PRISM / Portfolio optimization infrastructure

Personal portfolios.
Platform scale.

Built for advisors, RIAs, custodians, and wealth platforms.

Put portfolio computation behind your client experience. Evaluate account-specific holdings, tax lots, risk, and trading costs across a shared investment strategy.

Evaluate your platform workload See portfolio personalization
Direct indexing: own the stocks, manage individual holdings, and seek index-like exposure. An ETF opens into individually owned stock bubbles. Holdings settle into individual positions and resize as weights adjust. Multiple eligible loss-bearing lots are sold and different securities replace their exposure after review. Dashed steel index and cobalt portfolio lines fluctuate together, with the latest segment magnified. All data is illustrative, not a performance forecast. Playback controls are below the image. FUND SHARES Index ETF / mutual fund Your stocks. Your rules. Loss TAX-LOSS HARVESTING Stock-level choices Exclude Adjust weights LOT-LEVEL DECISIONSHarvest. Replace. Review.Eligibility, risk and wash-sale checksIllustrative · no realized tax benefit impliedLoss-bearing lot under review Seek index-like exposure INDEX EXPOSURE Follow the index LATEST WINDOW Index Portfolio INDEX & PORTFOLIO RETURNS Illustrative paths · before tax & fees Monthly return differences · their variability is tracking error COLOR = SECTOR · SIZE ≈ WEIGHT FACTOR ALIGNMENT Index target WASH-SALE CHECK Same / substantially identical −30 daysSale+30 days Screen linked accounts
Illustrative · before taxes & fees · *replacement eligibility assumed after reviewRead the workflow and assumptions
ONE STRATEGY → INDIVIDUAL HOLDINGS → INDEX EXPOSURE

Illustrative account personalization and benchmark exposure.
No live trades, measured returns, or realized tax benefits are depicted.

For the platform. Through to the client.

Your investment experience.
PRISM underneath.

Integrate portfolio optimization into the workflows through which you serve investors. Keep the investment mandate, review process, and client relationship in your platform.

  1. Your platformInvestment inputs

    Models, risk inputs, holdings, and account restrictions.

  2. PRISMPortfolio computation

    Construct and evaluate proposed allocations against the supplied objective and constraints.

  3. Your client accountsReview & implementation

    Return proposed weights and the available execution record to your workflow.

Integration scope and supported inputs are established during evaluation. Proposed outputs remain subject to your review and implementation controls.

01 / THE PERSONALIZED BOOK

The rules are personal.
The computation scales.

Investment platforms own the strategy and the client relationship.
PRISM supplies the decision layer underneath.

01.1 / DIRECT INDEXING

One book.
No two accounts alike.

Restrictions, cash flows, tax positions. Each client brings a different set of conditions to the same investment objective.

Explore direct indexing
01.2 / TAX-AWARE DECISIONS

The opportunity
is in the details.

Look inside the holdings. Evaluate eligible lots and proposed replacements against tax, risk, and trading-cost constraints.

Follow the tax-lot workflow
Tax-lot analysis
Cost basis changes the account decisionFour illustrative lots sit above or below the same reference market price. One loss-bearing lot is selected for eligibility review; another remains restricted. A loss alone does not establish eligibility.RELATIVE COST BASISMarket priceLOT 01LOT 02LOT 03LOT 04Embedded gainReview lossRestrictedRetainEligibility first
01 / Identify the lot02 / Check replacements03 / Review exposureConceptual lot positions. No prices, tax savings, or executed trades are shown.
01.3 / TRANSITIONS

A new target.
A considered path.

Move the legacy portfolio toward its destination under an agreed tax budget, tracking-error limit, and implementation cost.

Explore portfolio transitions
Portfolio transition
A portfolio transition narrows the gap to a new mandateThree allocation profiles show current holdings, a staged proposal, and the target. A retained position narrows progressively while a new exposure grows. The path is conceptual and must be evaluated under tax, turnover, and risk limits.CURRENTSTAGEDTARGETRetained positionNew exposureLegacy bookAgreed limitsNew mandate
Tax budgetTurnoverTracking objectivesConceptual allocation path. Band widths are illustrative, not a measured portfolio.

EVIDENCE / TWO DIFFERENT QUESTIONS

Fast is meaningful
when the rules match.

A generated portfolio-QP comparison and an executed-trade parity test answer different questions. Keep their data, models, timing, and quality gates separate.

01 / GENERATED COMMON FACTOR-QP

Portfolio profiles.
A defined quality gate.

26.25×

Median speedup at 100K assets against the stated tuned commercial CPU reference. All eight tested profiles passed in the quality-capped configuration.

Model, timing & acceptance gate
Data
Seeded/generated portfolio profiles; not real-market observations.
Model
Common factor-QP; not the full tax-lot/SOC/ledger model.
Acceptance
Objective gap ≤ 1e−3; feasibility ≤ 2e−6.
Comparison
GPU vs tuned 8-thread CPU reference. Isolated confirmation includes instance construction and solve after one excluded GPU kernel warm-up; not client workflow latency.
Read scope and provenance
02 / FULL TAX-LOT / SOC / LEDGER MODEL

Executed trades.
A frozen cent gate.

1K

Assets and lots at which strict executed-objective parity was established: 1.480 s vs 3.587 s, or 2.424×, under the stated protocol. At 5K, the frozen $0.01 gate is not met.

Model, timing & acceptance gate
Acceptance
$0.001496 executed difference within the $0.01 gate.
Boundary
At 5K, $0.018935 exceeds the $0.01 gate.
Scale
Capacity measured to 500K assets/lots; larger strict-parity speed superiority is not established.
Meaning
Numerical executed-trade comparison; not realized client tax savings.
Read the acceptance boundary
Other evaluated models and configurations

Keep the losses in. The large fixed-income profile did not converge under its gate; the commercial reference won. The separately retuned common-QP configuration has its own results and must not be blended with the 26.25× quality-capped result.

THE PLATFORM CONTRACT

A decision your
platform can inspect.

PRISM is an optimization layer for investment platforms. Your platform owns the investment offering, client experience, and order execution.

01 / INPUT

Holdings. Lots. Rules.

Supply the investment universe, account state, constraints, and agreed evaluation settings.

02 / DECISION

Optimize within the gate.

Evaluate candidate portfolios under the workflow’s quality, feasibility, and time requirements.

03 / OUTPUT

Review before execution.

Inspect proposed trades, status, latency, and the audit record before downstream execution.

An 8-week paid matched-workload pilot. Agree the pass/fail metric before work begins.

PRISM-Q / DISCRETE PORTFOLIO DECISIONS

More possibilities.
One checked decision.

Explore candidate baskets under your constraints. Refine them with PRISM and inspect the proposed decision.

Explore PRISM-Q

Simulation workflow built with NVIDIA CUDA-Q.
Method & evidence

Conceptual isometric compute and cryostat illustration of the PRISM-Q portfolio workflow
Conceptual quantum–classical architecture · Illustration: NVIDIA.

One investment mandate.
Many individual portfolios.

Share the investment objective. Preserve each account’s starting holdings, restrictions, and tax context through the construction process.

PRISM / Portfolio optimization

Optimization infrastructure
for investment teams.

PRISM by Asymmetry Computing supports portfolio construction, direct indexing, tax-aware rebalancing, and portfolio transitions for advisors, registered investment advisers (RIAs), custodians, and wealth platforms. Its B2B2C role is portfolio computation within the institutional workflows that serve individual investors.

Start with your investment objective, risk model, holdings, and account restrictions. Define the workload and acceptance criteria, then assess numerical quality, portfolio outcomes, and runtime together.

Read the product overview Inspect the benchmark data Explore the integration guide

A direct conversation

Your workflow.
Let’s work through it.

Meet with Debdoot Ghosh, founder of Asymmetry Computing. Bring your questions about portfolio construction, integration, or a potential pilot.

Book a meeting

30 minutes · Scheduled through Google Calendar

Original platform context and study wording

PLATFORM DOSSIER / EXISTING CONTEXT

The earlier cross-domain explanation is retained below. Reported performance belongs to its stated workload and date; illustrations and research are distinct from production evidence.

PRISM · One real-time optimization engine

Extreme scale. Hyper-personalized. In real time.

Fully deterministic·Provably feasible·Auditable

The optimization engine for problems too big and too personal for conventional solvers — solved orders of magnitude faster, the instant the world moves. One engine, proven in finance and expanding across real-time operations.

8-week paid pilot on your data — you get a benchmark report on your real problem and reference pricing for production. Prefer a form? Request a pilot

Historical evidence and earlier studies

HISTORICAL PUBLIC RECORD / MARCH–AUGUST 2026

The earlier studies, in context.

The original explanations and results below are preserved with their original study scope. Real-data statements apply to the historical studies, not the generated-QP results above. Reported tax budgets are not realized client tax savings. The 500K-account reference runtime is extrapolated from a smaller measured sample. Feasibility and timing claims do not establish universal product guarantees.

Start here

What this actually does, in plain English.

No background assumed. Every specialist term on this site is defined on the glossary page.

Some decisions have three awkward properties at the same time. There are far too many possible answers to check by hand. The answer has to obey strict rules. And it has to arrive before a particular moment, or it is worthless — however good it would have been.

Two real examples. A wealth manager runs fifty thousand individual investment accounts. Every night, before the market opens, each one needs its own list of what to buy and sell — respecting that client's tax position, restrictions and risk limits. Miss the open and the trades don't happen. Separately, an electricity grid operator has thousands of batteries spread across a region. Every few seconds, each one needs a new instruction as power prices move. The grid does not wait; miss the moment and the opportunity is gone.

These two look nothing alike. Underneath they are the same problem: search an enormous number of options, obey every rule, and finish on time.

PRISM is the software that does that. It returns an answer that breaks none of the rules, is as close to the best available as makes any practical difference, arrives inside the deadline, and comes with a record of how it was reached — so an auditor or regulator can check the decision later. Run it twice on the same inputs and you get exactly the same answer.

The honest short version. Existing tools already solve these problems well at small scale. They struggle when the number of accounts, assets or devices gets large and a hard deadline applies. That gap is the only thing we sell into — and where we don't win, we publish that too.

Same shape, every timeThree properties → one engine
An enormous option space Every rule obeyed Finished before the deadline PRISM Feasible · near-optimal · audited Deterministic, every run
The three properties on the left describe almost any hard operational decision. PRISM is what sits on the right of the arrow — the same engine, whatever the domain.
Where PRISM is proven

Anchored in finance. One engine, expanding.

The deepest, money-grade proof is in finance — direct indexing and real-time portfolio decisioning, on real data, losing cases kept in. The same core extends to any high-stakes decision that must be re-solved correctly, on time, the instant the world moves. Alongside it sits an open research lane in quantum methods — evaluated honestly, published either way — and, third, real-time energy dispatch on live grid data.

One optimization core · four validated lanesFinance → quantum research → energy
PRISM Real-time optimization core One engine 1 · Direct Indexing $238K captured 2 · Portfolio Decisioning 75K assets, <1s 3 · Quantum Research Pre-registered 4 · Energy Dispatch ~5ms deadline
Read top to bottom: the deepest, money-grade evidence is in finance; next, an open quantum research lane, evaluated honestly and published either way; third, real-time energy dispatch on live grid data. One engine underneath all four.

Beyond the four above, the same engine is expanding across real-time operations — telecom & satellite resource allocation, AI/HPC cluster scheduling, and pricing & budget allocation — with molecular candidate selection on the research roadmap. Earlier-stage work is under NDA.

Selected measured results

Measured on real data. Losing cases kept in.

Every figure here traces to a recorded run on real data — with losing cases kept in, not cherry-picked. The peer-reviewable write-up is arXiv:2606.23367; the evaluation artifacts are public on GitHub; the provenance of each number is on the research page.

Fleet throughput
10.2×
more accounts priced per core-second
100,000 accounts priced in ~11.6 min · deterministic & auditable
Tax alpha captured
$238K
harvestable tax budget on a $5M, 192-name book
full budget band vs commercial baseline on the structured lane
Real-time dispatch
~5 ms
feasible, audited grid dispatch plan
delivered within deadlines from 500 ms down to ~5 ms · real CAISO data
100,004-asset structured solve. A real-data structured solve across 100,004 assets and 47+ exchanges, completed deterministically with a full audit trail. The current recorded solve time on the 100k structured lane is 2.16 s.
Reproduce it yourself. On public Kenneth French (FF30) data — no licensed data, nothing to take on faith — PRISM runs 23.28× faster than a commercial baseline at matched quality. The fastest way to trust a benchmark is to run it on data you already have. See the methodology
Why PRISM

Built for operational reality.

The answer arrives inside the deadline that matters — fast where scale demands it, feasible and auditable every time. One engine, the same contract across every domain.

Speed

Built for the books where scale is the problem — hundreds of thousands of accounts, or universes in the tens of thousands of names, answered inside the window you actually operate in. Per-lane speed figures are being re-measured under a contention guard before republication.

Reproducible

Audit-ready delivery with explicit, documented execution paths. Deterministic, content-hashed, re-derivable — every trade traces to your constraints.

Works with your stack

HTTPS API. Slots into OMS/EMS, grid control loops, custodian APIs, and risk models. Deploy in your VPC, on-prem, or air-gapped — no per-seat solver license.

Proven at scale

75,257 real assets on the proof lane and a 100,004-asset structured suite. Evidence spans head-to-head, routed scale, and backtest surfaces.

Constraints
All constraints enforced simultaneously. Deterministic solve. Auditable.
Hard real-time deadlines (control-tick to overnight)
Ramp limits and state-of-charge (battery / DER)
Wash-sale handling, lot-level tax accounting
Tracking error and risk-factor caps
Turnover limits and transaction-cost penalties
Liquidity constraints (ADV, min trade sizes)
Position-count and cardinality limits
Sector and ESG exposure limits
The gap conventional solvers can't close

The answer that arrives too late is worth nothing.

Across grids, trading desks, and personalized books, the world moves on a clock you don't control. Exact solvers find a beautiful answer — long after the moment to act has passed. Simpler methods keep up but leave value, or feasibility, on the table.

PRISM lives in that gap: a feasible, near-optimal, fully auditable answer inside the deadline — the same engine, whether the window is a grid control-tick or an overnight book.

The contract, on every solve
Feasibleevery constraint honoured
Auditableevery leg reported
Ingest, per-account build and solve are broken out separately on every result — no hidden legs, losing cases kept in. Speed figures are being re-measured under a contention guard and will be republished per lane with the load recorded alongside.
Built for the desk Deterministic & content-hashed Fully reproducible audit trail VPC, on-prem, or air-gapped No per-seat solver license

Bring your hardest real-time problem. We'll prove it on your data.

8-week paid pilot on your data — you get a benchmark report on your real problem and reference pricing for production. You set the pass/fail metric before we start; every losing case is shown.

Start an 8-week paid pilot

Prefer a form? Request a pilot