Trenches

Onchain markets make a remarkable amount of activity visible. You can follow transfers, inspect holdings, and watch a new market form almost transaction by transaction. I built Trenches because that visibility creates a difficult research problem: there is much more to observe than there is to understand.

A cluster of wallets buying the same asset might reflect independent conviction. It might also reflect a shared source of funds, coordinated activity, or several addresses belonging to one participant. Treating every wallet as a separate vote gives the data a meaning it has not earned. Before asking whether an activity pattern predicts price, I want to know what that pattern actually represents.

That leads naturally to relationships. Where did the funds come from? How concentrated is ownership? Do the same groups repeatedly appear together? What happened after earlier instances of similar behaviour? A useful view of a market needs to preserve those connections. A list of popular tokens or profitable addresses leaves too much of the explanation out.

Trenches brings discovery and wallet research into a workflow where an observation can become an explicit hypothesis. I want to state what I think will happen, identify what would make me change my mind, and keep the record as the market develops. Otherwise, it is too easy to remember an idea in the form that would have worked.

The next question is whether the observation leaves room for an executable trade. A price move visible in a historical chart may have happened before a researcher could act. A quoted price may be available for only a small amount. Entering a position changes the problem again: the exit depends on the size actually held and the liquidity available when it is time to leave. These details decide whether an apparent opportunity can survive contact with a market.

I use paper portfolios and replay to study that gap. The aim is to revisit a decision with the information and timing available at the time, including the delay between noticing something and responding. Losing ideas belong in that history too. A research process that quietly drops them would give a flattering picture of its own judgment.

The project spans Solana, Base, and Robinhood-related market research, with different levels of evidence available in each. I treat gaps in coverage as part of the observation. An empty view cannot establish that nothing happened. The same discipline applies to a paper result: it is useful evidence about a hypothesis, with further questions still to answer about real execution.

My interest in Trenches is ultimately about how conviction is formed. Public activity can help explain a market, provided the researcher is willing to question the apparent independence, completeness, and tradability of what they see. The work begins with a transaction, but the useful part is the reasoning that follows it.