Before
Target state could advance immediately after an order call, even when the difference did not fill.
UPDATE / V0.1.1
IMPLEMENTED & VERIFIEDSeparated target weights, order requests, and actual fills to correct overstated performance and introduce retries.
Target state could advance immediately after an order call, even when the difference did not fill.
Suspension, price limits, insufficient cash, and sub-lot differences are not the same failure.
Retryable targets remain in pending_targets and are removed only after a fill or a confirmed non-executable difference.
Formal cost-aware total return moved from 10.286% to 4.473%.
A lower explainable result is more valuable than attractive performance built on invalid execution assumptions.