Coin Metrics Rebuilds Ethereum Flow History, Alters Outflows

Coin Metrics recomputed Ethereum’s Standard Flow Metrics from the network’s first block using updated exchange-wallet mappings. The company posted a completion notice on Oct. 1 at 17:04 UTC after announcing the recomputation plan on Sept. 28. The rebuild covers all ETH Flow Metrics at daily and hourly frequencies and corrected history is available for backfilling. The notice did not include aggregate revision amounts or side-by-side strategy comparisons.

The provider’s Standard metrics assign transfers to exchanges using all addresses currently known to belong to those entities, and each address’s history begins at its first nonzero balance. When additional exchange addresses are identified later, past Standard values can be restated. Coin Metrics’ Point-in-Time (PIT) series instead uses the set of addresses known to belong to an entity during each historical interval; later discoveries do not rewrite earlier PIT intervals.

A chart downloaded today can show a different pattern of past exchange outflows than a user would have seen at the time. A backtest that replays a trading rule must use the information that would have been available at each decision point. Coin Metrics’ notice emphasizes the need to identify the data vintage-the specific version of the data used in any test-while noting that measuring any effect on returns requires separate analysis.

Other data providers document similar limitations. CryptoQuant’s documentation warns that its ETH Exchange Flows endpoint does not support PIT accuracy and that historical values may change as exchange wallets are discovered and validated through periodic clustering updates; CryptoQuant runs automatic updates weekly on Tuesdays at 00:00 UTC and notes recent observations can change. Glassnode provided a March 13, 2026, hypothetical backtest using Binance’s BTC exchange balance in which a moving-average rule produced different results when run on PIT balances versus revised balances constructed with later knowledge.

Analysts who want to measure the impact of Coin Metrics’ rebuild need paired observations from the same provider and metric with matching exchange coverage, time intervals and timestamps. For the revision question, that requires a retained copy of pre-rebuild Standard values alongside the post-rebuild Standard history. The trading rule must remain fixed across comparisons, and a replay must model the information availability, publication timing, execution timing and trading costs a trader would have faced.

Availability and publication timing add further constraints. PIT history exists only from the date tracking began for each metric; a metric added later does not retroactively gain earlier PIT observations. Some providers record computed_at timestamps that show when a metric was computed and when it was published; without those fields reconstructing what a trader could access at a specific time is harder. Coin Metrics’ Oct. 1 notice marks the rebuild’s completion but does not by itself date every affected value’s original availability.

Revised historical balances can be used to study past supply movements with today’s address coverage. Demonstrating that exchange outflows offered a tradable edge requires reproducible inputs, documentation of when metrics and address mappings were available to users, and replayed trades built from those contemporaneous inputs.