Shared Capital Allocation: A Two-Sleeve Worked Example

Shared capital allocation: This synthetic example prorates two competing buy requests against one cash balance.

Shared capital means simultaneous strategies compete for one account balance. A standalone sleeve result cannot reserve money that another sleeve has already spent. This original numerical study uses two fictional assets to expose that accounting issue before discussing portfolio returns. It is an offline illustration, not an allocation recommendation or a reproduction of a community strategy.

Shared capital allocation: define the account before the sleeves

The account starts with 1,000 cash units. Asset A costs 10 per unit and asset B costs 20. Both sleeves have signals available before the execution bar. Each requests an 800-unit cash budget. We assume a long-only account, no leverage, zero costs, no cash yield and no settlement restriction. Those simplifications are part of the experiment, not claims about a broker.

Combined demand is 1,600 against 1,000 available cash. In this fixture the predeclared allocation policy prorates concurrent demand. The fill fraction is 1,000 / 1,600 = 0.625. Each sleeve receives a 500-unit budget. A buys 50 units; B buys 25. Cash is zero and marked holdings total 1,000 at the execution prices.

The policy does not decide which strategy is better. A sequential priority rule, minimum lot size or reservation policy could produce different quantities. Choosing one after observing which sleeve rises would create a new, retrospectively selected experiment. Freeze the allocation rule before comparing outcomes.

Mark positions without inventing a sale

At the next observation, A is marked at 12 and B at 18. A's position is worth 600; B's is worth 450. Equity is cash plus holdings: 0 + 600 + 450 = 1,050. The account's simple marked return is 5% over this fixture. Positions remain open. This is not realized sale proceeds, and an actual liquidation could have costs or different prices.

If two standalone simulations each spent 800, their combined positions would require 1,600 initial cash. Their 1,680 terminal value belongs to that larger-capital assumption. Dividing it by the shared account's 1,000 would manufacture a 68% return by ignoring the extra 600 needed at entry. Correcting the denominator alone also does not reproduce the actual shared-account quantities or scheduling.

Keep invariants alongside performance

At every observation verify cash plus marked holdings equals reported equity. Every open sleeve needs a current supplied mark. Reject ambiguous timestamps, repeated order IDs and aggregate demand that exceeds finite numeric representation. Preserve unfilled demand and the declared prioritization rule rather than inventing capital to satisfy all requests.

The released replay sells before prorating buys on each bar. That ordering is a declared model choice. It does not prove proceeds are settled or immediately available at every broker. Its guide explicitly limits the fixture to long-only, no-leverage research; margin, tax and real buying-power rules require separate verified contracts.

Extend one assumption at a time

The companion reproduce.py checks the 0.625 fraction, quantities of 50 and 25, and terminal equity 1,050. Download the standalone numerical checks and run python3 equity-research-labs.py to verify the arithmetic without a Tradewink account. Then rerun a separately versioned scenario with execution costs or declared cash yield. Cost-inclusive buy budgets change acquired quantity, so do not subtract a fee twice or apply it to an unrelated sleeve.

Try Shared-capital replay > Use sample in Advisory Research, then inspect the cash and position rows. Its synthetic example teaches accounting; it does not recommend proration, promise diversification, or compare actual strategies.

QuantConnect portfolio documentation separately displays equity, holdings and cash. The worked allocation, invariants and experiment design here are original Tradewink analysis. Portfolio efficacy remains unmeasured.

Continue the research

Circular Block Bootstrap: Full-Horizon Return Invariance and Intraday Equity Sampling: Audit Short-Hold Risk develop adjacent checks. Use a small synthetic fixture first; these educational diagnostics do not validate a trading edge or change account settings.