decision nodes sensitivity analysis

Decision analysis can be performed on decision nodes. Copilot suggests 3 approaches which could be considered.
- Option Removal Analysis
- Decision Leverage Analysis
- Decision Importance Ranking

---

Most people do not perform this analysis enough.

Question:

> "How valuable is each decision opportunity?"

Examples:

- Drill appraisal well?
- Delay decision?
- Expand facility?
- Farm down?
- Abandon?

Here we are no longer testing uncertainty.

We are testing the strategy space itself.

---
**Option Removal Analysis**

Remove one decision alternative and recompute MEU.

Example:
```
1 Base MEU:
2 500 M$
```
Remove:

```
1 Delay option
```

New MEU:

```
1 450 M$
```

Value of flexibility:

```
1 50 M$
```
Message
```
1 The delay option contributes 50 M$
2 of strategic value.
```

This is very different from uncertainty sensitivity.

---

**Decision Leverage Analysis**

Measure:

$$
MEU_{best}-MEU_{worst}
$$

at a decision node.

Example:

```
1 D3: Facility concept
2
3 Tieback 520
4 FPSO 500
5 Minimal 420
```

Decision leverage:

```
1 100 M$
```

Message

```
1 This decision is strategically important.
```

High leverage decisions deserve management attention.

---

**Decision Importance Ranking**

For each decision node compute:

$$
EV(\text{optimal}) - EV(\text{suboptimal})
$$

Ranking example:

| Decision | Impact |
| -- | -- |
| Development concept | 120 |
| Appraisal | 70 |
| Expansion | 30 |
| Monitoring | 10 |

Message
```
1 Not all decisions matter equally.
```

This is often a very powerful executive insight.

No files, tests, or implementation entry points are named. Start by locating the existing decision-node and MEU calculation entry points, then clarify which of the three proposed analyses is in scope and how results should be exposed; done means the selected analysis is specified with representative decision-node cases and validated outputs.

Written by the indexing model from the issue text.