PRISM

AI-native ALM analytics · demo access

Loading sign-in…

Demo with synthetic data — please don't enter confidential information.

PRISM

AI-native ALM analytics

Demo — work in progress. Runs on dummy / synthetic data; results are illustrative only. Please don't enter confidential information.

↵ send, Shift ↵ newline Working…

Summary

Tool explainer

What is PRISM?

PRISM is an AI-native analytics layer for Asset & Liability Management (ALM). Describe the analysis you want in plain language and PRISM selects the right model(s) from its registry, runs them, and assembles a dashboard — summary, metrics, charts and tables — alongside a written explanation. Follow-up messages can either refine the dashboard (e.g. "show it as a bar chart") or trigger a new calculation (e.g. "now add a +100bps shock"). The Model inputs and Model outputs tabs show exactly what was sent to, and returned by, each model run.

ALM models implemented in this demo

Curve model

Bootstraps a swap curve into a linearly-interpolated zero curve and discount factors. The model returns the linearly-interpolated zero curve and discount factors.

Non-Maturing Deposits (NMD) model

Values a non-maturing deposit portfolio using a Hull-White short-rate model for interest-rate simulation and a client-rate model for deposit pricing. The model returns NPV, BPV and duration.

NMD replication model

Derives an optimal replicating portfolio of fixed-rate instruments that matches the deposits' interest-rate risk (BPV). The model returns the resulting replication profile and the replication NPV, BPV and duration.

Mortgage valuation model

Values a mortgage portfolio incorporating a prepayment function. The model returns NPV, BPV and duration.

Mortgage replication model

Determines the optimal replicating portfolio of fixed-rate instruments for the mortgage portfolio, matching its interest-rate risk (BPV). The model returns the resulting replication profile and the replication NPV, BPV and duration.

Data & scope

This is a demonstration. The models are simplified versions of production ALM models and run on dummy / synthetic data. Results are illustrative only and should not be used for decision-making. Please don't enter confidential information.