Introduction

Data shows that the success rate of transactions on Terra is declining (see the graph below).

In investigating the causes of this issue, some point the finger of suspicion at bots. These commentators argue that the expansion of the UST market has increased bot activities on Terra network which in turn led to a lower success rate of the network overall. The vital role of LUNA-UST burn-mint mechanism in stabilizing UST, as Terra native stablecoin, incentivizes arbitrageurs to be involved in and paves the way for bot activities on Terra. The assumption behind this argument is that bots' success rate is lower than the average user success rate.

This dashboard tests the above mentioned assumption by comparing the success rate of bot wallets with the other wallets.

Definitions, method, and data

Bots are defined as wallets that operate by software programs. These software programs execute functions, in most cases for trading, using artificial intelligence.

These wallets very rarely execute transactions manually or by human actions.

In this dashboard, we distinguish bots based on the number of transactions per day. Bots can be varied in terms of the daily number of transactions. Therefore, some specific thresholds of activities are used to categorize wallets within the spectrum of activities. Activities range from solely human administrated wallets to purely highly active bots.

In the appendix the thresholds and details of categories can be found.

Appendix

Categories

• '0 low_active_user': n_txns > 0 AND n_txns <=10
• '1 mid_active_user': n_txns >10 AND n_txns <=50
• '2 high_active_user': n_txns >50 AND n_txns <=100
• '3 bot-M': n_txns >100 AND n_txns <=500
• '4 bot-L': n_txns >500 AND n_txns <=2000
• '5 bot-XL': n_txns >2000

note1: n_txns = number of transactions in a day.

The graph below shows the distribution of wallets in terms of number of transactions per day.

note 2: wallet-day is an independent unit of analysis. It should not be confused by wallets.

Analysis

The graph below shows the daily average success rate per category. We can see that the group with the highest number of transactions in the day (bot-XL) has the highest success rate overall. The second higher success rate is for low active users. Users with less than 10 transactions in the day.

Here we can also see the scatter plot that shows the pattern differently. The highest and lowest transactions rate lead to highest success rate.

We can see that the lowest success rate is most likely happened for medium and large number of transactions bots.

Conclusion

The graph below summarizes and shows the average success rate at aggregated level. In general we can see a U-curve in relationship between number of transactions and success rate.

Bots with the medium level of number of transactions have the worst experience in terms of success rate.

Assuming all wallets do in average the same actions or operates the same functionality, the highly active (perhaps more sophisticated) bots as well as cautious low active humans have the highest level of success rate while bots with the medium level activities have the lowest success rate.

Further studies should look into the activities and protocols.