While most existing analytical models treat account interactions as a network and examine it from a static and global perspective, they neglect the exploration of dynamic and micro-level account characteristics. Therefore, we have conducted pioneering work in this chapter to delve into these aspects of various account types on Ethereum, such as exchanges and phishing entities. Our research involves describing and comparing the trading dynamics of these accounts. Subsequently, we create trading ego networks for each account and carefully examine their microscopic characteristics. The experimental results reveal the unique characteristics of different account types in terms of transaction characteristics and ego network properties, revealing their respective roles. It is worth noting that there are obvious differences between normal accounts and illegal accounts in terms of transaction neighbors and interaction patterns. Additionally, our observations indicate that criminal gangs may be involved in phishing schemes. Based on the conclusions of our account analysis, we designed a variety of account features for classification tasks. Experimental results confirm the utility of our proposed dynamic and micro-features in differentiating various account types. We believe our analysis will provide valuable insights into account classification efforts within Ethereum as well as other blockchain platforms.