The Forecasting page in ZoneLiquidity displays projected bank balances over your selected forecast period, with interactive charts, drill-through explainability, forecast comparison tools, and what-if scenario modelling to support cash planning decisions.
Note: Clicking on the images within this guide will enlarge them for better viewing.
Overview
Once a forecast has been generated and its status is Completed, navigate to Setup > ZoneLiquidity > Forecasting to access the Cash Forecast page. All forecasts you have previously submitted are listed here, starting from the earliest. Select the forecast you wish to review.
Important: Forecasts are currently identified by their internal system ID only. Custom naming is not yet supported. If you have submitted multiple forecasts, identify the correct record by cross-referencing the creation date and configuration details in the ZR Forecast Definition record.
Forecast chart
The forecast chart visualises projected bank balances over the forecast period. 2 chart types are available: a line chart and a bar chart. Toggle between them using the chart type controls at the top of the chart.
Line chart
The line chart displays the projected cash balance over time. The following reference lines are overlaid on the chart where destination rules have been configured:
- Liquidity floor — a horizontal line marking the minimum balance you want to keep on hand. If the projected balance drops below it, the forecast is flagging a potential shortfall.
- Cash milestone target — a horizontal line marking the balance you're aiming to reach by the end of the forecast period.
Two forecast methods are plotted on the line chart:
| Method | How it works |
|---|---|
| Deterministic | Used when a clear recurring pattern is detected — for example, a rent payment processed on the same date each month. The system identifies the cadence, the vendor, and the expected amount, and applies them directly as the forecast. |
| AI | Used when no clear pattern is detected. The system groups smaller or irregular transactions and applies a statistical model based on historical transaction volume to generate a forecast value. |
Below are the key practical differences:
| Deterministic | AI (TimeGPT) | |
|---|---|---|
| Method | Fixed interval rule | ML time-series model |
| Series ID prefix | Entity-based | TimeGPT: |
| Output | Single amount | Amount + low/high range |
| Explainability | Full pattern detail (interval, anchor, match rate, source transactions) | Prediction interval level only |
| Best for | Regular, predictable payments | Irregular or complex patterns |
Hover over any point on the line chart to see a tooltip showing the forecasted balance, the liquidity floor value, and the cash milestone target for that date.
Bar chart
Switch to the bar chart view to see inflows and outflows broken down by day. Click any bar on the chart to load the corresponding date in the Breakdown section below the chart.
Breakdown (transparency layer)
The Breakdown section below the chart lists the transactions contributing to the inflows and outflows for a selected date. Select a date in the Breakdown to load the transactions for that day.
Each transaction line in the Breakdown includes a Forecast Lineage link. Click Forecast Lineage to open a detailed explanation of how that transaction's forecast amount was calculated.
Understanding forecast lineage
The Forecast Lineage panel provides full transparency into the AI model's reasoning for a specific transaction or transaction group. The following fields are displayed:
| Field | Description |
|---|---|
| Occurrence score | A measure of how frequently this transaction type occurs in the historical data. |
| Amount consistency | An indicator of how consistent the transaction amount has been across historical occurrences. |
| Pattern match rate | The confidence level that the detected pattern is a reliable predictor of future behaviour. |
| Mean and prediction range | The mean forecast amount, along with a low estimate and a high estimate derived from the standard deviation of historical values. The low estimate is the minimum expected amount; the high estimate is the upper bound of the prediction range. |
| Source transactions | The historical transactions used to calculate the forecast, displayed as individual records. |
| Upcoming forecasts | The projected dates within the forecast period on which this transaction is expected to recur. |
The Forecast Lineage panel content differs depending on whether a pattern was detected:
- For named vendor or customer lines: the lineage confirms a recurring pattern was detected and provides full detail including pattern match rate and upcoming forecast dates.
- For grouped lines (e.g. Other AP): the lineage confirms no specific pattern was identified. The system grouped these transactions and applied a statistical model. Occurrence score, amount consistency, and pattern methodology are still reported.
Transactions sublist
The Transactions subtab on the Cash Forecast page lists all transactions that were sent to the AI model for this forecast. This sublist serves as an audit log and contains the same number of records as the Sync Count shown in the ZR Forecast Definition record.
Variance analysis
The Variance Analysis subtab compares the forecasted bank balance against the actual bank balance for each period in the forecast. Use this view to evaluate forecast accuracy after the period has passed.
For forecast periods that have not yet ended, the Actual and Variance columns will be empty. These columns populate automatically as actual bank balance data becomes available in NetSuite.
Forecast history
The Forecast History table provides a summary of inflows, outflows, net cash flow, and the number of days remaining in the forecast period, giving an at-a-glance overview of the forecast's overall trajectory.
Comparing forecasts
Use the Compare Forecasts feature to compare 2 forecast records side by side. This is useful for evaluating planned versus actual cash positions across different periods — for example, comparing a Q2 forecast against a Q3 forecast.
To compare forecasts:
- Navigate to Setup > Treasury > Forecasting and open the Cash Forecast page.
- Select Compare Forecasts.
- Choose 2 forecast records from the pick-lists provided.
- Review the summary metrics and cash flow comparison displayed for each forecast.
Tip: When selecting forecasts to compare, check the creation date and date range displayed for each record to confirm you are comparing the intended periods.
Refreshing the forecast
Click Refresh Forecast to re-send transaction data to the AI model and update the forecast with the most recent transactions. The Last Sync timestamp updates to reflect when the refresh completed.