LOGRETTA

The architecture I am developing separates observation, interpretation and action. Data supplies market context. A model proposes a candidate decision. Risk determines the conditions under which it can be considered. Trade management follows orders and positions. Records preserve the observations that allow the whole sequence to be reviewed.

The wiring matters as much as the model. A decision should remain connected to its input context, configuration and account state, so it can be examined later. An execution record should distinguish an intended order from an acknowledgement, a fill and a confirmed position. That separation makes the system easier to reason about when information arrives late, a component fails or an account observation changes.

A signal is only one part of a trading system. Position sizing, available margin, total exposure, trading costs and drawdowndrawdowndrawdownA decline from a previous high. This portfolio’s dashboard calculates it from the selected period’s running closing-balance peak; daily closing observations do not describe intraday losses or prove risk compliance.× boundaries determine the conditions in which a candidate trade can be considered. My research treats risk as explicit operating constraints, with the account and the quality of the available information visible at the point of review.

Trade management follows the full lifecycle: intent, submission, acknowledgement, partial or complete fills, open-position observations and closure. The architecture needs to account for rejected requests, stale information and interrupted sessions. An exit also has a different purpose from an entry: reducing an existing exposure must remain distinguishable from proposing a new one.

The example below keeps one candidate fixed while its account context changes. Missing information, an older observation and a recent record from another account each answer the review question differently. The record has to retain why it was held, rather than making uncertainty disappear when the view changes.

The workspace offers Standard, Prop and Research views over the same sample ledger. Standard pairs closing balances with performance and recent trades. Prop opens the risk view with illustrative daily-loss and drawdown limits. Research replaces reference limits with observed risk and adds a distribution of trade outcomes against initial risk.

Changing the reporting period updates the balances, statistics, charts and journal together. A chart observation leads to that session’s recorded trades; a day with no closed trades remains visible as a carried-forward balance. The journal can be filtered, sorted and exported. Expanding, reopening or returning from the compact still preserves the selected controls. These interactions use invented data within the portfolio, with no connection to a broker or live account.

The interface connects headline measures to a closing-balance curve, monthly net profit and loss, trade outcomes, instrument attribution and the trade journal. Each report reconciles to the same sample ledger. Commissions are deducted from every trade; win rate, profit factor and expectancy use those net outcomes.

Sharpe compares average differential return with the variability of those differential returns. The benchmark, observation interval and assumptions behind annualisation matter. Historical ratios do not establish future returns; the review also needs drawdown, trading costs, outcome distribution and the size of the observed sample.