Anomaly Detection is a service that helps identify which trades were abnormal for a specific wallet in relation to its entire transaction history.
The service detects 7 types of anomalies:
This type of anomaly looks at the frequency of trades within specific time intervals: from midnight to 4 AM, from 4 AM to 9 AM, and so on. If a wallet usually doesn’t trade during a certain time period but starts doing so suddenly, it is considered an anomaly.
This type of anomaly examines the gas fees incurred by the wallet during trades. If the gas fees for a particular trade were significantly lower or higher than usual, it is considered an anomaly.
Max priority gas costs:
This type of anomaly examines the costs of accelerating a transaction (such as a "swap") incurred by the wallet. If the acceleration costs for a specific swap were unusually low or high, it is considered an anomaly.
PNL:
This type of anomaly looks at the change in the wallet's PNL after a trade. If the PNL increased or decreased significantly following a particular swap, it is considered an anomaly.
Trade volume in dollars:
This type of anomaly examines the trade volume in dollars. For example, if the wallet usually executes swaps worth ~$1000, but suddenly trades for $10,000 or $10 (a significant volume change in either direction), this is considered an anomaly.
Spread to trade volume ratio:
This type of anomaly looks at the ratio obtained by dividing the spread (the percentage difference between the buy and sell prices of a token) by the trade volume in dollars. A significant change in this ratio in either direction is considered an anomaly.
Gas to trade volume ratio:
This type of anomaly looks at the ratio obtained by dividing the gas fees incurred during a trade by the trade volume in dollars. A significant change in this ratio in either direction is considered an anomaly.
To use the Anomaly Detection service, go to the "Anomaly Detection" tab within the wallet analysis section:
Once you enter this tab, you will see the settings panel and the default display of anomalies for the first automatically selected type:
Let’s take a closer look at the buttons:
The most interesting part is the chart. Let's look at an example:
At the top of the chart, there is a "legend":
Great! Now we have two time intervals displayed:
Next, in the chart’s legend, there is a switch "Show DEMA values" – this button enables the display of a line around which the confidence interval is built.
The confidence interval is a range within which the parameter is allowed to fluctuate. In the image above, the confidence interval is the pink area around the red line.
If a parameter (in this case, the number of trades executed within a specific time interval) goes beyond the confidence interval, it is considered an anomaly.
Anomalies are displayed as red dots on the chart – by clicking on one, the transaction list will scroll, and the anomaly will be highlighted in blue:
Now you have a convenient tool for simplified analysis of wallet trading history – nothing will escape your attention!