Bitcoin Forum

Other => Meta => Topic started by: Zeroxal on December 31, 2015, 09:00:30 PM



Title: Trust score algorithm
Post by: Zeroxal on December 31, 2015, 09:00:30 PM
As we have this trust score that is under our forum name/avatar, I wanted to ask if there is an official algorithm that has been released that determines that score.
I know that the btc amount and timespan play an important role, so is there a specific amount+time that will make your score higher?

I see many members with a positive trust score(like me), but they are not green but black. What does your score need to be to get a green trust?


Title: Re: Trust score algorithm
Post by: tarsua on December 31, 2015, 09:02:25 PM
it needs to be positive 5 to get green, so 5 months
so you will be green in about ~4 months


Title: Re: Trust score algorithm
Post by: Zeroxal on December 31, 2015, 09:06:54 PM
it needs to be positive 5 to get green, so 5 months
so you will be green in about ~4 months
I think it isn't that easy. What about the trusted amount?


Title: Re: Trust score algorithm
Post by: tarsua on December 31, 2015, 09:16:21 PM
it needs to be positive 5 to get green, so 5 months
so you will be green in about ~4 months
I think it isn't that easy. What about the trusted amount?
the more trust you get, the more trust you get, each positive from a trusted user gives you a +10 rating spread over 10 months


Title: Re: Trust score algorithm
Post by: prodigy8 on December 31, 2015, 09:42:12 PM
As we have this trust score that is under our forum name/avatar, I wanted to ask if there is an official algorithm that has been released that determines that score.
I know that the btc amount and timespan play an important role, so is there a specific amount+time that will make your score higher?

I see many members with a positive trust score(like me), but they are not green but black. What does your score need to be to get a green trust?
From what i have seen until now you can get max of 10 points from a single user.

Even if he makes more than one feedback to your account, and it takes 10 months to get the max of 10 points from a single user.

1 month after a user gives you positive trust you will see 1 black color, then from 5 to 14 it is Light Green and from 15+ it is Dark Green


Title: Re: Trust score algorithm
Post by: Quickseller on December 31, 2015, 09:56:01 PM
If you assume that you have no negative ratings then it is easy. Only ratings from people who are in your trust network are taken into consideration. For each person/account that gives you positive trust you get one trust point (the number on the very left) for each month since that account left you the first positive rating with a maximum of 10 trust points from each person.

The second number is the number of people who have given you negative trust.

The third number is the number of people who have given you positive trust.

If the last rating is your first negative rating then the highest score you can have is three question marks. The number of negative ratings you have, to the power of two is the number of trust points that are removed from your trust score (so if you have two negative ratings then 22 trust points are subtracted from your trust score, and if you have four negative trust ratings then you have 24 trust points removed from your trust score). If you have more unique positive trust ratings then the number of negative trust points that your negative trust removes, then your trust score is the number of unique positive trust ratings since your first negative rating minus the number of negative trust ratings.

I believe the above accurately explains it, however the code is below:

Code:
if there are no negative ratings
score = 0
for each rating, oldest to newest
if this rater has already been counted
continue
score += min(10, round_up(months since rating))
else
score = unique_positive - 2^(unique_negative)
if score >= 0
start_time = time of first negative
score = unique_positive since start_time - unique_negative since start_time
if(score < 0)
return ??? (orange)

move score to range [-9999,9999]
return score

This algorithm is