Solution · Collateral Optimization

Post the cheapest eligible collateral — across every book.

Cheapest-to-deliver is easy to say and hard to compute when eligibility schedules, haircuts, concentration limits, and funding costs all interact across margin, repo, and securities-lending obligations at once. PRISM treats the whole thing as the single optimization it is, and finds the allocation that costs you the least while staying compliant everywhere.

In plain English. When a bank borrows or trades, it has to post assets as security — collateral. It usually owns many assets that would qualify, and they differ in what they cost the bank to tie up. So there is a genuinely cheapest combination to hand over, but working it out means respecting which assets each counterparty accepts, how much each one counts for, and limits on concentrating in any one thing. This page is about finding that cheapest legal combination. Unfamiliar words? See the glossary.

Cost-sorted inventory Allocating to eligible bins
−18%
funding cost vs greedy cheapest-to-deliver
100,004
Asset structured solve, completed deterministically
2.16s
Structured solve time on the large-N GPU lane
3
Books optimized together — margin, repo, securities lending
1
Objective across eligibility, haircuts, and funding cost
Cost-sorted collateral inventory allocated across margin, repo and lending books
PRISM · Collateral Optimization Margin, repo and securities-lending obligations solved as one cost-minimizing allocation — eligibility-, haircut- and concentration-aware — instead of one expensive silo at a time.
01 · The problem

Silos leave money on the table.

Every desk that posts collateral knows the principle: deliver the asset that satisfies the obligation at the lowest economic cost, keeping the valuable, liquid, low-haircut assets free for where they earn more. The principle is simple; the computation is not. Each obligation has its own eligibility schedule and haircut; each asset has its own funding cost and opportunity cost; concentration and wrong-way-risk limits constrain the whole picture; and an asset posted here cannot be posted there. Optimizing one obligation at a time — the way siloed systems do — produces a locally sensible, globally expensive answer, because the best asset for this obligation might have been far more valuable held back for that one.

The right answer requires looking at all obligations and all available collateral at once and solving for the allocation that minimizes total cost subject to every constraint simultaneously. That is a genuine large-scale constrained optimization, and it is exactly the shape of problem PRISM exists to solve.

02 · How PRISM fits

One optimization across the whole inventory.

You bring the obligations, the eligible inventory, and the cost and constraint schedules; PRISM returns the cost-minimizing allocation and a reproducible audit trail.

Inputs
  • Obligations & eligibility
  • Available inventory
  • Haircuts & funding costs
  • Concentration limits
PRISM
Optimization core

Minimizes total cost across all obligations and inventory at once.

Outputs
  • Cost-minimizing allocation
  • Per-obligation assignments
  • Cost & savings summary
  • Audit log

Global, not greedy

Solves the whole inventory-against-obligations problem at once, capturing savings siloed allocation misses.

Constraint-faithful

Eligibility, haircuts, and concentration limits are hard constraints, not after-the-fact checks.

Deterministic

Same inputs, same allocation — re-derivable for treasury, risk, and audit.

Finance workflow · Illustrative model

See the capital your book is carrying.

Set your notional and risk-class mix, then move hedge effectiveness and diversification to see how the standardised capital charge responds. Illustrative — not a regulatory calculation.

Bring a real book. We'll show you the savings.

In a matched-workload pilot, run your real obligations and inventory through PRISM and compare the allocation cost against your current process — every number traceable.

Request a pilot →