ALMAZOV TRADE
ALMAZOV TRADE
AccountDocumentationContact the managerEN/RU
← Research & Development

Execution engineering

Contracts, partial exits and actual execution

A theoretical position size must become an order the venue can accept. After a partial exit, both the remaining size and the quantities of later orders need to stay consistent.

A continuous model meets discrete contracts

In December 2025, our C++ position calculations were converted to contracts. The work exposed minimum-size restrictions and the interaction between the available position and the number of partial exits. Rounding was also corrected in a Python implementation.

The same nominal percentage can produce a different executable quantity when the contract step changes. If the desired exit is below one step, the mathematical intention and the order that can be placed are different objects.

The remaining position changes the next calculation

A July 2026 development discussion examined what happens when a target is filled after other exit orders have already been calculated. A percentage of the original position and the same percentage of the remaining position produce different order sizes.

The model below makes that distinction explicit. It uses an abstract position and sequential fills, not the parameters of a current trading algorithm or the specifications of a particular exchange.

Execution model · abstract units

Two exits, two possible sizing bases

Initial position100 units
Quantity step
First fill
40
Second fill
10
Final remainder
50

Second order requested: 18 units. An abstract position of 100 units. Each fill is capped at the remaining position and rounded down to the chosen quantity step. This is an explicit model assumption, not a venue specification.

Execution differences need their own investigation

The working history also contains comparisons of slippage across instruments and discussions of how entry and exit slippage contribute differently. These observations are a reason to examine fills and costs for the particular instrument and model version.

They do not establish that every historical tester contains a complete slippage or funding model. A numerical performance report needs to identify which fees and execution assumptions were actually used.