Serialization

All major curve types and the Solver support JSON serialization via to_json() and from_json(). This enables persisting calibrated curves to disk, transmitting them over a network, or inspecting their structure for debugging. Links to DiscountCurve for the primary example.

Discount Curve Round-Trip

The serialization pattern is the same for all curve types: call to_json() to get a JSON string, and the corresponding static from_json() to restore the object. The restored curve produces identical query results.

import datetime
import json
from vade import DiscountCurve, LineCurve, NelsonSiegel, Solver, Deposit, IRS

effective = datetime.date(2025, 6, 16)

curve = DiscountCurve(
    {
        effective: 1.0,
        datetime.date(2025, 12, 16): 0.98,
        datetime.date(2026, 6, 16): 0.96,
        datetime.date(2027, 6, 16): 0.925,
        datetime.date(2030, 6, 16): 0.87,
    },
    interpolation="log_linear",
    convention="act360",
)

json_str = curve.to_json()
isinstance(json_str, str)  # True

restored = DiscountCurve.from_json(json_str)
float(restored.discount_factor(datetime.date(2026, 6, 16)))  # 0.96
assert float(restored.discount_factor(datetime.date(2026, 6, 16))) == float(
    curve.discount_factor(datetime.date(2026, 6, 16))
)

LineCurve and Parametric Curves

The same to_json() / from_json() pattern applies to LineCurve and parametric curve types like NelsonSiegel. Each type's from_json() is a static method that returns the correct type.

line = LineCurve(
    {
        effective: 0.040,
        datetime.date(2026, 6, 16): 0.039,
        datetime.date(2027, 6, 16): 0.038,
    },
    interpolation="linear",
    convention="act360",
)

line_restored = LineCurve.from_json(line.to_json())
float(line_restored.discount_factor(datetime.date(2026, 6, 16)))  # 0.9612299019102728
assert float(line_restored.discount_factor(datetime.date(2026, 6, 16))) == float(
    line.discount_factor(datetime.date(2026, 6, 16))
)

ns = NelsonSiegel(
    beta0=0.04, beta1=-0.02, beta2=0.03, tau=1.5,
    base_date=effective, convention="act365f",
)

ns_restored = NelsonSiegel.from_json(ns.to_json())
float(ns_restored.zero_rate(effective, datetime.date(2026, 6, 16)))  # 0.03189622964353336
assert float(ns_restored.zero_rate(effective, datetime.date(2026, 6, 16))) == float(
    ns.zero_rate(effective, datetime.date(2026, 6, 16))
)

Solver State

Persisting a calibrated Solver captures the full calibration state -- curves, instruments, and weights. This is useful for reproducibility: save after calibration, restore later without re-running the optimization.

cal_curve = DiscountCurve(
    {
        effective: 1.0,
        datetime.date(2025, 12, 16): 1.0,
        datetime.date(2026, 6, 16): 1.0,
        datetime.date(2027, 6, 16): 1.0,
    },
    interpolation="log_linear",
    convention="act360",
    id="sofr",
)

dep = Deposit(effective=effective, termination="6m", rate=0.0, convention="act360")
irs_1y = IRS(
    effective=effective, termination="1y", frequency="a",
    fixed_rate=0.0, convention="act360", float_convention="act360",
)
irs_2y = IRS(
    effective=effective, termination="2y", frequency="a",
    fixed_rate=0.0, convention="act360", float_convention="act360",
)

solver = Solver(
    curves=[cal_curve],
    instruments=[(dep, 0.0430), (irs_1y, 4.00), (irs_2y, 3.85)],
)
result = solver.iterate()
assert result.converged

df_before = float(solver.get_curve(0).discount_factor(datetime.date(2026, 6, 16)))

solver_json = solver.to_json()
restored_solver = Solver.from_json(solver_json)
df_after = float(restored_solver.get_curve(0).discount_factor(datetime.date(2026, 6, 16)))

df_before  # 0.9610212072035146
df_after   # 0.9610212072035146
assert abs(df_before - df_after) < 1e-10

Inspecting JSON Structure

Use json.loads() on the serialized string to inspect the internal structure. This is useful for debugging and integrating with external tools that consume JSON.

data = json.loads(curve.to_json())
sorted(data.keys())  # ['calendar', 'convention', 'id', 'interpolator', 'modifier', 'nodes']

Next Steps

Bootstrap & Parametric Curves -- Iterative bootstrap and parametric curve models
Bonds -- Bond analytics and spreads

MarketContext Serialization

For serializing and restoring a complete market environment (curves, fixings, FX rates, calendars), see the Market Context guide.