What thirty years of working reserves taught us about building a trading system

Petroleum reserves are a strange kind of number. Nobody has ever seen a barrel of proved reserves. The figure is an estimate, bounded by rules, that a professional signs and a company files. If it is wrong in the optimistic direction, the correction arrives later, in public, with a name attached.

Three habits keep that number honest, and when we started building SUTP we found ourselves installing all three.

1. Reconcile the movement, not the balance

A reserves close does not ask “what are the reserves?” It asks “how did we get from last year’s number to this year’s?” Production, revisions, extensions, acquisitions and divestitures each have to account for their share of the change, and the total has to tie.

SUTP’s ledger works the same way. We do not ask whether the book’s position looks right. We reconcile every fill, assignment, rename and corporate action to the broker’s own record, and when the movement does not tie, the symbol is held and the operator is alerted. A balance that merely looks right is not evidence.

2. The category is a claim about the future, so it needs a rule

Proved, probable and possible are not adjectives. Each has conditions, and the conditions decide what can be booked and what must wait. A proved undeveloped location that has no development plan within five years is not proved, however good the geology looks.

A strategy in SUTP is classified the same way. It is a trial until it has a full quarter, matched dates, an attested data feed and overfitting-aware statistics that survive the test. Until then it may run, but it cannot be promoted and it cannot grow its size. The verdict is a rule, not a feeling.

3. Someone signs, so someone must be able to stop

A reserves engineer can refuse to book. That refusal is the whole value of the role. In an automated system the equivalent is the kill switch and the breaker that latches until a person clears it, and the authorisation chain that stands between a paper book and live capital.

We would not put our name on a reserves report whose numbers we could not stop. We will not put it on a trading platform whose orders we cannot stop either.

Why this matters to a counterparty

An allocator evaluating a systematic strategy is in the auditor’s chair. The question is not whether the backtest is attractive. The question is whether the process that produced it would catch its own errors, and whether anyone has the standing to say no. That is the discipline reserves work demands, and it is the one we brought with us.