Read episodes instead of isolated transfers
Cexlens groups verified MM transfers by asset and direction into fixed 30-minute episodes. A repeated transaction is deduplicated, while several desks or exchanges can contribute to one episode. Onto CEX and off CEX describe movement relative to an exchange. They do not prove a sale or purchase.
- Gross flow The sum of verified transfer value inside the selected period.
- Net off CEX Off-exchange value minus onto-exchange value.
- Consensus An episode where at least two verified MM desks moved the same asset in the same direction.
Separate directional follow-through from correlation
Price measurement starts after the final transfer in an episode. The reaction chart shows median directional follow-through at 1h, 4h and 24h with the middle 50% band. For an off-CEX episode, a positive return counts as follow-through; for an onto-CEX episode, a negative return does. Non-BTC assets are compared with BTC when benchmark data is available. The scatter plot keeps the actual signed return and reports Spearman correlation. Neither metric establishes causation.
- Median The middle observed directional outcome, less sensitive to extreme moves.
- Follow rate The share of mature episodes that moved in the expected direction.
- Sample n The number of mature outcomes behind a metric. Small samples are marked as early or insufficient.
Use pace, routes and receipts as one workflow
Start with the asset radar and compare the latest hour with that asset's own previous seven-day hourly baseline. Open Price Reaction Lab for the historical distribution, then check which desks and venues formed the route. Finish with On-chain Receipts, which links the aggregate back to capped transaction evidence. Stablecoins stay in Dry Powder because their price reaction is not comparable with risk assets. History is capped at 30 days and older horizons unlock by plan; the page is research evidence, not trading advice.
Frequently asked questions
Does an MM deposit onto a CEX mean the desk sold?
No. The chain proves movement to exchange infrastructure. The desk may rebalance inventory, provide liquidity, post collateral or prepare to trade. Price outcomes describe what followed, not the desk's private intent.
Why can a reaction metric show n/a?
The episode may be too recent, the asset may not have supported candle history, or the accessible plan slice may not contain enough mature observations. Cexlens does not fabricate a result from an incomplete sample.