PumpkinVine predictive analytics dashboard interface used for portfolio risk assessment

AI-Assisted Decision Analytics

Predictive modelling for capital preservation, built on backtested market data

PumpkinVine analyses historical and real-time market data to identify risk-adjusted allocation paths, helping retirees and private investors reduce exposure to volatility without abandoning growth objectives.

View Performance Data Request Technical Overview

Backtested Model Snapshot

15yr

Historical window

0.4s

Re-analysis cycle

Volatility bands

Why Predictive Modelling Matters at Retirement

Retirement portfolios carry a different risk profile than accumulation-phase investing

Once regular income withdrawals begin, sequence-of-returns risk becomes a primary concern: a downturn early in retirement can permanently reduce the sustainable withdrawal rate, even if long-term average returns remain acceptable. PumpkinVine models this risk directly, rather than relying on static, average-return assumptions.

The platform's role is to surface statistically significant patterns in historical drawdowns and recovery periods, presenting the resulting risk-adjusted scenarios for the investor or adviser to weigh against personal circumstances. The final allocation decision remains with the account holder.

Historical backtesting depth15 years
Portfolio re-analysis frequencyDaily
Volatility scenarios modelledLow / Med / High
Human sign-off on allocation changesRequired

Core Capabilities

How the platform reduces exposure to unnecessary risk

Each capability below addresses a distinct stage of the analysis pipeline, from raw data ingestion through to a recommendation an investor can review before acting.

Stage 01 — Ingestion

Real-time data aggregation

Market pricing, macroeconomic indicators, and portfolio holdings are consolidated into a single dataset, refreshed continuously during market hours rather than on a fixed reporting cycle.

Refresh intervalIntraday

Stage 02 — Modelling

Risk-adjusted scenario analysis

Predictive models run multiple drawdown and recovery scenarios against the current portfolio composition, flagging concentrations that historically correlated with above-average volatility.

Scenarios per cycleMulti-path

Stage 03 — Recommendation

Decision-support output

Findings are presented as a ranked set of allocation adjustments with the underlying reasoning shown, so the investor or adviser can evaluate the trade-offs before approving any change.

Approval stepManual

Methodology

Recommendations are derived from backtested performance, not projections alone

Drawdown Recovery — Illustrative Backtest

Simulated allocation response across historical downturn periods (indexed, non-representative of any live client outcome)

Every recommendation surfaced by PumpkinVine is first tested against historical market conditions, including periods of significant contraction, to assess how a given allocation would have performed in practice rather than in theory.

The engine does not search for the highest theoretical return. It searches for allocations that historically reduced the depth and duration of drawdowns while preserving a comparable long-run outcome — a trade-off most relevant to investors drawing an income.

Transparency statement: Backtested results describe past market behaviour and the model's historical response to it. They are not a projection of future performance, and no allocation strategy eliminates market risk entirely.

Applications

From an individual retirement account to broader asset allocation strategy

Individual Retiree Portfolio

A retiree drawing a fixed monthly income has their portfolio continuously re-assessed for sequence-of-returns risk. Adjustments are proposed when modelled downside exposure exceeds the investor's stated tolerance band, with the reasoning shown alongside each suggestion.

DailyRisk re-check

Self-Managed Super Fund (SMSF)

Trustees managing an SMSF use the platform to stress-test the fund's current allocation against historical downturn scenarios, identifying concentration risk across asset classes before it becomes material to compliance reporting.

Multi-assetStress testing

Adviser-Managed Client Books

Financial advisers apply the same backtesting engine across a book of client portfolios, using standardised, evidence-based outputs to support conversations about risk tolerance rather than relying on generic model portfolios.

Portfolio-levelReporting

Institutional Asset Allocation

Larger allocators use the platform's scenario engine to evaluate strategic asset allocation shifts at scale, cross-referencing proposed changes against decades of historical market cycles before committing capital.

DecadesOf historical data

Security & Trust

Built for the constraints of the Australian financial environment

Data encryption in transit and at rest

Portfolio and account data is encrypted throughout transmission and storage, with access limited to systems required for the analysis pipeline to function.

Privacy aligned with Australian standards

Personal and financial data handling follows the Australian Privacy Principles, with data retention limited to what is required to maintain an accurate model of the account.

Human oversight on every recommendation

The platform generates decision support, not automated trades. Every allocation change requires explicit approval from the investor or their adviser before it is actioned.

About the Approach

An analytical layer that sits alongside human judgement, not in place of it

PumpkinVine was built on the premise that retirees and private investors are better served by transparent, evidence-based analysis than by opaque automated trading. The platform's models are designed to explain their reasoning, not just their conclusion.

Every output includes the historical basis for the recommendation, so it can be reviewed, questioned, and either accepted or declined by the person responsible for the capital.

PumpkinVine analyst reviewing backtested portfolio data on a workstation

Frequently Asked Questions

Common questions about the role of AI in the decision process

Does PumpkinVine make investment decisions automatically?

No. The platform produces ranked recommendations with supporting historical evidence. Any change to a portfolio's allocation requires explicit approval from the account holder or their adviser.

How is "backtested" performance calculated?

Backtesting applies the current model logic to historical market data to observe how a given allocation would have behaved during past periods, including known downturns. Results are historical simulations, not guarantees of future performance.

Is this suitable for a Self-Managed Super Fund?

Trustees commonly use the platform's scenario analysis to inform SMSF allocation decisions. The output is designed to support, not replace, the trustee's own compliance and investment obligations.

What happens to my data if I stop using the platform?

Data retention is limited to what is necessary to maintain an accurate model while the account is active. On closure, data is handled in line with the platform's privacy policy and applicable Australian retention requirements.

Can I see the reasoning behind a specific recommendation?

Yes. Each recommendation is presented with the historical scenarios and risk factors that informed it, so it can be reviewed on its own terms before any decision is made.