[Transcoder Analytics Part 2]: Do you really understand the income of a transcoder?

In my previous post, I talked about some very basic concepts of Livepeer transcoders.
In this post, let’s take a deep look into what are the factors that impact a transcoder’s income.

First, let’s look at the income that top two transcoders are pulling in right now. And I am benchmarking them against the community node 0x4f47. In the “Income Breakdown” chart below, the x axis is the round number and the y axis is the number of LPT tokens. The green line is the total income, the yellow line is the income from the reward cut and the deep blue line is the income from transcoder’s own staked tokens. I will refer it as “self staking” or Pending Stake from here on.

0x4f47: Reward Cut 50%, Total Stake: 134k+, Pending Stake: 34K
0x481e: Reward Cut 0.9%, Total Stake: 704k+, Pending Stake: 1.25K
0x4ff0: Reward Cut 1.25%, Total Stake: 699k+, Pending Stake: 3.67K

In the most recent round 1286, the most staked transcoder 0x481e made less than 30 LPT; The second most staked transcoder 0x4ff0 made a little over 40 LPT. But 0x4f47 made 300 LPT! What is happening here? Well, I think there are two elements here:

  1. Pay attention to your Pending Stake. In the “Income Breakdown” chart above, 40% of 0x4f47's income is from its “self staking” while 29% and 16% for 0x4ff0 and 0x481e respectively. As I mentioned in my previous post, the reward amount generated based on the size of Pending Stake 100% goes to the transcoder. At this stage of the development of Livepeer network, I think this is the most predictable income source. As the owner of the transcoder, you have full control of whether and how you want to unbond or keep bonding the “self staking” portion towards the transcoder. There are still external factors (such as current mintable tokens, total bonded tokens) but their impact is small. If I was going to optimize my transcoder income, this would be the first place I would look into: How would I increase the “self staking” portion?
  2. Reward Cut, Total Stake, delegators’ incentives and transcoders dynamics, I think these four things impact the second portion of a transcoder’s income. Take a look at the sudden increase in the Total Stake charts of 0x481e and 0x4ff0 at round 1264-1265. Both had a drastic increase in the amount of staked tokens (200k+ tokens). If you notice that there is no increase in the “self staking” portion from the Pending Stake chart, it’s clear that both of them received new delegations. So why?
  • Is this because the transcoders are doing important work for the community and some delegators decide to support them? If the transcoders raise the reward cut in the future, would the delegators still stay with them?
  • Maybe the delegators just want to maximize the income and decided to switch from other transcoders which have higher reward cut? And they will likely switch again if there are new transcoders offering lower reward cut?
  • How much halo effect did these two transcoders get by being listed as the top two transcoders on the explorer page?

I certainly don’t have answers for all the above questions but my point is simple: Regardless of all the above variables which are beyond your control, a transcoder is ONLY getting a cut of that portion of the revenue. And if you increase your percentage of cut, you will likely lose some delegation unless your value to the community is NOT JUST providing a stable place for people to stake their tokens and earn some dividends. The pricing war on reward cut will (if not already) quickly reach a point where it stops making sense to any transcoders.

Before I end this post, let me share the income breakdown of our transcoder. Yea, I know our chart looks like a totally bad boy. I have definitely upset some delegators by doing this experiment without announcing it on the forum. See you in my next post.

Here is the link to the dashboard of the transcoder comps. It’s slow right now since we haven’t optimized the data yet.

Disclaimer: I don’t know any people behind the transcoders 0x4ff0 and 0x481e. I picked them only because they are listed as the top 2 transcoders on the explorer page at the moment.