Anchor Liquidations

    Compare Anchor liquidations with LUNA price over time, discuss any potential correlations

    Loading...

    Sudden & Spikey

    Anchor is a well known market protocol on Terra ecosystem. It allows for borrowing UST in exchange of collateralizing with bLUNA & bETH. If the price of these two assets falls to a certain threshold, a liquidation might be triggered.

    The following is a chart prepared to show LUNA price action and liquidations taking place on Anchor. In it, liquidations are shown in USD equivalent. Only bLUNA is included.

    The time unit for this chart differs from the usual hourly or daily timestamp due to data collection infrequencies. In general, the mentioned data were collected about 1-6 times per day, and should be interpreted as such.

    We can see many sudden liquidation spikes, especially after September 2021, a period when Terra ecosystem was booming and many new retailers were flooding in.

    Studying the Liquidations: Stats & Inferences

    To find out more about the dynamics of liquidations, the data were analyzed in R software.

    In particular, the analysis concentrated on events when more than 1M USD in liquidations took place at a given timestamp. There were 41 of such events or moments. From now on we call these events "The Big Liquidations".

    It is hypothesized that a liquidation takes place when the price of LUNA falls suddenly and quickly from a local maximum price to a lower price. The price collapse should be quick and wide enough to cause big liquidation impact.

    Exploring these dynamics requires the analyst to generate helper tools, and a slight creativity in operationalizing notions like "price collapse". Most recent local maxima in LUNA price before liquidation were to be specified. A new data had to be generated to reflect this, which included some raw and other hard calculated data:

    1. Timestamp of liquidation (raw)
    2. Amount of liquidation in USD (raw)
    3. Price of LUNA at liquidation (raw)
    4. The local maxima of LUNA price (i.e. the most recent high price LUNA was at before the liquidation event) (hard calculated)
    5. Timestamp of the local maxima (raw)
    6. Time difference (in hours) between the LUNA liquidation price and the recent local maxima (hard calculated)
    7. Percentage of difference in the price of LUNA between the liquidation price and local maxima.
    8. Speed of collapse: classic physics (speed=distance/time), it shows how quickly the price is falling (hard calculated)
    db_img

    Indeed, non-Big Liquidations serve as a benchmark here. Take speed of collapse (SoC) for example. During normal times, SoC is -0.36% per hour, meaning that at times of red market, the price falls on average 0.36% each 1 hour. But during Big Liquidations, SoC on average is -1.02%, this is 1.02% fall in price per hour. In other words, at times of big liquidations, SoC is 181% faster than other times.

    db_img

    Other stats of use might be direct correlations. Taking a look at the following figure, we notice:

    • A stronger relationship between Big Liquidations and LUNA price. The higher LUNA is, the more liquidations there are. Basic intuition indicates that LUNA's price has been growing in time, and time has brought many new and inexperienced retailers, ones who were easier to fall for a liquidation crisis. Another issue is related to volatility; it might be that as LUNA's price made a big run recently, so did its volatility in the market, which can make sudden ups and downs more likely than before, causing more liquidations.

    • A correlation test between LUNA price and Liquidated USD amounts was conducted using the full liquidation data (without separating between Big and non-Big liquidations). The correlation was significant, with a coefficient of 0.1072. In other words; LUNA's price and liquidated amounts are in a positive relationship, when one grows, so does the other.

    • Speed of Collapse is an indicator significant to all variables in the Big Liquidations data, but was less important in the non-Big Liquidations set. This tells that it might be good in discriminating and summarizing the different aspects in play during liquidations in particular.

    The Big Liquidations had peculiar stats, taking means, the following was found:

    • Average amount_liquidated USD: 5433082.055
    • Average LUNA liquidation_price: 22.43284185
    • Average recent_local_maxima: 26.18820977
    • Average % fall in price from recent maxima to liquidation price: -16.188605 (%)
    • Average time_diff that spanned the fall from maxima_to_liquidation_price (in hours): 48.04878049
    • Average speed_of_collapse (% per hour): -1.029474 (% per hour)

    If we take a look to non-big liquidations (i.e. small liquidations and no liquidations), we find the following:

    • Average amount_liquidated USD: 33454.3
    • Average LUNA liquidation_price: 20.751
    • Average recent_local_maxima: 21.952
    • Average % fall in price from recent maxima to liquidation price: -5.945 (%)
    • Average time_diff that spanned the fall from maxima_to_liquidation_price (in hours): 33.3
    • Average speed_of_collapse (% per hour): -0.3663 (% per hour)

    I'll throw a picture to depict these relationships for quick reference.