Steem Experiments: Is it better to sell your vote with Smartmarket or Minnowbooster?

There are two services on Steem currently which allow you to sell your vote directly, rather than through delegation: MinnowBooster and SmartMarket. You give them posting access to your account, and they use it to give your vote to people who pay them for it, and pass that payment on to you after taking a percentage for the service. (Since I'm crass those are my affiliate links.)

There are lots of reasons someone might sell their vote:

1. Because they need cash
2. So they don't have to micromanage their account to make sure it never sits at 100%
3. Because they're going on vacation, or have other life reasons not to be here for a significant period of time
4. Because they're leaving Steem but want to make use of their voting power while they power down
5. More I'm sure.

But it makes sense to compare the two services when deciding how to go about selling your vote, and there's not a lot of easy information available. I was intrigued by an experiment I ran into a few weeks ago (and which I now cannot find again) where a user ran a week with his account on each service and compared them. But since it was his active account, and the weeks were different, I didn't think the experiment was well-enough controlled for changes in SP and Steem price.

I have some testing accounts, which I'm able to against each other at the same time, with the same starting SP. So I set out to replicate the experiment under more controlled conditions.

Methodology

I set up two accounts with precisely 120 SP at the beginning of the experiment through delegation. I allowed the SP to grow through the system's growth mechanism, so they have 120.045 SP now, but did not claim any rewards. I voted both accounts down to as close to 79% voting power as I could with no slider (there was about 0.8% extra in the MinnowBooster account; this difference should be trivial) and signed each of them up with one of the services, with the vote-selling set to stop when the account fell below 80%.

I made one error: Midweek I accidentally voted with one of the accounts. Fortunately I noticed and immediately voted the same post with the other one. That shouldn't matter for comparisons but it does skew weekly value by a little over 1%.

Observations

The two services work very differently. SmartMarket votes like clockwork; if you set the voting limit to 80%, anytime you check the account it will be between 77% and 83%. On the other hand, MinnowBooster can go long periods of time without voting, and then vote many times over a short period. That means the voting power is highly variable, and this weekend that account has spent substantial chunks of time sitting at 100%, which is naturally going to hurt Minnowbooster's raw numbers.

At the beginning of the experiment SmartMarket allowed customers to buy votes with Steem as well as SBD, but thankfully my account only got SBD payments, so they can be compared directly.

This analysis does not consider curation rewards, because I'm writing it at the end of the voting week. That would just reinforce the huge effect of the missed votes on the MinnowBooster side anyway. If I try to replicate the experiment in a situation where MB will vote all the time I'll include them then.

Results

Metric	SmartMarket	MinnowBooster
Votes Sold	89	53
Total VP Used	8686%	4699%
Total Payout SBD	0.62583857	0.424404
Payout/10k VP	0.7205	0.9032
Ending VP	78.6%	100%

Really I should probably find a way to do this with rshares, since 100% VP used at 95% and 100% VP used at 80% aren't the same. I'm not prepared to spend a lot of time on that right now though, and I'm not even sure how to find the rshares value of a single vote without writing code. I'm just going to hope the error is small.

Conclusions

If your goal is to keep an account from wasting voting power by sitting at 100%, use SmartMarket. This is a really clear difference there, and if you set SmartMarket to vote when your account is over 97% it will do it. Even with the account set at 80% Minnowbooster let it spend quite a bit of time wasting power.

On the other hand, the edge in payment for votes when sold went to MinnowBooster, significantly enough that it shouldn't be accounted for just by the difference in VP. So without the flaw of letting the account sit at 100%, MinnowBooster may very well have the edge. One way to accomplish that might be to set the limit much smaller, say 50%. This is something I can do another experiment to confirm at some point.

It's also an option to use both services at the same time. You could use both at the same level and let them fight it out for your votes, or set MinnowBooster at a low threshold and SmartMarket at a higher one, so that MinnowBooster sells more votes but SmartMarket protects against the times when it has long pauses.

That last is probably the most profitable strategy.

More Steem Experiments?

I'm interested in setting up more interesting experiments with these accounts, and I'm looking for ideas. I have a plan for one account for next week, but beyond that I don't have a lot in mind yet. Are there any questions you want answered?