An agenda for on-chain and crypto market structure research. These are proposed study protocols—not completed studies, live feeds or trading signals.
Research status
Questions first. Findings only with evidence.
No data collection, external API or new market result is connected to this page. Sources must be selected, checked and recorded before any study claims are published. Existing source-linked briefings remain a separate format.
01 / On-chain flows
Does an exchange inflow tell us what happened next?
Proposed study: compare timestamped exchange-labeled transfers with later market behavior, without treating a transfer as proof of a sale.
Required evidence: chain, block time and finality, transaction reference, address-label provenance and when that label became available.
Alternative explanations: internal transfers, custody changes, collateral movements and incorrect labels.
Test design: predefine event windows, baseline and exclusion rules; preserve null and adverse findings.
Stop condition: do not infer investor identity or intent when the data cannot support it.
Protocol only · no dataset or conclusion published
02 / Derivatives structure
What does leverage add beyond the chart?
Proposed study: examine open interest, funding and spot/perpetual differences as context—not as stand-alone direction calls.
Required evidence: venue, contract specification, funding interval, timestamp, unit conversions and data coverage.
Test design: specify a hypothesis and benchmark before observation; account for funding, fees and slippage.
Stop condition: open interest measures outstanding exposure, not who is right or which way price must move.
Protocol only · no live derivatives feed
03 / Liquidity and execution
Can the quoted price survive a real fill?
Proposed study: compare spread, depth and execution assumptions across venues, with centralized order books and on-chain pools treated as different mechanisms.
Required evidence: venue/pool, instrument, timestamp, order size assumptions, quoted versus executable depth and fee schedule.
Costs: spread, price impact, slippage, network costs where relevant and failed execution assumptions.
Test design: keep paper/hypothetical and actual fills separate; record stale data and missing intervals.
Stop condition: visible depth is not a guarantee of liquidity when an order arrives.
Protocol only · no execution engine
A reusable research note
Question, hypothesis and what would disprove it.
Source URLs, retrieval time, data permissions and coverage gaps.
Method, sample, baseline, costs and excluded observations.
Findings, including negative outcomes—or an explicit “not tested” label.
Limitations and the next question, without a buy/sell recommendation.
AI may be proposed to help organize sources. It must not generate missing observations or substitute a confident summary for evidence.
