Introduction: Thorchain swaps

Thorchain is a decentralised cross-chain liquidity protocol which allows users to swap native assets from different blockchains with the clear mission to facilitate this cross-chain liquidity at the same time as it is censorship resistant.

All Thorchain liquidity pools (LP) consist of a vault which contains RUNE, the native Thorchain token, and a different token, either native or synthetic. Direct swaps between RUNE and each asset are possible as well as asset-to-asset swaps which are treated as a double swaps in the protocol (asset 1 - RUNE, RUNE - asset2). Swaps pay the inbound and outbound fees on each asset blockchain.

Thorchain calculates each of the asset's price as a function of the pool balances - it does not rely on oracles or weighted averages external sources and it does not peg or wrap assets. For this to work as an effective market, arbitrageurs need to compensate any difference in price to reach equilibrium again.

Methodology

This analysis will dig into the bot swap activity in Thorchain. For that, I will define a bot as an address which at least one day of the time period analysed did more than a certain number of transactions. Both the time period in days and the number of swaps to consider a bot will be parametrised, with 90 days and 200 swaps per day as the default values.

The thorchain.swaps table was used for the analysis.

Results

Swaps stats - number of swaps and swapped volume by user type

Figures 1. and 2. show the daily number of swaps by each type of user - human and bot - as a percentage and in absolute value respectively.

Usually, bot activity is around 80% of the total swaps, with some less dominance by the end of March (70%).

In absolute terms, there is a steady increase of activity since the last 90 days and 2 high activity periods: late March - begining of April and begining of May. Humans interacted more similarly in both periods while bots saw the top 4 days by swap volume from May 9th to May 12th - during the Terra crash were plenty of arbitrage opportunities were available.

Figures 3. and 4. show the daily swaps volume in USD by each type of user - human and bot - as a percentage and in absolute value respectively.

Dominance of bots vs humans is slighty lower than by number of swaps - most days between 60% and 80% with a low of 53.9% in mid March.

In absolute terms, a similar upward trend as in number of swaps can be seen. May 9th crash volume was not bigger than in mid April's peak - the latter saw a max daily volume of over 100M USD for bots on April 6th and over 50M USD for humans on April 14th.

Figures 5. and 6. show the cumulative number of swaps and swap volume by user type.

Over 1.13M swaps were done in the last 90 days by bots compared to the 270k done by humans (76% bot average dominance).

3.57B USD were swaped by bots compared to almost 1.6B by humans in the last 90 days (55% bot average dominance)

Bot distribution

There are a total of 44 unique bots as of May 17th (with the default parameters). Let's look at how are they distributed by number of swaps, swapped volume and blockchain.

Figures 7. and 8. shows the distribution of swaps by blockchain, by number of swaps and by volume swapped). Thorchain (THOR) and Binance Smart Chain (BNB) are the most popular chains for the swaps, with THOR having a little more volume and BNB more swaps. Terra is the third blockchain in both metrics despite the events of the last days. This is just one of many proofs of the importance that this projet reached in the whole crypto industry. DOGE has some activity which is irrelevant compared to the other chain's volumes (below 1% for both metrics).

As a side note, THOR blockchain refers to synthethic swaps while each of the other blockchains refers to a swap to or from RUNE into another blockchain's native asset.

Figures 9. and 10. show the bot address distribution by number of swaps and swapped volume. Both distributions are very skewed meaning the bots with more activity have a significant advantage over the rest:

• The top 3 bots by number of swaps account for almost 400k swaps from a total 1.25M, so 1,2% of all bots make 32% of the swaps.

• The top 3 bots by swapped volume account for 1.3B USD from a total 4.35B USD, which means 29% of all volume AND the top bot has more volume that the top 2 and 3 combined.

• Top bot in both metrics is thor1yep703ewakef0h2l9qel93xh96tvkm004pesq7.

Figure 11. shows a scatter plot for number of swaps vs. volume swapped. There are 3 distinct strategies:

• Our top 1 address is an outlier, with over 160k swaps for 760M USD.

• The next 8 addresses range between 150M and 300M USD in volume and 20k and 120k swaps.

All these 9 addresses operate either on THOR or BNB blockchain.

• The rest of addresses swap below 100M USD and 30k swaps.

Figures 12. and 13. show donut graphs for the distribution of volume swapped per strategy and per bot respectively.

• Our mega swapper bot is the only one with more than 300M USD volume and accounts for 21.4% of all volume.

• There are 8 bots with a volume between 150M and 300M USD, accounting each between 4.5% and 8%% of all volume.

These 9 addresses together account for 69% of all volume,

• The low volume strategy (below 150M USD) is followed by 36 different bots. They account for 31% of the total volume.

Conclusions

Swaps in Thorchain in the last 90 days amounted to 5B USD and 1.4M swaps, both by humans and bots. Assuming the default parameters, a total of 44 can be identified with this method. A summary of their activity:

• The top 9 addresses make around 70% of the total volume, each swapping more than 150M USD in total since February 15th.

• The top 7 addresses make around 60% of the total swaps, each with more than 50k swaps since Febraury 15th.

• Bots interacting in Thorchain do more volume in USD while more swaps happen in Binance Smart Chain.

• Average bot dominance is 76% in number of swaps and 54% in volume in USD.

• The top 1 bot swapper by both metrics accounts for 21.4% of the volume and 14.7% of all swaps. He really is the ThorChad!

As an analogy, figure 14. shows the distribution by number of swaps:

• Our mega swapper accounts for 14.7% of all swaps (166k).

• The next 6 bots have more than 50k swaps, weighting between 10.8% and 5.4% of total swaps.

• The remaining 37 bots account for 40.2% of the total swaps, with less than 50k swaps each.

Interestingly, only the mega swapper is a Thorchain address, the next 6 bots are BNB addresses.

Based on these results, I can confirm that most bot swap activity on Thorchain is made by a low number of bots which concentrate both volume and swaps metrics.