Measuring realized Cardano staking yield from cardano-db-sync: epoch rewards over epoch stake, split by pool size

The Cardano Record measures claimed Cardano staking yield from their own cardano-db-sync node. They examine what delegators actually earned for staking, rather than the advertised pool rewards. The key metric is realized yield, calculated by dividing rewards earned by the stake that generated those rewards. This ratio is determined using data from the reward table and epoch stake table in the db-sync node.

Rewards are recorded separately for member rewards (paid to delegators) and leader rewards (paid to pool operators). They are tallied by the epoch in which they were earned, not paid. Member and leader rewards are then divided by the stake active in each epoch to obtain the headline yield figure. To illustrate this, the annualized yield is computed by multiplying the member rewards over delegated stake ratio by the number of epochs in a year, which is 73.

The figures are checked for internal consistency before being broadcast. The segment confirms that rewards add up correctly, stake balances, and the stake bands sum to the total network stake. They also demonstrate that the ratio yields 1.765% annualized for the epoch in question, with the gross yield on all stake being 2.104%. The performance of pools is compared across stake size bands, revealing a wide disparity in yields, from 0.542% for very small pools to 1.797% for the largest pools.

The material emphasizes the importance of distinguishing between pool owners and delegators. Pool owners pledge their own stake but do not receive member rewards. Instead, any return is realized through the leader rewards. Counting owners' stake as delegated stake results in an inaccurate picture, inflating the yield of delegators while understating the owners' true returns.

The segment also addresses other nuances, such as pools that made no blocks, zero-block pools, and the non-monotonic nature of the yield curve across stake size bands. The analysis concludes that while larger pools tend to have higher yields, the data should not be used to advise delegators to allocate their stakes based on size alone.