Skip to content
LQM Registry / research

Rolling Probabilistic Forecast Calibration

Out-of-sample rolling AR(1) residual-bootstrap calibration with empirical interval coverage, width, interval score and explicit look-ahead guard.

Model keylq.research.rolling-forecast-calibration
Version1.0.0
Runtimepython-process
Model contract

Inputs are explicit.

{
    "type": "object",
    "required": [
        "values"
    ],
    "properties": {
        "values": {
            "type": "array",
            "items": {
                "type": "number"
            },
            "minItems": 40
        }
    }
}
Output contract

Outputs are explicit.

{
    "type": "object",
    "required": [
        "evaluation_protocol",
        "nominal_coverage",
        "empirical_coverage",
        "coverage_absolute_error",
        "average_interval_score",
        "lookahead_guard"
    ]
}
Default parameters
{
    "initial_window": 24,
    "interval": 0.90000000000000002220446049250313080847263336181640625,
    "scenarios": 300,
    "expanding_window": true,
    "rolling_window": 24
}
Example input
{
    "values": [
        100,
        100.400000000000005684341886080801486968994140625,
        100.7999999999999971578290569595992565155029296875,
        100.599999999999994315658113919198513031005859375,
        101.099999999999994315658113919198513031005859375,
        101.5,
        101.2999999999999971578290569595992565155029296875,
        101.900000000000005684341886080801486968994140625,
        102.2000000000000028421709430404007434844970703125,
        102,
        102.599999999999994315658113919198513031005859375,
        102.900000000000005684341886080801486968994140625,
        103.2000000000000028421709430404007434844970703125,
        103,
        103.7000000000000028421709430404007434844970703125,
        104.099999999999994315658113919198513031005859375,
        103.900000000000005684341886080801486968994140625,
        104.5,
        104.7999999999999971578290569595992565155029296875,
        105,
        105.400000000000005684341886080801486968994140625,
        105.2000000000000028421709430404007434844970703125,
        105.900000000000005684341886080801486968994140625,
        106.2999999999999971578290569595992565155029296875,
        106.099999999999994315658113919198513031005859375,
        106.7999999999999971578290569595992565155029296875,
        107,
        107.5,
        107.2999999999999971578290569595992565155029296875,
        107.900000000000005684341886080801486968994140625,
        108.2000000000000028421709430404007434844970703125,
        108,
        108.7000000000000028421709430404007434844970703125,
        109.099999999999994315658113919198513031005859375,
        108.900000000000005684341886080801486968994140625,
        109.5,
        109.7999999999999971578290569595992565155029296875,
        110.2000000000000028421709430404007434844970703125,
        110,
        110.7000000000000028421709430404007434844970703125,
        111.099999999999994315658113919198513031005859375,
        111.400000000000005684341886080801486968994140625,
        111.2000000000000028421709430404007434844970703125,
        111.900000000000005684341886080801486968994140625,
        112.2999999999999971578290569595992565155029296875,
        112.099999999999994315658113919198513031005859375,
        112.7999999999999971578290569595992565155029296875,
        113.2000000000000028421709430404007434844970703125
    ]
}
Capabilities
cpuresearchprobabilisticcalibrationwalk-forwardlqbench