🐳@bellyrub powered up to $50+ An analysis of recent auctions shows some huge returns! 💰💰💰

Have you noticed that the auction-style upvote-bot @bellyrub has been growing recently? It's now in possession of a total upvote worth in excess of $50! Have you taken advantage of the potential returns to boost your post? There have been some huge opportunities over the last few days!

If you aren't already aware @bellyrub is an auction-style upvote-bot. You can read about its new, extra-big belly, the auction process, and how much you can bid by following this link.

---

Background

I've been starting to learn SQL over the past couple of weeks and I though I'd put my new found knowledge into practice by analysing the value-for-money of some of the upvote bots. I've started by looking at @bellyrub under its increased Steem power. This analysis is based on the data in sql.steemsql.com, a fantastic tool run by @arcange. Thank you for making this available!

Before I start it's worth noting that the main purpose of upvote bots is to give your work visibility. Getting your beautifully crafted post launched onto the hot list or the trending list can help to get it seen by more people who may share your interests. For minnows and new Steemians (welcome!) this can be a valuable resource. As such, the "value-for-money" is far from the main story. Still, it's always nice to understand how much of a boost you may get for your hard earned SBD!

---

Analysis

The basic statistic I've looked at is simply:

the upvote value a post has received from @bellyrub / the SBD bid to obtain that upvote

Since bidders obtain an amount of the total upvote proportional to their SBD bid, this statistic is (broadly) equal for all bidders in each upvote window.

I've not considered any reduction for the 25% of a post's value that goes to curators. The amount that a poster retains of this 25% varies depending on the amount of previous upvotes in the first half-hour of a post's life and the timing of the @bellyrub upvote (which some people try to target to arrive in the first half-hour).

I've also not taken into account that some SBD amounts were returned by @zeartul (who runs @bellyrub) but the upvotes were still made. This free lunch only benefited some bidders.

I've carried out the analysis at a very granular level (the level of each individual bidder in fact) but then grouped the results into a total statistic for each upvote window. There are roughly nine windows per day (I always thought it was ten but it's 24 hours / 2hours-40ish-minutes). Due to a slight (and variable) delay at the end of each window, the upvote periods are not at the same time each day. However over the three days that I've analysed (Saturday through Monday) the nine widows line up sufficiently to allow some consideration of the effect of bidding at different times of day.

---

Results

The @bellyrub delegated steem power increased to around a $30 upvote a week ago and then to over $50 at the end of Saturday. From the results it does not appear that many people were aware of the increase. The upvote value far exceeded the bid value in some windows. In one case $5 upvotes for each $1 bid were available!

This first graph, with all windows in chronological order, shows that:
(a) upvote value / SBD bid has been consistently above 100% - 26 out of 27 windows
(b) upvote value / SBD bid was huge in some windows on Saturday!
(c) these huge percentage reduced as more people become aware of the increased power available but were still around 200% for some windows on Monday

This second graph gathers together the results: nine daily windows (from midnight to midnight UTC) for each of three days. It demonstrates that there is some "hourly seasonality" in windows with the mid-afternoonn window the most likely to be overbid (the data is limited but @booster has a similar effect).

The full results are shown below.

---

Further work

I'll be looking at @booster next with its mysterious front-runners guild! How does it compare?

I'll be keeping an eye on @bellyrub though. I aim to provide some form of regular report update across the main upvote bots. Let me know what you think would be useful!

---

Final notes

Please remember that the returns from auction-style upvote-bots are not guaranteed and are dependent on the amount of bids made in the window. Your post may receive less of an upvote than the SBD you have bid.

I note that I have no affiliation with @bellyrub (or any other upvote-bot) and my analysis is entirely independent.

---