This is a followup to an earlier post
One way to find posts that will trend is to look for the luckiest curators.
These are people who seem to choose posts that are going to trend, well in advance of trending, i.e. before the post hit the 12% mark.
In order to do this you simply poll the top 100 in trending, then look at each vote one at a time until you get to the vote where the post crossed 12% of the total reward shares. After that list is compiled, simply scan the list counting which voters got in early on 2 or more posts.
This approach is sound, but we found it was favoring the same 15 people constantly.
On further analysis it turns out that several people who were higher in our ranking almost always vote only after a few other people have voted.
In otherwords, those individuals were already running their own lucky list and we wanted to find some way to filter that out, this way we find the curators, not the trails following them.
To maximize our approach we needed a refinement to the model to uncover the actual influencers, i.e. the first people to vote even if they individually caught less trending posts than some of the others.
As it turns out the change was relatively simple. As we count votes, we keep a running total of reward shares for the post prior to each vote being cast. From there instead of sorting by count, we sort by average rshares.
What this gives us is a much better overview of the luckiest influencers, i.e. those who manage to get in early and often on posts before they trend, and before the people who are trailing them.
So now the lucky curator list looks like the one below.
Keep in mind with AVG RShares, lower is better.
Thanks for tuning in!
Speaking of luck, we have created a lottery to help reduce the SBD debt ratio, all SBD earnings from this post will go into the jackpot for it. Good Luck!