Skip to content
Parth Parekh
STAGE 09PROMOTIONEXECUTION PLANE
← BACK TO THE MAP

From a cleared backtest to a position in the book

Most portfolios stop at “backtested”. The interesting half is everything after it: paper on the live order path, then real money at minimum size, then a rollout across accounts, and finally life alongside every other strategy already running.

strategies to live
8
THE SHAPEPROMOTION
FIG. J · STRATEGY LIFECYCLEITERATE1IDEATIONmicrostructure, desk, data2HYPOTHESISa reason to be paid3SPECIFYlegs, entries, exits4BACKTESTnext-bar fill, real costsshelved5WALK-FORWARDtrain frozen, OOS scoredshelved6PAPERthe live OMS doorshelved7LIVEclient capitalTHE GATES ARE THE PRODUCT.Most ideas leave through the side, not the end.Eight cleared every gate into live capital.
The gates are the product. Most ideas leave through the side of this diagram, not the end.
WHAT I BUILTPROMOTION

Built here.

  • 01

    Paper that leaves through the live order path — same routing, same timeouts, same fan-out, with only the fill simulated. A strategy that clears paper has already exercised the machinery capital will use.

  • 02

    A frozen configuration hash: live refuses to start unless its resolved parameters match the set that was tested, so a hand-edited value weeks later cannot quietly void the evidence.

  • 03

    A small-capital stage at minimum size, which is the first place slippage, queue position and partial fills stop being modelled and start being measured.

  • 04

    A comparison of live fills against what the simulator predicted for the same session, so the gap between the two is a number rather than a feeling.

  • 05

    Multi-client rollout through the same control layer the desk already uses — each account sized against its own capital and routed to its own broker, with no strategy-side change.

  • 06

    Portfolio-level accounting: margin netted across strategies rather than summed, exposure and drawdown limits enforced on the combined book as well as per strategy.

WHAT IT SURVIVESFAILURE MODES

Failure modes.

  • An idea that looked good in simulation and does not survive real queue position — caught at minimum size rather than at full size.
  • Parameter drift between what was tested and what is running, which the config hash turns into a refusal to start rather than a silent divergence.
  • Two strategies that look independent in isolation and lose on the same days, which only shows up in the combined drawdown.
  • Capital that is committed on paper but unavailable in practice once margin is netted across the whole book.

A strategy that works alone is not finished. Capital is shared rather than granted, margin nets rather than sums, and correlation only appears once the book is whole — which is why the last gate is the portfolio, not the backtest.

STACKON THIS STAGE
Pythonthe live OMS pathconfig hash freezemargin nettingattribution