**Average Holding Time Subscore Report** **TL;DR:** This Xerberus Labs paper (March 12, 2025) introduces the Average Holding Time Score, which measures how long wallets hold a token before transacting again, averaged across users ($H(T) = \frac{1}{N} \sum (t_t - t_{prev,i})$). It’s a gauge of token turnover—short times mean fast trading (high liquidity), long times mean holding (low liquidity). Why it matters: it shows user behavior and ecosystem health. High variability (big $\sigma_H$) suggests a lively mix of traders and holders, while low variability with short holds hints at stagnation (e.g., bot-driven activity). Results from Ethereum, Polygon, Cardano, and Base show older chains have longer, varied holds, while newer ones like Base lean shorter, reflecting adoption and vitality. ### **1 Introduction** The Average Holding Time Score is a metric that measures how long wallets hold a given token before transacting again. In essence, it captures the duration between a wallet’s consecutive transactions for a specific token. By averaging this duration across all active wallets, we obtain a snapshot of user holding behavior. This metric provides insight into user activity and liquidity in a token’s ecosystem. A short average holding time typically indicates rapid token turnover – users are frequently trading or using the token, which can imply high liquidity and active participation. In contrast, a long average holding time suggests that users predominantly hold the token (i.e. “HODL”), indicating lower liquidity and a user base with a longer-term outlook. For example, Coinbase observes that a long typical hold time often signals an accumulation trend (investors holding with a long-term view), whereas a short hold time points to increased trading activity (1). Similarly, industry analyses show that tokens with longer holding periods tend to reflect stronger investor confidence and stability, whereas shorter holding periods often coincide with higher volatility and speculative trading (2). This makes the Avg Holding Time Score a valuable indicator of a token’s economic vitality – bridging concepts of token velocity (rate of circulation) and holder sentiment. Notably, some token economic models equate holding time to the inverse of token velocity ($V$), defining $H = \frac{1}{V}$ as the average time a user holds a coin before using it (3). A high holding time (low velocity) implies users are willing to retain the asset, whereas a low holding time (high velocity) means the token rapidly changes hands. By quantifying these patterns, the Avg Holding Time Score helps characterize whether a blockchain network’s token is primarily used as a medium of exchange (fast turnover) or as a store of value (slow turnover), thereby providing a window into the token’s liquidity profile and user engagement. ### **2 Mathematical Definition** Formally, let today’s date be $t_t$. For each wallet $i$ that transacts with the token on $t_t$, we look at the timestamp of its last prior transaction involving this token, denoted $t_{\text{prev},i}$. The Average Holding Time $H(T)$ on day $T$ is defined as: $ H(T) = \frac{1}{N} \sum_{i=1}^{N} \Big(t_t - t_{\text{prev},i}\Big) $ where $N$ is the total number of distinct wallets that made a transaction in the token today. Essentially, for each active wallet we calculate the holding duration (in days, for example) since its previous transaction, then average these durations. This gives the typical holding time between transactions for active participants on day $T$. In addition to the mean, we can compute the standard deviation of holding times, denoted $\sigma_H(T)$, as: $ \sigma_H(T) = \sqrt{\frac{1}{N} \sum_{i=1}^{N} \Big[(t_t - t_{\text{prev},i}) - H(T)\Big]^2} $ This $\sigma_H(T)$ measures the dispersion or variability of individual holding times around the average. It serves as a gauge of how consistent or varied the holding patterns are across different wallets. A high $\sigma_H(T)$ means the holding times of active wallets differ widely – some wallets might hold the token for a long time before transacting, while others transact again very quickly. In other words, a large standard deviation indicates a broad spread of holding durations, reflecting high unpredictability or randomness in user transaction timing (analogous to high entropy or uncertainty (4)). Conversely, a low $\sigma_H(T)$ implies most active wallets have similar holding periods, meaning their behavior is more uniform or predictable. Together, $H(T)$ and $\sigma_H(T)$ give a quantitative summary of how long and how variably users hold a token between transactions on a given day. ### **3 Why it Matters** The Avg Holding Time Score and its variability provide insight into the health and dynamics of a token’s ecosystem. A token whose active users exhibit a wide range of holding times (high $\sigma_H(T)$ and a broad range of $H$ values across users) can be thought of as “lively” and engaged. A high standard deviation in holding times means there is no single dominant pattern of usage – some users might be frequent transactors (short hold times) while others are long-term holders (long hold times). This diversity often signals a broad and active user base with varied strategies (for example, a mix of traders, investors, and everyday users), contributing to a vibrant token economy. In statistical terms, it indicates high unpredictability in individual user behavior, which is characteristic of an active ecosystem with many independent actors (4). Such a token might have ongoing organic activity, news-driven spikes, and participation from different types of holders – all signs of a network with randomized, robust holding patterns. On the other hand, a token where most active wallets have similar (and typically short) holding periods – reflected by a low $\sigma_H$ (low variance) – may indicate a form of stagnation or monotony in usage. If almost everyone transacting with the token today also transacted very recently (e.g., yesterday or within a few days), the average holding time will be low and tightly clustered. This scenario often occurs when the only entities still actively moving the token are those with automated or routine transactions. Examples include trading bots, reward distribution scripts, or backend services that move tokens on a fixed schedule. These automated wallets produce regular, predictable transaction patterns, leading to a narrow distribution of holding times. A low variance and consistently short holding time can thus suggest that organic user-driven activity has dwindled, leaving primarily algorithmic or maintenance-driven transactions. In effect, the token’s circulation might be on “auto-pilot,” lacking new infusions of user interest or spontaneous activity. This is analogous to what has been observed with token velocity extremes: as one industry source notes, extremely low velocity (tokens rarely changing hands) can signal a stagnant market or hoarding behavior, where tokens aren’t being used for their intended purpose and overall ecosystem activity is hindered (5). In summary, high variability in holding times points to an energetic and diverse user engagement, whereas low variability (especially coupled with short average hold times) raises a red flag for potential dormancy in the token’s community. Monitoring these patterns is important because it helps distinguish between a token that is broadly adopted (with many different usage patterns) and one that is only being kept alive by a narrow set of automated behaviors. ### **4 On Chain Results** ![[Pasted image 20250312192441.png]] These histograms show how each chain’s parameter is distributed across its user base. Ethereum and Polygon have slightly higher average holding scores, likely due to being older, well-established networks with broader adoption. Cardano’s distribution lies in a similar range but trends lower overall, reflecting its relatively smaller user base. Base, as the youngest network, has the lowest holding scores and a narrower distribution, suggesting shorter average holding times compared to the more mature chains. ![[Pasted image 20250312192530.png]] **References** (1) [Shiba Inu HODLers Strong: Coinbase Data Shows Longer Hold Times Than Bitcoin and Ethereum – What Does It Mean?](https://bitcoinworld.co.in/diamond-hands-shiba-inu-shib-median-hold-time-on-coinbase-reaches-250-days/) (2) [DeFi Token Holding Periods Analyzed by IntoTheBlock](https://blockchain.news/flashnews/defi-token-holding-periods-analyzed-by-intotheblock) (3) [Token Economics Considering “Token Velocity” - Scott Locklin](https://basicattentiontoken.org/static-assets/documents/token-econ.pdf) (4) [SoK: Measuring Blockchain Decentralization](https://arxiv.org/abs/2501.18279) (5) [Understanding Velocity in the Token Economy](https://nextrope.com/token-utility-balancing-supply-demand-and-velocity/)