Repository
https://github.com/steemit/steem
Introduction
Like it or not, bid-bots are now a widely used part of Steemit.
For some, they are a tool of promotion to get their post out of the "noise" and for some, it is a way of profit.
The general action of a bid-bot is :
- The user sends Steem or SBD to the bid-bot with the URL of a post in the memo
- The bid-bot collects the bids
- Depending on the total of the bids collected, the bot starts giving upvotes using its total %2 of total power ( until 100% vote) giving upvotes proportional to each users bid to total bids collected.
- After voting window ends, the bot rests for 2,4 hours - or until it reaches 100% power to regain power.
As the ROI ( the upvote user will receive / the money user sent ) totally depends on the total bids collected ( vote value of the bot is constant) there should be a more profitable time frame, where there are fewer users sending bids to use these bots.
In this analysis, it is searched, if such a timeframe really exists.
Scope of Analysis
Analysis date: 15 May 2018
Analysis timeframe: 07 May 2018 00:00 to 13 May 2018 24:00
Bots analyzed:
All times are in UTC.
Tools
Finding the upvotes given to all users
For this analysis the tool SteemBotAnalysis GitHub is used.
I have mainly built this tool to analyze the voting behaviors ( vote date-post date) of the users but it is very suitable for such analysis with the added properties like :
- Timeframe selection
- % Vote given indication
(thanks from here to @transisto for his recommendations on the added properties)
Selecting the bot and the time frame, all the upvote data of the bot is received.
Finding the bid sent
For the bid sent, two methods should be chosen.
1- Write a script to actually take the transfer from block-chain data
2- Put the data to a spreadsheet
- Determine the voting windows
- Take the minimum bet of the bot from @yabapmatt's SteemBotTracker
- Define the minimum vote given for each voting window
- Estimate the bids sent as minimum vote given for minimum bid
I have taken the second approach and validated that it works perfectly unless the minimum bid changes within the analysis week.
Calculate the Gross - ROI
The gross ROI is basically (SBD Value of upvote received) / (Bid Sent as SBD)
Since we have all the data, it is just spreadsheet calculation.
Results
The Gross-ROI findings for 3 bots during 1 week are as follows :
Looking at the 1 week data we can conclude that :
- It is difficult to define a maximum ROI voting window time but we can easily see a much stable time which is between 00:00 UTC to 9:30 UTC for everyday.
- The stable zone generally gives more ROI than unstable zone, less gambling.
- For all three bots, it is clearly seen that bidding on Sunday gives much more ROI compared to other days.
As a final word, I really would like to see if Hawthorne Effect ( observation changes behavior) will work on this analysis.
Just because we have determined a "stable timeframe" and "max ROI date" can change the voting behavior of people and a new study would be required.
Proof of Authorship
Contact
- @FireDream - Steemit
- @firedream#3528 - Discord