{
 "bundle_of": "C12",
 "schema_version": "0.44",
 "paper": {
  "id": "entrylib:2026.000012",
  "version": 1,
  "title": "Complex exponential smoothing",
  "licence": "CC-BY-4.0",
  "status": "active",
  "date": "2026-11-17"
 },
 "claims": [
  {
   "id": "C12",
   "statement": "On the [[D32:M4 data without 10 series]], [[D5:SCUM]] has lower mean and median [[D15:MASE]], [[D14:RMSSE]] and [[D16:MSIS]] than [[D6:SCUM(noCES)]].",
   "type": "descriptive",
   "role": "principal",
   "novelty": "no_prior_found",
   "scope": {
    "population": "M4 competition time series",
    "excludes": [
     "10 series on which ETS broke down"
    ]
   },
   "support": [
    {
     "kind": "empirical_quant",
     "evidence": "E8",
     "results": [
      "E8.T1.r5",
      "E8.T1.r6"
     ]
    }
   ],
   "limitations": [
    "L3",
    "L5"
   ],
   "source_spans": [
    "p10.9",
    "p11.27"
   ],
   "provenance": "author_stated",
   "about": {
    "subject": "D5",
    "relation": "has lower mean and median errors than",
    "object": [
     "D15",
     "D14",
     "D16"
    ],
    "comparator": "D6",
    "conditions": [
     "D32"
    ]
   },
   "verifiability": "open",
   "verifiability_flags": []
  },
  {
   "id": "C60",
   "statement": "The error term $\\varepsilon_t$ of [[D1:CES]] is additive white noise.",
   "type": "descriptive",
   "role": "supporting",
   "support": [
    {
     "kind": "assumption"
    }
   ],
   "assumption_category": "statistical",
   "check": "unchecked",
   "source_spans": [
    "p4.5",
    "p4.14"
   ],
   "provenance": "author_stated",
   "verifiability": "not_applicable",
   "verifiability_flags": []
  },
  {
   "id": "C62",
   "statement": "The error term $\\varepsilon_t$ of [[D1:CES]] has no autocorrelation.",
   "type": "descriptive",
   "role": "supporting",
   "support": [
    {
     "kind": "assumption"
    }
   ],
   "assumption_category": "statistical",
   "check": "unchecked",
   "source_spans": [
    "p4.5",
    "p5.14"
   ],
   "provenance": "author_stated",
   "verifiability": "not_applicable",
   "verifiability_flags": []
  },
  {
   "id": "C63",
   "statement": "The error term $\\varepsilon_t$ of [[D1:CES]] has constant variance.",
   "type": "descriptive",
   "role": "supporting",
   "support": [
    {
     "kind": "assumption"
    }
   ],
   "assumption_category": "statistical",
   "check": "unchecked",
   "source_spans": [
    "p4.5",
    "p5.14"
   ],
   "provenance": "author_stated",
   "verifiability": "not_applicable",
   "verifiability_flags": []
  },
  {
   "id": "C64",
   "statement": "The error term $\\varepsilon_t$ of [[D1:CES]] is normally distributed.",
   "type": "descriptive",
   "role": "supporting",
   "support": [
    {
     "kind": "assumption"
    }
   ],
   "assumption_category": "statistical",
   "check": "unchecked",
   "source_spans": [
    "p4.14-p4.15"
   ],
   "provenance": "author_stated",
   "verifiability": "not_applicable",
   "verifiability_flags": []
  }
 ],
 "definitions": [
  {
   "id": "D1",
   "term": "CES",
   "definition": "Complex exponential smoothing: a forecasting method that combines each past observation $y_t$ and its forecast error $e_t$ into the complex variable $y_t + i e_t$ and forecasts with an exponentially weighted sum of these values, $$\\hat{y}_t + i\\hat{e}_t = (\\alpha_0 + i\\alpha_1) \\sum_{j=1}^{t-1} (1 + i - (\\alpha_0 + i\\alpha_1))^{j-1} (y_{t-j} + i e_{t-j})$$, or recursively $$\\hat{y}_t + i\\hat{e}_t = (\\alpha_0 + i\\alpha_1)(y_{t-1} + i e_{t-1}) + (1 - \\alpha_0 + i - i\\alpha_1)(\\hat{y}_{t-1} + i\\hat{e}_{t-1})$$, where $\\alpha_0 + i\\alpha_1$ is the [[D2:complex smoothing parameter]] and $\\hat{e}_t$ carries the information on past forecast errors.",
   "kind": "method",
   "assumptions": [
    "C60",
    "C62",
    "C63",
    "C64"
   ]
  },
  {
   "id": "D2",
   "term": "complex smoothing parameter",
   "definition": "The smoothing parameter of [[D1:CES]], the complex number $\\alpha_0 + i\\alpha_1$; its real part $\\alpha_0$ and imaginary part $\\alpha_1$ are the two real parameters estimated, and they set how much of the past observations and of the past forecast errors is carried into the forecast.",
   "kind": "variable",
   "part_of": [
    "D1"
   ]
  },
  {
   "id": "D3",
   "term": "nonlinear trend component",
   "definition": "The state $c_t$ of the [[D1:CES]] state-space model, which carries information on both its own previous value $c_{t-1}$ and the previous level $l_{t-1}$; it does not correspond to the trend component of exponential smoothing.",
   "kind": "concept",
   "part_of": [
    "D1"
   ]
  },
  {
   "id": "D4",
   "term": "seasonal CES",
   "definition": "[[D1:CES]] extended with a second pair of level and [[D3:nonlinear trend]] states taken at lag $m$, the seasonal period, with its own [[D2:complex smoothing parameter]] $\\beta_0 + i\\beta_1$: $y_t = l_{0,t-1} + l_{1,t-m} + \\varepsilon_t$, $$l_{0,t} = l_{0,t-1} - (1 - \\alpha_1) c_{0,t-1} + (\\alpha_0 - \\alpha_1) \\varepsilon_t$$, $$c_{0,t} = l_{0,t-1} + (1 - \\alpha_0) c_{0,t-1} + (\\alpha_0 + \\alpha_1) \\varepsilon_t$$, $$l_{1,t} = l_{1,t-m} - (1 - \\beta_1) c_{1,t-m} + (\\beta_0 - \\beta_1) \\varepsilon_t$$, $$c_{1,t} = l_{1,t-m} + (1 - \\beta_0) c_{1,t-m} + (\\beta_0 + \\beta_1) \\varepsilon_t$$.",
   "kind": "method",
   "broader": [
    "D1"
   ]
  },
  {
   "id": "D5",
   "term": "SCUM",
   "definition": "Simple combination of univariate models: the median of the forecasts of [[D9:ETS]], [[D10:ARIMA]], the [[D11:Theta method]] and [[D17:CES]].",
   "source": "B13",
   "kind": "method"
  },
  {
   "id": "D6",
   "term": "SCUM(noCES)",
   "definition": "The median of the forecasts of [[D9:ETS]], [[D10:ARIMA]] and the [[D11:Theta method]]: [[D5:SCUM]] without CES.",
   "kind": "method",
   "broader": [
    "D5"
   ]
  },
  {
   "id": "D7",
   "term": "the six compared methods",
   "kind": "method",
   "members": [
    "D17",
    "D9",
    "D10",
    "D11",
    "D5",
    "D6"
   ],
   "definition": "The [[D12:four compared methods]] and the two combinations, [[D5:SCUM]] and [[D6:SCUM(noCES)]]."
  },
  {
   "id": "D9",
   "term": "ETS",
   "definition": "Exponential smoothing in a state space framework with automatic selection among its models; here es() from the smooth package with model ZXZ, selecting by AICc.",
   "source": "B2",
   "kind": "method"
  },
  {
   "id": "D10",
   "term": "ARIMA",
   "definition": "Automatic state space ARIMA; here auto.ssarima() from the smooth package.",
   "source": "B15",
   "kind": "method"
  },
  {
   "id": "D11",
   "term": "Theta method",
   "definition": "The Theta method of forecasting; here thetaf() from the forecast package 8.14.",
   "source": "B16",
   "kind": "method"
  },
  {
   "id": "D12",
   "term": "the four compared methods",
   "kind": "method",
   "members": [
    "D17",
    "D9",
    "D10",
    "D11"
   ],
   "definition": "[[D17:CES with automatic seasonal selection]], [[D9:ETS]], [[D10:ARIMA]] and the [[D11:Theta method]], as the benchmark runs them."
  },
  {
   "id": "D14",
   "term": "RMSSE",
   "definition": "Root mean squared scaled error: the root mean squared forecast error over the horizon, scaled by the mean absolute first difference of the in-sample data; reported as its mean and median over the series.",
   "source": "B17",
   "kind": "variable"
  },
  {
   "id": "D15",
   "term": "MASE",
   "definition": "Mean absolute scaled error: the mean absolute forecast error over the horizon, scaled by the mean absolute first difference of the in-sample data; reported as its mean and median over the series.",
   "source": "B18",
   "kind": "variable"
  },
  {
   "id": "D16",
   "term": "MSIS",
   "definition": "Mean scaled interval score: the interval score of a prediction interval averaged over the horizon and scaled by the mean absolute first difference of the in-sample data; here of the 95% prediction interval, reported as its mean and median over the series.",
   "source": "B19",
   "kind": "variable"
  },
  {
   "id": "D17",
   "term": "CES with automatic seasonal selection",
   "definition": "[[D1:CES]] fitted by auto.ces() from the smooth package 3.1.2, which chooses between [[D4:seasonal CES]] and non-seasonal CES by AICc.",
   "kind": "method",
   "broader": [
    "D1"
   ]
  },
  {
   "id": "D32",
   "term": "M4 data without 10 series",
   "definition": "The 100,000 series of the M4 competition (23,000 yearly, 24,000 quarterly, 48,000 monthly, 359 weekly, 4,227 daily and 414 hourly) except series 19357, 43160, 51552, 62738, 92370, 92435, 93159, 93196, 93412 and 93491, whose influential outliers and structural breaks made ETS break down.",
   "source": "B9",
   "kind": "data"
  },
  {
   "id": "D37",
   "term": "holdout at the M4 horizons",
   "definition": "Each series split into an in-sample part and a holdout of the length used in the M4 competition: 6 observations for yearly, 8 for quarterly, 18 for monthly, 13 for weekly, 14 for daily and 48 for hourly series; forecasts made from the end of the in-sample part.",
   "kind": "design"
  }
 ],
 "evidence": [
  {
   "id": "E8",
   "profile": "quant",
   "design": "benchmark",
   "data_origin": "secondary",
   "preregistered": false,
   "n": 99990,
   "description": "The same six methods compared on the [[D32:M4 data without 10 series]] by error measures on a single holdout per series, without a significance test.",
   "artefacts": [
    "A7",
    "A4",
    "A5",
    "A8"
   ],
   "tables": [
    {
     "id": "T1",
     "description": "Mean and median MASE, RMSSE and MSIS of six methods on the M4 competition data, excluding 10 series",
     "row_labels": [
      "D7"
     ],
     "columns": [
      {
       "variable": "D15",
       "statistic": "mean"
      },
      {
       "variable": "D14",
       "statistic": "mean"
      },
      {
       "variable": "D16",
       "statistic": "mean"
      },
      {
       "variable": "D15",
       "statistic": "median"
      },
      {
       "variable": "D14",
       "statistic": "median"
      },
      {
       "variable": "D16",
       "statistic": "median"
      }
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "D17"
       ],
       "values": [
        2.921,
        2.339,
        2.379,
        1.803,
        1.529,
        0.592
       ]
      },
      {
       "id": "r2",
       "label": [
        "D9"
       ],
       "values": [
        3.31,
        2.688,
        4.751,
        1.831,
        1.542,
        0.632
       ]
      },
      {
       "id": "r3",
       "label": [
        "D10"
       ],
       "values": [
        3.037,
        2.436,
        2.272,
        1.903,
        1.606,
        0.617
       ]
      },
      {
       "id": "r4",
       "label": [
        "D11"
       ],
       "values": [
        2.98,
        2.394,
        2.13,
        1.865,
        1.574,
        0.597
       ]
      },
      {
       "id": "r5",
       "label": [
        "D5"
       ],
       "values": [
        2.737,
        2.198,
        1.751,
        1.742,
        1.475,
        0.568
       ]
      },
      {
       "id": "r6",
       "label": [
        "D6"
       ],
       "values": [
        2.784,
        2.241,
        1.798,
        1.769,
        1.498,
        0.583
       ]
      }
     ],
     "source_spans": [
      "p10.11-p10.20"
     ]
    }
   ],
   "source_spans": [
    "p9.22-p10.9"
   ],
   "setup": {
    "studied": [
     "D17",
     "D9",
     "D10",
     "D11",
     "D5",
     "D6"
    ],
    "cases": [
     "D32"
    ],
    "procedure": [
     "D37"
    ],
    "lens": [
     "D15",
     "D14",
     "D16"
    ]
   },
   "availability": {
    "level": "public"
   }
  }
 ],
 "limitations": [
  {
   "id": "L3",
   "category": "sample_size",
   "statement": "The effect of the length of a time series on the accuracy of CES is not analysed.",
   "affects": [
    "C3",
    "C4",
    "C11",
    "C12"
   ],
   "impact": "reduces_transferability",
   "mitigation": "none",
   "source_spans": [
    "p12.1"
   ],
   "provenance": "author_stated"
  },
  {
   "id": "L5",
   "category": "sampling",
   "statement": "The 10 M4 series on which ETS broke down were excluded from the M4 comparison, so the series were selected on the failure of one of the compared methods.",
   "affects": [
    "C4",
    "C12"
   ],
   "impact": "could_bias_magnitude",
   "mitigation": "none",
   "source_spans": [
    "p10.8"
   ],
   "provenance": "author_stated"
  }
 ],
 "artefacts": [
  {
   "id": "A4",
   "kind": "software",
   "description": "smooth package for R (ces, auto.ces, es, auto.ssarima)",
   "uri": "https://cran.r-project.org/src/contrib/Archive/smooth/smooth_3.1.2.tar.gz",
   "version": "3.1.2",
   "reference": "B10",
   "availability": {
    "level": "public"
   },
   "archive_uri": "https://cran.r-project.org/src/contrib/Archive/smooth/smooth_3.1.2.tar.gz"
  },
  {
   "id": "A5",
   "kind": "software",
   "description": "forecast package for R (thetaf)",
   "version": "8.14",
   "reference": "B11",
   "availability": {
    "level": "public"
   },
   "uri": "https://cran.r-project.org/src/contrib/Archive/forecast/forecast_8.14.tar.gz"
  },
  {
   "id": "A7",
   "kind": "data",
   "description": "M4 competition data",
   "reference": "B9",
   "availability": {
    "level": "public"
   }
  },
  {
   "id": "A8",
   "kind": "software",
   "description": "R, the environment for statistical computing in which the experiments were run",
   "uri": "https://www.R-project.org/",
   "reference": "B24",
   "availability": {
    "level": "public"
   }
  }
 ],
 "references": [
  {
   "id": "B2",
   "id_type": "doi",
   "target": "10.1016/S0169-2070(01)00110-8"
  },
  {
   "id": "B9",
   "id_type": "doi",
   "target": "10.1016/j.ijforecast.2019.04.014"
  },
  {
   "id": "B10",
   "id_type": "url",
   "target": "https://github.com/config-i1/smooth",
   "archive_uri": "https://cran.r-project.org/src/contrib/Archive/smooth/smooth_3.1.2.tar.gz"
  },
  {
   "id": "B11",
   "id_type": "doi",
   "target": "10.18637/jss.v027.i03"
  },
  {
   "id": "B13",
   "id_type": "doi",
   "target": "10.1016/j.ijforecast.2019.01.006"
  },
  {
   "id": "B15",
   "id_type": "doi",
   "target": "10.1080/00207543.2019.1600764"
  },
  {
   "id": "B16",
   "id_type": "doi",
   "target": "10.1016/S0169-2070(00)00066-2"
  },
  {
   "id": "B17",
   "id_type": "doi",
   "target": "10.1016/j.ijforecast.2021.11.013"
  },
  {
   "id": "B18",
   "id_type": "doi",
   "target": "10.1016/j.ijforecast.2006.03.001"
  },
  {
   "id": "B19",
   "id_type": "doi",
   "target": "10.1198/016214506000001437"
  },
  {
   "id": "B24",
   "id_type": "reference",
   "target": "R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/"
  }
 ]
}
