Skip to main content

​
Why scenario comparison matters

SIMCEL’s core value is running multiple “what-if” simulations simultaneously. The Insights API lets you extract and compare those scenarios in your own code. Here are the most useful patterns.

​
Pattern 1: Delta table

Generate a variance table between a baseline (Committed) and one or more alternatives. 
def delta_table(base_name: str, scenarios: list[dict]) -> None:
    base = next(s for s in scenarios if s["scenarioName"] == base_name)
    others = [s for s in scenarios if s["scenarioName"] != base_name]

    metrics = ["netFIESales", "grossProfit", "ebit", "dpm"]

    print(f"{'Metric':<22} {'Base':>14}", end="")
    for s in others:
        print(f"  {s['scenarioName']:>16}  {'Δ vs Base':>12}", end="")
    print()
    print("─" * 80)

    for m in metrics:
        base_val = base["pnl"][m]
        print(f"{m:<22} {base_val:>14,.0f}", end="")
        for s in others:
            val = s["pnl"][m]
            delta = val - base_val
            print(f"  {val:>16,.0f}  {delta:>+12,.0f}", end="")
        print()

​
Pattern 2: Best-case / worst-case bounds

Automatically identify the min and max across all scenarios for a given metric. 
def scenario_bounds(scenarios: list[dict], metric: str) -> dict:
    values = {s["scenarioName"]: s["pnl"][metric] for s in scenarios}
    best = max(values, key=values.get)
    worst = min(values, key=values.get)

    return {
        "metric": metric,
        "best": {"scenario": best, "value": values[best]},
        "worst": {"scenario": worst, "value": values[worst]},
        "spread": values[best] - values[worst],
        "spread_pct": (values[best] - values[worst]) / values[worst],
    }

bounds = scenario_bounds(data["scenarios"], "ebit")
print(f"EBIT range: ${bounds['worst']['value']:,.0f} ({bounds['worst']['scenario']}) "
      f"→ ${bounds['best']['value']:,.0f} ({bounds['best']['scenario']})")
print(f"Spread: ${bounds['spread']:,.0f} ({bounds['spread_pct']:.1%})")

​
Pattern 3: Monthly trend divergence

Find the month where two scenarios start to diverge significantly. 
def find_divergence_month(
    scenario_a: dict,
    scenario_b: dict,
    metric: str = "demandValue",
    threshold_pct: float = 0.05,
) -> str | None:
    """Return the first month where scenarios diverge by more than threshold_pct."""
    series_a = {row["period"]: row[metric] for row in scenario_a["series"]}
    series_b = {row["period"]: row[metric] for row in scenario_b["series"]}

    for period in sorted(series_a.keys()):
        a, b = series_a[period], series_b.get(period, 0)
        if a == 0:
            continue
        if abs(a - b) / a > threshold_pct:
            return period

    return None

month = find_divergence_month(
    data["scenarios"][0],   # Committed
    data["scenarios"][1],   # Optimistic
    metric="demandValue",
    threshold_pct=0.05,
)
print(f"Scenarios diverge from: {month}")

​
Pattern 4: Scenario ranking by KPI

Rank scenarios on any KPI in a single call. 
def rank_scenarios(scenarios: list[dict], metric: str, higher_is_better: bool = True):
    ranked = sorted(
        scenarios,
        key=lambda s: s["pnl"][metric],
        reverse=higher_is_better,
    )
    print(f"\nScenario ranking by {metric}:")
    for i, s in enumerate(ranked, 1):
        print(f"  {i}. {s['scenarioName']}: {s['pnl'][metric]:>14,.0f}")

rank_scenarios(data["scenarios"], "ebit")
rank_scenarios(data["scenarios"], "supplyChainCost", higher_is_better=False)