote:
    Don't forget that if your escape attempt fails and you use retreat, you will reenter the battle with the same RNG seed. I you don't change anything on your attack run, you will get the same result on the escape attempt.
    So, not all data is valid - the number of "fail" may easily be overestimated.


    An excellent point! This could easily be a skew. I will need to cull the data sets to account for this...
  • "Ambassador;c-1197376" wrote:
    Don't forget that if your escape attempt fails and you use retreat, you will reenter the battle with the same RNG seed. I you don't change anything on your attack run, you will get the same result on the escape attempt.
    So, not all data is valid - the number of "fail" may easily be overestimated.


    Pretty sure the RNG seed is only fixed for GW, not raids.
  • "thelebk;c-1197323" wrote:
    "leef;c-1195911" wrote:
    when i'm trying to escape a team because i want to die without doing anymore damage, i usually am able to escape more than 1 character. Granted, i don't really pay attention because dying is just as good as escaping for me personally, but it seems closer to 40% than 15% in my experience.


    Leef what is the N of your current data set and did any of the trials involve Zylo? Thanks.


    i'm not tracking data, mainly because it seemed to be around 40% when playing so i didn't feel the need to track my succesrate with escaping.