The current estimation system faces challenges from inaccuracies caused by ignoring transactions with unconfirmed parents and assuming confirmations are due solely to fee rates. It has been suggested that if adequate data can be obtained from standalone transactions in the mempool, these could provide reasonable fee estimates without considering every transaction's specific confirmation cause.

There is an approach where tracking only transactions without in-mempool descendants may currently suffice but could change with an increase in package transactions. To manage the complexities of transaction fee estimation, transactions are segmented into chunks with assigned feerates based on processing time. A timestamp-based system adjusts chunk feerates throughout the day, emphasizing the importance of using the latest update time for mempool duration.

MapMempoolTxs plays a crucial role by measuring the time transactions spend in the mempool, providing insights into transaction evaluation and prioritization. Furthermore, in scenarios involving Child Pays For Parent (CPFP) and Replace-By-Fee (RBF), it is recommended to consider the fee rate and mempool duration of the youngest transaction in a chunk for accurate estimations.

The email proposes examining each transaction's chunk feerate individually to optimize validation and scalability. Upon the arrival of a new block, it suggests removing matched chunks from consideration and updating the fee estimator with evictions, accepting mismatches due to different mempool compositions.

Finally, the CBlockPolicyEstimator is being designed to improve accuracy in predicting transaction confirmations. To address its limitations, there is a proposal to use a "chunk mining score" that updates each transaction's potential for mining upon entry into the mempool and when it improves. This score guides the estimator in disregarding non-tracked transactions as failures and evaluating successes based on the sponsor transaction within a chunk. This novel tracking method might enhance the precision of the CBlockPolicyEstimator's forecasts.