Introduction

With the help of the Flipside discord, this query was made a lot easier than it could have been. Identifying all sorts of bots can be quite hard. There are a lot more kinds of bots, other than arbitrage bots, ones who wait long periods of time to execute strategies and maybe have some of the oracle ping-pong strategies outside of a contract. These are harder to track.

For the scope of this bounty, I will be using the query provided as a proxy definition, but this is specific to arbitrage bots.

Bot Swaps vs Average Swaps

The assumptions here are as such:

• Bot swappers: at least 10 or more (>=10) transactions a minute
• Average swappers: at least 2 or fewer (<3) transactions a minute

First let's look at transaction counts, individually and together...

The following make it quite clear that bot swappers are in fact the minority and not the majority. It also seems that there is some seasonality with these bots.

Seasonality vs LUNA price

Let's now look at this potential seasonality vs the price of LUNA, as these bots could very well be arbitrage bots, by definition.

From the above, we can see that there definitely is some correlation between that mid-May dip in price and swap spamming on bot activity, as well as average swappers.

We can better look at their ratios, by looking at the % on the inner join for bot swaps vs average swaps.

We can see that on that date where luna cascaded down in price, there was a ~1.05% in transactions that could be classified as bot transactions, per our definitions. Meaning that as expected our definition has captured, at least some, of the arbitrage bot swaps.

Summary & Evaluation

Given our definition, we can see that there is some seasonality in the arbitrage bot swapping occurring in the network, which is expected. What should be noted is that here, we're only looking at the swaps and not all transactions, we are not even looking at Successful vs Failed transactions, to get an even deeper understanding of what took place.