Thank you for the contribution. It has been approved.
I've not seen a better effort at drilling down into the curation engine and trying to display where the curation sweet spot is at. An awesome effort sir!
That's really cool, crokkon.
I'm not too concerned about the graph, but the percentage helps us quantify our results. Could you crunch the numbers on me, please?
It'd be interesting to see the percentage change over time.
Great analysis.
Fantastic and hugely complex analysis @crokkon, particularly switching the timing to find the optimum voting time; a very impressive inclusion. Thanks also for including me and mentioning my posts!
For an explanation of my curation approach above, for anyone reading who is interested in the technique I applied, all of my voting is manual. In the period you covered I was challenging myself to win @abh12345's curation challenge and I based my approach on hunting for good curation opportunities rather than a specific time-interval based approach (which seems the approach for the other two users).
I simply searched for posts where I had a good idea a whale vote would follow. Posts on utopian, or OCD, also potentially DTube (although I wasn't looking at this at the time) fall into this category.
Once you are out of the 30 minute reverse auction period, the specific timing is no longer important, just the volume of upvotes that arrive before you. So a curation opportunity that can be found at 31 minutes may still be unchanged at 5 hours, as long as no one voted on it in-between. This explains why a lot of my upvotes are much later than under an interval approach.
I could definitely have received more curation rewards if I had voted earlier, but the problem is being certain that the post will receive the whale vote - for that you have to wait, or take a risk and guess. There are certainly some bot trails now that are taking the latter approach, I assume based on poster success history, and I would expect their curation rewards to be very strong.
First of all, thanks for the warm words, the resteem, the vote (at a probably well-paying position) and the approval!!!
Curie certainly creates its own sweet spot, but I'm not sure about biophil. He wouldn't have to vote before 30 mins with a reasonable trail in hand.
Keep in mind that this analyzes only the position of the votes and not the selection of the posts. You can still have pretty poor curation rewards while voting at the exact optimal position because the selection of posts or vote-% is wrong.
I switched all my autos to 15 minutes a few weeks ago, but this morning (honestly!) I changed 99% of them to 24 minutes. I'm pretty sure that analysing @abh12345 would show i'm voting too early (my rewards have halved!)
Are you only running the analysis on top level comments? If not, @zapncrap would be interesting :)
Keep in mind that this analyzes only the position of the votes and not the selection of the posts. You can still have pretty poor curation rewards while voting at the exact optimal position because the selection of posts or vote-% is wrong.
Hi - this is superbly detailed analysis - the kind of thing that should be linked to in the FAQ page... I know for a fact that I vote too early!
You might want to add in a little conclusion at the end (if I've read it correctly) - for maximum curation rewards... vote at between 20-30 minutes!
Interesting comments below about the bots... which I guess means that profiting from manual curation is, in the future, going to be more of a challenge!
Thanks! This post was not really meant as a FAQ-level post and I expected readers that make it through the article to be aware of "vote at between 20-30 minutes" as a rule of thumb :)
The 3 links at the beginning of the post should give all necessary information for readers without this background, also summarized to FAQ-level, but additionally with a lot of details.
Actually this post was more meant to point out that within this window between 20-30 minutes it can make a big difference when you vote, but this strongly depends on your selection of posts, SP, VP, %, ....
Bots play an interesting role at the moment. I think they can generate quite some curation rewards on the short run, but I doubt that they can replace manual curation long term.
Cheers - I checked out your post because I was following a 'data analytics people trail', not because I was looking for curation tips, but hey that's useful to know, I'll check it out.
It's one of the downsides of steemit ATM that it doesn't really do the freeze frames of your graphs justice!
Great work though. I will check out those other 'basics' posts.
sure, here are your graphs. The statistics for @josephsavage and @sneakyninja are a bit low, so don't draw too many conclusions on those.
In median @josephsavage is a few minutes too late, but also contains a few votes that came too early.
@sneakyninja hits the sweet spot pretty well for votes around the first 30 min, a few votes too late and only rarely too early.
@steembasicincome is interesting, there are almost no votes too early. In general it is however a couple of minutes too late. How does this bot vote - does this have a fixed voting time after a post is created?
To be honest, I care more about the statistics for @steembasicincome than for the other two, as different curation strategies are followed.
@steembasicincome uses steemvoter at specified delay to provide regular income to participants, through upvotes. Since optimizing curation rewards would help the program to deliver better results, we really appreciate this report. It looks like moving all the votes forward 2-3 minutes would make a nice improvement on curation rewards and give each participant a little more income.
This is fascinating, @crokkon, and very helpful - as I'm fairly new, here, and still trying to figure out why some posts do better than others (when they are, basically, of the same quality).
Looking forward to seeing graph of my account & I thank you, in advance!
hey @yahialababidi, glad you like it!
Your graph unfortunately doesn't say very much because there is not much data from you. You have voted only on around 100 posts in the time range from Dec. 7th 2017 to Dec. 28th. A third of your votes in that time would not have created curation rewards even at the optimal position and another third was spent on your own posts. From the remaining third, only 13 fell into the first 5 hours. Well, drawing conclusions from 13 votes is a bit hard. But congratulations, you basically never voted too early on those! :)
Impressive post and analysis. @biophil is the king IMO. I would be interested in seeing mine, although it will likely be skewed in some weird ways because I often front run the Curie vote on a post that is older than 5 hours (I am a top @curie curator). A good chunk of my votes are actually set to auto follow my @r-bot and @humanbot account votes, but I set a 5 minute delay to give some curation reward to the vote trailers who signed up to follow @r-bot and @humanbot, so that definitely drags down my overal numbers. Still I have placed top 10 in Asher's league quite a few times I believe, including last week. My placement there is dragged down a bit because I spend a good chunk of my SP upvoting comments, yielding next to no curation reward, so if your analysis is on top level posts only my performance may be better than in Asher's league. Cheers - Carl
Hi Carl, thanks, here are your stats:
You clearly have the "curie pattern" in your actual voting times, however with a few more votes around the first 30 minutes. Those are rarely too early, but sometimes a few minutes "too late". More than half of the votes are outside the 5h window and not included and shown here. So all in all it fits your description pretty well!
Awesome thanks for whipping that out so fast! Yeah, no real surprises for me there but sometimes it is nice to see the data confirming what you had already intuited. Thanks!
Thank you for the contribution. It has been approved.
I've not seen a better effort at drilling down into the curation engine and trying to display where the curation sweet spot is at. An awesome effort sir!
Highly approved!
You can contact us on Discord.
[utopian-moderator]
That's really cool, crokkon.
I'm not too concerned about the graph, but the percentage helps us quantify our results. Could you crunch the numbers on me, please?
It'd be interesting to see the percentage change over time.
Great analysis.
Thanks @mattclarke!
here are your stats:
Seems like you tend to vote a little too early.
A percentage change over time is really interesting, thanks for the idea! I'll think about that!!!
Thanks mate. That's really cool.
Fantastic and hugely complex analysis @crokkon, particularly switching the timing to find the optimum voting time; a very impressive inclusion. Thanks also for including me and mentioning my posts!
For an explanation of my curation approach above, for anyone reading who is interested in the technique I applied, all of my voting is manual. In the period you covered I was challenging myself to win @abh12345's curation challenge and I based my approach on hunting for good curation opportunities rather than a specific time-interval based approach (which seems the approach for the other two users).
I simply searched for posts where I had a good idea a whale vote would follow. Posts on utopian, or OCD, also potentially DTube (although I wasn't looking at this at the time) fall into this category.
Once you are out of the 30 minute reverse auction period, the specific timing is no longer important, just the volume of upvotes that arrive before you. So a curation opportunity that can be found at 31 minutes may still be unchanged at 5 hours, as long as no one voted on it in-between. This explains why a lot of my upvotes are much later than under an interval approach.
I could definitely have received more curation rewards if I had voted earlier, but the problem is being certain that the post will receive the whale vote - for that you have to wait, or take a risk and guess. There are certainly some bot trails now that are taking the latter approach, I assume based on poster success history, and I would expect their curation rewards to be very strong.
Anyway, great post, great read!
Pretty sure biophil has a trail and his own bot engine.
Could you say that because of this he is creating the voting sweet-spot for the post?
miniature has been hunting utopian-io posts that will take a nice up-vote from their bot - you have hours/days in hand there as you know.
not sure on bugged-out but i think he would be interested see that a minor change to his technique could pay (more!) dividends.
Sweet post!
First of all, thanks for the warm words, the resteem, the vote (at a probably well-paying position) and the approval!!!
Curie certainly creates its own sweet spot, but I'm not sure about biophil. He wouldn't have to vote before 30 mins with a reasonable trail in hand.
Keep in mind that this analyzes only the position of the votes and not the selection of the posts. You can still have pretty poor curation rewards while voting at the exact optimal position because the selection of posts or vote-% is wrong.
No problem, really great work!
I switched all my autos to 15 minutes a few weeks ago, but this morning (honestly!) I changed 99% of them to 24 minutes. I'm pretty sure that analysing @abh12345 would show i'm voting too early (my rewards have halved!)
Are you only running the analysis on top level comments? If not, @zapncrap would be interesting :)
Understood, cheers!
seems like you have a spread of too-early and too-late votes around the optimal time for @abh12345:
sorry, I'm only running on top level comments a.t.m.
For the autos: did you check the actual voting times? I noticed streemian being around 2 minutes late in my case, which can make quite a difference.
I'm running Steemvoter and it has been very hit and miss recently.
May request a 7 day view in a weeks' time to see if that's improved anything.
Cheers!!
Hi - this is superbly detailed analysis - the kind of thing that should be linked to in the FAQ page... I know for a fact that I vote too early!
You might want to add in a little conclusion at the end (if I've read it correctly) - for maximum curation rewards... vote at between 20-30 minutes!
Interesting comments below about the bots... which I guess means that profiting from manual curation is, in the future, going to be more of a challenge!
Thanks! This post was not really meant as a FAQ-level post and I expected readers that make it through the article to be aware of "vote at between 20-30 minutes" as a rule of thumb :)
The 3 links at the beginning of the post should give all necessary information for readers without this background, also summarized to FAQ-level, but additionally with a lot of details.
Actually this post was more meant to point out that within this window between 20-30 minutes it can make a big difference when you vote, but this strongly depends on your selection of posts, SP, VP, %, ....
Bots play an interesting role at the moment. I think they can generate quite some curation rewards on the short run, but I doubt that they can replace manual curation long term.
Cheers - I checked out your post because I was following a 'data analytics people trail', not because I was looking for curation tips, but hey that's useful to know, I'll check it out.
It's one of the downsides of steemit ATM that it doesn't really do the freeze frames of your graphs justice!
Great work though. I will check out those other 'basics' posts.
This is really interesting! Would you be willing to do this for my account (and for my projects @sneakyninja and @steembasicincome)?
Thank you so much!
sure, here are your graphs. The statistics for @josephsavage and @sneakyninja are a bit low, so don't draw too many conclusions on those.
In median @josephsavage is a few minutes too late, but also contains a few votes that came too early.
@sneakyninja hits the sweet spot pretty well for votes around the first 30 min, a few votes too late and only rarely too early.
@steembasicincome is interesting, there are almost no votes too early. In general it is however a couple of minutes too late. How does this bot vote - does this have a fixed voting time after a post is created?
Thanks @crokkon, this is very helpful.
To be honest, I care more about the statistics for @steembasicincome than for the other two, as different curation strategies are followed.
@steembasicincome uses steemvoter at specified delay to provide regular income to participants, through upvotes. Since optimizing curation rewards would help the program to deliver better results, we really appreciate this report. It looks like moving all the votes forward 2-3 minutes would make a nice improvement on curation rewards and give each participant a little more income.
I'm glad to see that you can make real use of this analysis!
This is fascinating, @crokkon, and very helpful - as I'm fairly new, here, and still trying to figure out why some posts do better than others (when they are, basically, of the same quality).
Looking forward to seeing graph of my account & I thank you, in advance!
hey @yahialababidi, glad you like it!
Your graph unfortunately doesn't say very much because there is not much data from you. You have voted only on around 100 posts in the time range from Dec. 7th 2017 to Dec. 28th. A third of your votes in that time would not have created curation rewards even at the optimal position and another third was spent on your own posts. From the remaining third, only 13 fell into the first 5 hours. Well, drawing conclusions from 13 votes is a bit hard. But congratulations, you basically never voted too early on those! :)
Thanks, for chart & taking time to explain that to me -we Steemit & learn
this is something new to me, thanks for sharing..
Impressive post and analysis. @biophil is the king IMO. I would be interested in seeing mine, although it will likely be skewed in some weird ways because I often front run the Curie vote on a post that is older than 5 hours (I am a top @curie curator). A good chunk of my votes are actually set to auto follow my @r-bot and @humanbot account votes, but I set a 5 minute delay to give some curation reward to the vote trailers who signed up to follow @r-bot and @humanbot, so that definitely drags down my overal numbers. Still I have placed top 10 in Asher's league quite a few times I believe, including last week. My placement there is dragged down a bit because I spend a good chunk of my SP upvoting comments, yielding next to no curation reward, so if your analysis is on top level posts only my performance may be better than in Asher's league. Cheers - Carl
Hi Carl, thanks, here are your stats:
You clearly have the "curie pattern" in your actual voting times, however with a few more votes around the first 30 minutes. Those are rarely too early, but sometimes a few minutes "too late". More than half of the votes are outside the 5h window and not included and shown here. So all in all it fits your description pretty well!
Awesome thanks for whipping that out so fast! Yeah, no real surprises for me there but sometimes it is nice to see the data confirming what you had already intuited. Thanks!
Most Excellent!!!!!!!
Resteeming and on my way over to approve this masterpiece!