How Heron decides

How Heron decides

Every few minutes Heron runs the same check. Understanding it takes about a minute and makes the rest of the product readable.

The question it asks

Does any market pay more than it costs to enter it?

Not "which market pays most". A market paying a headline 12% is worth nothing if the fees to enter, hold and exit come to 13%. Heron subtracts all of it first, then compares. That is why it often does nothing.

What it checks, in order

  1. Is the market alive? A market nobody has touched for months can show an attractive rate that no borrower is actually paying. Heron requires a recent sign of life and refuses stale markets outright.
  2. What does it really pay? The rate after fees, slippage and the cost of the transactions themselves.
  3. Is that enough? If the margin is not clearly positive, Heron declines.
  4. Record the answer. Either way.

Why a refusal is written down

If Heron only recorded the times it acted, you could not tell the difference between "it looked and decided not to" and "it stopped working". Those look identical from outside. Recording refusals is what makes that difference visible, and it is the main thing separating Heron from a yield product that goes quiet.

What you can check yourself

Every decision can be replayed from published inputs, without asking Heron for anything. See Verify a decision.

What Heron does not claim

Heron does not promise a return, and it does not run inside special secure hardware on this deployment. The published record shows what was decided and on what inputs. It does not, by itself, prove that no decision was left out. That limit is real and worth knowing.