01Benchmark & risk
Target weights, the eligible universe, and the supplied covariance or factor model.
Active weights, tracking objectives, and the agreed exposure limits.
Wealth & asset management / Direct indexing
Construct individual equity portfolios around a common benchmark—with each account’s holdings, tax lots, and investment restrictions in view.
For advisors, RIAs, custodians, and wealth platforms serving individual investors.
Retain part of an appreciated position. Review the remaining benchmark deviation alongside the permitted gain realization.
Illustrative weights and choices. No live optimization or performance result.
Account-aware portfolio construction
A model allocation is only the starting point. An existing portfolio brings embedded gains, restricted securities, cash flows, and a different path toward the investment objective. PRISM brings these inputs into a defined optimization problem.
Target weights, the eligible universe, and the supplied covariance or factor model.
Active weights, tracking objectives, and the agreed exposure limits.
Acquisition lots, cost basis, gain budgets, and supplied tax rules.
Proposed lot changes and the balance between tax cost and portfolio alignment.
Security exclusions, position bounds, retained holdings, and permitted replacements.
Whether the proposed holdings satisfy the account’s investment policy.
Contributions, withdrawals, cash targets, turnover limits, and transaction-cost assumptions.
Proposed purchases and sales, cash balance, and implementation costs.
The supported model, tax treatment, and any cross-account rules are agreed during evaluation.
From model to portfolio review
Connect portfolio inputs to proposed rebalances and retain the context needed to review each account.
Load holdings, lots, cash, and the current mandate. Identify incomplete inputs before evaluation.
Evaluate the supplied target and constraints together, within the agreed computation window.
Review proposed holdings, lot changes, portfolio exposures, and constraint checks.
Pass the reviewed output to your portfolio workflow with its input and evaluation record.
Your investment process stays in view. Define the data contract and review steps with your portfolio, risk, and engineering teams.
Explore the APIEvaluation on your account universe
A useful evaluation includes accounts that are difficult to rebalance, as well as straightforward ones. Agree the reference, quality criteria, and timing boundary before measuring throughput.
Read the benchmark methodologyInclude embedded gains, concentrated positions, security restrictions, and cash events.
Check the same objective, tax assumptions, and feasibility criteria against an agreed reference.
State the account count, universe size, hardware, and included processing steps with every timing result.
The account example illustrates construction choices and does not calculate a portfolio result. Published benchmarks distinguish generated numerical workloads from strict economic-parity evaluations. Model availability, data requirements, and acceptance criteria are established during a scoped evaluation.
Historical fleet-study scope: the 500,000-account PRISM wall time was measured on one GPU. The whole-book CPU comparator timing and resulting ratio were extrapolated from a measured 2,000-account sample. This study is separate from the generated common-QP and full-workflow strict economic-parity records. Harvestable tax budget is not realized customer tax savings.
Direct indexing / Evaluation questions
Direct indexing constructs an account from individual securities around a reference index or benchmark. Individual holdings make it possible to reflect account-specific restrictions and tax considerations alongside the desired market exposure. The resulting portfolio can differ from the benchmark because those constraints matter.
Two accounts can share an investment objective while starting with different holdings, cost bases, cash balances, and restricted securities. A proposed rebalance therefore needs to consider the account’s starting point as well as its target. The illustration above shows how these inputs can change a proposed allocation.
Agree the benchmark, risk model, account data, and constraint definitions first. Review tracking error, turnover, tax-budget usage, and feasibility, then measure runtime across a representative set of accounts. The published engine benchmarks describe their own test conditions; they do not establish an individual account’s after-tax outcome.
Work with Asymmetry