Diamond Hands Lightning Routing Report(Jan,2022)

TL;DR

1. The Diamond Hands node’s transaction volume has grown about 30% monthly on average since April 2021 and it now routes more than 10 BTC/month.
2. The median transaction size for the last 3 months is 47,298 sats and the average size is 103,962 sats. Many transactions are so-called rebalance transactions, thus this does not necessarily reflect the true average transaction size for Lightning payments.
3. The median fee rate for the last 3 months is 50ppm (0.005%) while the average fee rate is 132ppm (0.0132%). There are mainly low (1 to 10ppm), medium (100ppm) and high (1000ppm) transaction categories for our node depending on the opposite nodes’ characteristics.
4. Optimal fee rates depend on each node and its strategy, but the median fee rate seems to be in decline over time. This may be due to the transaction demand of LN not keeping up with growth in liquidity.
5. 21.3% of the Diamond Hands’ adjacent nodes have not routed any transactions for the past 3 months (inactive) and close to 80% are not making meaningful contributions to routing. The inactive ratio is particularly high for us as we incentivize beginners to open channels to us, but we suspect the majority of new Lightning nodes on the network are also struggling to route transactions regularly.

Diamond Hands is the largest and most active Lightning Network user and developer community in Japan. The community started in June last year and has grown quickly, exceeding even my own personal expectation significantly.

Now, there are about 300 Lightning nodes, mostly new node operators, who have joined us and we are also sponsored by several top crypto companies in Japan to onboard more users and educate companies alike.

Our current enterprise sponsors. We use the sponsorship fund to give away sats to participants.

The most significant factor that helped bootstrap the community in Japan for us was to give out sats rewards to nodes who open direct channels to the Diamond Hands(DH)node. These sats rewards were initially based on channel sizes, or in other words, how much inbound liquidity they provide to us. This also helped the Diamond Hands node to improve its presence on the network and route more transactions.

In this article, we share some statistics of our routing experiment so far. This is our first attempt, and we intend to publish more statistics and analyses in the future. We hope this will interest other node operators and/or more people to recognize our activities in Japan.

Transaction volume growth

Here is the routing volume by month from April 2021 to Dec 2021. Since May last year, we have been growing more than 30% month to month on average and routed just about 10BTC in Dec.

We saw the first spike in trading volume in July as we announced our routing incentive plan in late June and many Japanese node operators have opened channels to us since then.

Transaction size distribution

Next, here is the distribution of transaction sizes for the last 3 months from Oct 2021 to Dec 2021.

Evidently, most transactions are in the 10k to 100k sats range and we have few transactions over 1 million sats (0.01BTC).

This, however, does not necessarily represent a typical transaction size of organic Lightning payments as many of the transactions are smaller, artificial rebalancing transactions.

Also, despite my suspicion that transaction sizes are getting larger over time on the LN, I couldn’t find strong evidence for this theory. This may differ for other nodes but nonetheless, we will continue to monitor the shift in transaction size in future.

Fee rate distribution

Next, the DH node’s median transaction fee rate for the last 3 months was 50ppm (0.005%) while the average fee rate was at 132ppm(0.0132%). The chart below indicates that we have 3 rough fee categories.

• Low (0 to 10ppm): Some wallets and payment processors
• Medium (100ppm): Rebalance nodes
• High (up to 1000ppm): Exchanges

I will likely dig deeper into these categories and their trends respectively in the next article.

Also, interestingly, the weekly median fee rate is on a downtrend for us. This may indicate that demand for lightning transactions is not keeping up with the rapid growth of liquidity since last year.

To be clear, adequate fee rates for each node or channel depends on its capacity, connectivity and strategy and there is no single “optimal” fee rate for everyone.

The above chart maps a trend for the Diamond Hands’ node, for which the strategy is to maximize routing volume while adjusting fee rates to organically balance capacity without manually rebalancing it.

With that said, we argue that adding even more liquidity to the network now may be suboptimal. Routing profitability will eventually depend on how much and how strong the demand is for LN transactions. Active routing nodes actually have a strong incentive to help onboard more users and companies. I think this is an interesting dynamic for routing and lightning adoption.

Regardless, monitoring changes in median/average fee rates over time will likely help measure the demand for Lightning transactions and profitability of routing in general. We will keep an eye on this and dig deeper in future analyses.

Adjacent nodes’ activeness

The next item is node activeness, that is, how many of our connected nodes are active participants in routing.

I have summarized the number of transactions for each node in the last 3 months (Oct-Dec2021) and categorized them in the following manner.

(The breaking points are rather arbitrary but we will assume it is good enough for this purpose)

• Active: More than 100 (more than 1 tx /day on average)
• Normal: 51 to 100 (several transactions per week)
• Barely alive: 11 to 50(up to 2–3 transactions/week) Zombie: 1 to 10(almost no transaction)
• Inactive: 0 (No transaction. Could be offline or closed channels with us) The result is the following.

Out of the 473 nodes on the list, 103 (21.7%) are ‘Inactive’ and 180 (38%) are ‘Zombie’. If we include ‘Barely alive’, close to 80% of them are not contributing to routing, unfortunately.

This is likely a very high inactive ratio compared to other nodes due to the aforementioned incentive experiment we are running to encourage the participation of small, beginner nodes. At the same time, I suspect it is also true that most nodes, especially small-scale nodes, route very few transactions, not to mention being profitable. Routing is still a very new and immature field but if this trend continues, I’m afraid most small individual nodes will eventually be driven out of the market.

Being connected to many inactive nodes slows down the DH node’s performance and can hurt our node scoring as well, so we intend to mitigate this issue soon by helping our community nodes with automated fee adjustment setup and node selection. We will revisit this data again in a few months.

What’s next?

Future areas of further analysis:

• The effect of cleaning up inactive channels

We will close inactive channels and test whether it improves the node’s performance.

• Transaction volume share by node

We will identify which nodes are the best partner nodes for us and how it is changing over time.

• Volatility in routing volume
• Estimating optimal channel sizes
• The relationship between node capacity and routing volume

Is opening channels to large nodes really the best strategy?

• Routing trends across different categories (Exchanges, Payment processors, Wallets, Lapps, etc.)

We are still new to the routing endeavor but we will continue to work with the Lightning community in Japan and aim to eventually establish a strong presence in the routing ecosystem.

In addition to onboarding users and companies, we also have worked on a small experimental project together with developers in the community. It’s a channel lookup/scanner where you can pay the DH node directly and check the real channel capacity.

Our hypothesis is that a tool like this will help serious routing operators with routing strategies and rebalancing.

We may add more features such as a routing volume scan feature in the future or pivot to build something else. Either way, you can test it out already on this website and look up our capacity details by channel. We appreciate any feedback or suggestion.

Happy routing!