{
 "bundle_of": "C10",
 "schema_version": "0.44",
 "paper": {
  "id": "entrylib:2026.000013",
  "version": 3,
  "title": "What does sMAPE reward? An analytical and simulation study",
  "licence": "CC-BY-4.0",
  "status": "corrected",
  "date": "2026-11-18",
  "status_notice": {
   "date": "2026-11-18",
   "reason_code": "correction",
   "statement": "Version 3 answers a reader's comments. Corrected: C3 no longer says sMAPE is minimised by the median for small errors, only near it, with a bound; C4 keeps the weighted median, and C12 now holds what was said about the ordinary median; the measure called msMAPE in version 2 is the Suilin loss. Added: how well a test set tells two forecasts apart as the actuals spread out (C18); the large-sample distribution of the sMAPE minimiser (C19); Smyl's msMAPE (C20) and what it rewards on intermittent demand (C21). Also added: sections, the definition of sMAPE first, and a limitation on C6.",
   "issued_by": "Ivan Svetunkov (ORCID 0000-0001-7826-0281)"
  }
 },
 "claims": [
  {
   "id": "C10",
   "name": "Expected sMAPE flattens as the actuals spread out",
   "summary": "For widely spread actuals, every forecast scores close to the worst value. The differences between forecasts shrink, and the forecast sMAPE rewards moves from the median to the most likely value on the log scale.",
   "statement": "Let $\\ln y = \\mu + \\sigma\\varepsilon$, where $\\varepsilon$ has a fixed density $h$ with bounded continuous first and second derivatives, and write a forecast as $f = e^{\\mu + \\sigma c}$ (its quantile level, the probability that $\\varepsilon \\le c$, does not depend on $\\sigma$). As $\\sigma$ grows, $\\mathrm E[s(y,f)] = 1 - 4\\ln 2\\,h(c)/\\sigma + O(\\sigma^{-3})$ uniformly in $c$. (a) The [[D3:expected sMAPE]] of every forecast tends to its bound 200, and the difference between forecasts at two fixed quantile levels shrinks like $1/\\sigma$. (b) When $h$ has a single highest point $c_0$, the quantile level of the [[D6:sMAPE target]] tends to that of $c_0$, the mode of $\\ln y$, which differs from 1/2 when the mode of $\\varepsilon$ is not its median. (c) As $\\sigma \\to 0$, when $\\varepsilon$ has a unique median and a finite $\\mathrm E|\\varepsilon|^3$, the quantile level of the target tends to 1/2: sMAPE targets the median for narrow distributions and the mode of $\\ln y$ for wide ones.",
   "about": {
    "subject": "D3",
    "relation": "flattens towards its bound as the spread of log actuals grows, with",
    "object": "D6",
    "comparator": null,
    "conditions": []
   },
   "type": "formal",
   "role": "principal",
   "novelty": "no_prior_found",
   "support": [
    {
     "kind": "formal_proof",
     "evidence": "E28"
    },
    {
     "kind": "empirical_quant",
     "evidence": "E1"
    }
   ],
   "provenance": "author_stated",
   "hypothesis": {
    "lab_event": "H30",
    "outcome": "supported"
   },
   "planning": "confirmatory",
   "planned_at": "2026-11-17T09:21:09Z",
   "planned_at_anchor": "git_remote",
   "test": {
    "framework": "formal",
    "supported": "The proof of the hypothesis, as now worded, in the study's final proof document (proofs.md, AR1.R2) is complete, every step naming what it rests on; the independent checker AC4 has a check of the current wording with outcome confirmed; and the deciding run's numerical checks of its formulas meet the tolerances of PL5: identities on random pairs within 1e-9 (on the scale of sMAPE), quantities computed by quadrature within 1e-4 relative (or absolute, for a residual the proof says is zero), counts of violations 0, and the expansion's error times sigma^3 below the proof's bound.",
    "not_supported": "A counterexample to the statement as now worded, found by the proof, a checker, the supervisor or the deciding run, that a correction of the proof cannot remove.",
    "verdict": "judged",
    "runs": 1,
    "lab_plan": "PL9"
   },
   "checks": [
    {
     "by": "P3",
     "outcome": "confirmed",
     "extent": "full",
     "notes": "Derived from the statement alone (integral of 2/(1+e^|v|) is 4 ln 2, odd term vanishes, remainder at most 3 zeta(3) sup|g''|/sigma^3 uniformly in c; checked numerically for the normal), then compared with the final proofs.md (paper H11), which agrees; the constant 4 ln 2 is my earlier 800 ln 2 rescaled. (b) and (c) are claims about the quantile level, which is what holds in general."
    },
    {
     "by": "P2",
     "outcome": "confirmed",
     "extent": "full",
     "notes": "The supervisor read the proof of paper H11 in proofs.md (AR1.R2; rendered as AR2.R2) and confirmed it: first \"Confirmed\" (DC6), then, after the last corrections of paper H4, H5 and H9, \"All looks fine now. NExt\"."
    }
   ],
   "verification": "agent_checked",
   "relations": [
    {
     "type": "supports",
     "note": "C6 recommends against sMAPE for point forecasts; C10 shows that on widely spread actuals every forecast scores close to the worst value and the target moves to the mode of ln y.",
     "reference": "B6"
    },
    {
     "type": "supports",
     "reference": "B13",
     "note": "Kolassa (2026) concludes that sMAPE should not be used; C10 shows that on widely spread actuals every forecast scores close to the worst value."
    }
   ],
   "depends_on": [
    "C3"
   ],
   "verifiability": "open",
   "verifiability_flags": []
  },
  {
   "id": "C1",
   "name": "sMAPE depends only on the ratio of actual to forecast",
   "summary": "Only the ratio of the actual to the forecast matters to sMAPE, not their size.",
   "statement": "For $y > 0$ and $f > 0$, the [[D1:sMAPE term]] $s(y,f) = |y-f|/(|y|+|f|)$ depends on $y$ and $f$ only through their ratio $r = y/f$: $s = |r - 1|/(r + 1)$.",
   "about": {
    "subject": "D1",
    "relation": "depends only on",
    "object": "the ratio of actual to forecast",
    "comparator": null,
    "conditions": []
   },
   "type": "formal",
   "role": "principal",
   "novelty": "extension",
   "support": [
    {
     "kind": "formal_proof",
     "evidence": "E18"
    },
    {
     "kind": "empirical_quant",
     "evidence": "E1"
    }
   ],
   "provenance": "author_stated",
   "hypothesis": {
    "lab_event": "H21",
    "outcome": "supported"
   },
   "planning": "confirmatory",
   "planned_at": "2026-11-17T09:21:09Z",
   "planned_at_anchor": "git_remote",
   "test": {
    "framework": "formal",
    "supported": "The proof of the hypothesis, as now worded, in the study's final proof document (proofs.md, AR1.R2) is complete, every step naming what it rests on; the independent checker AC4 has a check of the current wording with outcome confirmed; and the deciding run's numerical checks of its formulas meet the tolerances of PL5: identities on random pairs within 1e-9 (on the scale of sMAPE), quantities computed by quadrature within 1e-4 relative (or absolute, for a residual the proof says is zero), counts of violations 0, and the expansion's error times sigma^3 below the proof's bound.",
    "not_supported": "A counterexample to the statement as now worded, found by the proof, a checker, the supervisor or the deciding run, that a correction of the proof cannot remove.",
    "verdict": "judged",
    "runs": 1,
    "lab_plan": "PL9"
   },
   "checks": [
    {
     "by": "P3",
     "outcome": "confirmed",
     "extent": "full",
     "notes": "Derived from the statement alone (first wording, with the factor 200), then compared with the final proofs.md (paper H1), which agrees after rescaling. No conditions needed."
    },
    {
     "by": "P2",
     "outcome": "confirmed",
     "extent": "full",
     "notes": "The supervisor read the proof of paper H1 in proofs.md (AR1.R2; rendered as AR2.R2) and confirmed it: first \"Confirmed\" (DC6), then, after the last corrections of paper H4, H5 and H9, \"All looks fine now. NExt\"."
    }
   ],
   "verification": "agent_checked",
   "relations": [
    {
     "type": "extends",
     "reference": "B12",
     "note": "Tofallis (2015) lists the sMAPE term, |y-f|/[(|y|+|f|)/2], among Tornqvist et al.'s measures of relative change; C1 shows it depends on the actual and the forecast only through their ratio."
    }
   ],
   "verifiability": "open",
   "verifiability_flags": []
  },
  {
   "id": "C3",
   "name": "sMAPE is a hyperbolic tangent of the log error",
   "summary": "sMAPE measures the error on the log scale. For small errors it is half the absolute log error; for large errors it levels off. For small errors its best forecast is close to the median, but not in general the median.",
   "statement": "For $y > 0$ and $f > 0$, the [[D1:sMAPE term]] is $s = \\tanh(|\\ln y - \\ln f|/2)$ ([[D4:log error]] $\\ln y - \\ln f$). (a) It is 0 when $y = f$, strictly increasing and concave in $|\\ln y - \\ln f|$, and bounded by 1, which it approaches as $|\\ln y - \\ln f|$ grows. (b) $0 \\le \\tfrac12|\\ln y - \\ln f| - s \\le \\tfrac1{24}|\\ln y - \\ln f|^3$; so when all log errors are small, sMAPE is about 100 times the mean absolute error of $\\ln f$ as a forecast of $\\ln y$ ([[D19:log-MAE]]). (c) The log-MAE of a test set is minimised by its median; sMAPE is then minimised near the median, not in general at it. For positive actuals $y_1, \\dots, y_n$ with a median $m$, a forecast $f^*$ that minimises their sMAPE satisfies\n$$\\frac1n\\sum_t|\\ln y_t - \\ln f^*| - \\frac1n\\sum_t|\\ln y_t - \\ln m| \\le \\frac1{12}\\cdot\\frac1n\\sum_t|\\ln y_t - \\ln f^*|^3:$$\nits mean absolute log error exceeds the smallest possible by at most a twelfth of its mean cubed absolute log error. The minimiser is the median itself only under further conditions on the distribution of the actuals.",
   "about": {
    "subject": "D1",
    "relation": "is a bounded function of",
    "object": "D4",
    "comparator": null,
    "conditions": []
   },
   "type": "formal",
   "role": "principal",
   "novelty": "extension",
   "support": [
    {
     "kind": "formal_proof",
     "evidence": "E20"
    },
    {
     "kind": "empirical_quant",
     "evidence": "E31"
    }
   ],
   "provenance": "author_stated",
   "hypothesis": {
    "lab_event": "H36",
    "outcome": "supported"
   },
   "planning": "confirmatory",
   "planned_at": "2026-11-18T12:02:12Z",
   "planned_at_anchor": "git_remote",
   "test": {
    "framework": "formal",
    "supported": "The proof of the hypothesis as worded here, in the study's proof documents (proofs.md, proof-suilin.md), is complete, every step naming what it rests on; the independent checker (an agent of another model than the author's: Claude Opus 5.5, registered as an actor before it checks) records a check of the hypothesis as worded with outcome confirmed; and the deciding run's numerical checks of its formulas meet the tolerances of the design.",
    "not_supported": "A counterexample to the hypothesis as worded, found by the proof, the checker, the supervisor or the deciding run, that a correction of the proof cannot remove.",
    "verdict": "judged",
    "runs": 1,
    "lab_plan": "PL12"
   },
   "checks": [
    {
     "by": "P3",
     "outcome": "confirmed",
     "extent": "full",
     "notes": "Derived from the statement alone (earlier wording: old H3 and H9), then compared with the final proofs.md (paper H3), which agrees; the bound in (b) follows from x - x^3/3 <= tanh x <= x. Concavity is in the absolute log error on [0, inf); 'minimised by the median' is the small-error approximation, as stated."
    },
    {
     "by": "P2",
     "outcome": "confirmed",
     "extent": "full",
     "notes": "The supervisor read the proof of paper H3 in proofs.md (AR1.R2; rendered as AR2.R2) and confirmed it: first \"Confirmed\" (DC6), then, after the last corrections of paper H4, H5 and H9, \"All looks fine now. NExt\"."
    },
    {
     "by": "P5",
     "outcome": "confirmed",
     "extent": "full",
     "notes": "Derived independently. With $z=\\ln y-\\ln f$, $s=|e^z-1|/(e^z+1)=\\tanh(|z|/2)$. (a) $\\tanh(u/2)$ on $u\\ge0$ is 0 at 0, strictly increasing, concave (derivative $\\tfrac12\\operatorname{sech}^2(u/2)$ positive and decreasing), below 1 with limit 1. (b) $x-x^3/3\\le\\tanh x\\le x$ for $x\\ge0$ (the lower bound: $\\frac{d}{dx}(\\tanh x-x+x^3/3)=x^2-\\tanh^2x\\ge0$); with $x=|z|/2$ this gives $0\\le|z|/2-s\\le|z|^3/24$, so $200\\cdot\\mathrm{mean}(s)\\approx100\\cdot\\mathrm{mean}|z|$. (c) Averaging (b): $\\bar s\\le a/2\\le\\bar s+k/24$; with $\\bar s(f^*)\\le\\bar s(m)$, $a(f^*)/2\\le\\bar s(f^*)+k(f^*)/24\\le\\bar s(m)+k(f^*)/24\\le a(m)/2+k(f^*)/24$, which is the stated bound; the log-MAE is minimised at any median of the $\\ln y_t$, i.e. at any median $m$. The proof (proofs.md H3) does exactly this; its only unproved step is the elementary inequality $x-x^3/3\\le\\tanh x$, closed by the derivative argument above. I recomputed the counterexample of step 8 (actuals $1,1,e,e^4,e^4$: sums 2.734 at the median $e$ against 2.390 at 1), which supports \"not in general at it\". Every part holds as worded."
    }
   ],
   "verification": "agent_checked",
   "depends_on": [
    "C1"
   ],
   "relations": [
    {
     "type": "extends",
     "reference": "B12",
     "note": "Tofallis (2015) relates sMAPE to the log accuracy ratio among relative measures; C3 shows the sMAPE term is exactly the hyperbolic tangent of half the absolute log ratio."
    }
   ],
   "verifiability": "open",
   "verifiability_flags": []
  }
 ],
 "definitions": [
  {
   "id": "D2",
   "term": "sMAPE",
   "kind": "variable",
   "definition": "Symmetric mean absolute percentage error of a point forecast $f$ over a test set of actuals $y_1, \\dots, y_n$: $\\frac{200}{n}\\sum_t |y_t - f|/(|y_t|+|f|)$, where a term with $y_t = f = 0$ counts as 0. Introduced as the symmetric MAPE by Makridakis (1993), whose denominator had no absolute values; for positive actuals and forecasts the two agree.",
   "source": "B8"
  },
  {
   "id": "D1",
   "term": "sMAPE term",
   "kind": "variable",
   "definition": "The contribution of one actual $y$ and one point forecast $f$ to [[D2:sMAPE]]: $s(y,f) = |y-f|/(|y|+|f|)$, between 0 and 1, with $s = 0$ when $y = f$ (including $y = f = 0$, where the fraction is $0/0$).",
   "symbol": "$s(y,f)$"
  },
  {
   "id": "D3",
   "term": "expected sMAPE",
   "kind": "variable",
   "definition": "The expected sMAPE of a fixed forecast $f$ under the distribution of the actual $y$: $200\\,\\mathrm{E}[s(y, f)]$, where $\\mathrm E$ is the expectation over the random actual $y$. Computed from the distribution (by integration or exact sums), never from a sample."
  },
  {
   "id": "D6",
   "term": "sMAPE target",
   "kind": "concept",
   "definition": "The value or values of $f$ that minimise expected sMAPE for a given distribution of the actual $y$, $\\arg\\min_f \\mathrm E[s(y,f)]$: the functional sMAPE rewards. It may be a single value or several."
  },
  {
   "id": "D4",
   "term": "log error",
   "kind": "variable",
   "definition": "For a positive actual $y$ and a positive forecast $f$, $z = \\ln y - \\ln f$.",
   "symbol": "$z$"
  },
  {
   "id": "D5",
   "term": "weight",
   "kind": "variable",
   "definition": "For a positive actual $y$ and a positive forecast $f$, $w(y,f) = y/(y+f)^2$: how much the actual pulls the forecast that minimises expected sMAPE, since the derivative of $\\mathrm E[s(y,f)]$ in $f$ is $2\\,\\mathrm E[w(y,f)(\\mathbf 1\\{y<f\\} - \\mathbf 1\\{y>f\\})]$. Largest at $y = f$, the same for $y/f = r$ and $1/r$.",
   "symbol": "$w(y,f)$"
  },
  {
   "id": "D10",
   "term": "RMSE",
   "kind": "variable",
   "definition": "Root mean squared error of a point forecast $f$ over a test set of actuals $y_1, \\dots, y_n$: $\\sqrt{\\frac1n\\sum_t (y_t-f)^2}$."
  },
  {
   "id": "D11",
   "term": "MAE",
   "kind": "variable",
   "definition": "Mean absolute error of a point forecast $f$ over a test set of actuals $y_1, \\dots, y_n$: $\\frac1n\\sum_t |y_t-f|$."
  },
  {
   "id": "D19",
   "term": "log-MAE",
   "kind": "variable",
   "definition": "100 times the mean absolute error of $\\ln f$ as a forecast of $\\ln y$: $\\frac{100}{n}\\sum_t |\\ln y_t - \\ln f|$. Unbounded; sMAPE's small-error limit (C3); minimised by the median."
  },
  {
   "id": "D28",
   "term": "Suilin term",
   "kind": "variable",
   "definition": "The contribution of one actual $y$ and one forecast $f$ to the Suilin loss:\n$$\\tilde s(y,f) = \\frac{|y-f|}{\\max(|y|+|f|+\\delta,\\ 0.5+\\delta)}, \\qquad \\delta = 0.1;$$\nthe Suilin loss is 200 times its mean over a test set. It replaces the denominator $|y|+|f|$ of the [[D1:sMAPE term]]. Proposed by Suilin (2017) in code for a Kaggle competition, as a smoothed variant of sMAPE; defined in print by Hewamalage, Bergmeir and Bandara (2021, Eq. 15), who write the constant $\\delta$ as $\\epsilon$ and the factor 2 into the measure differently (a constant factor changes no property here). Hewamalage, Ackermann and Bergmeir (2023) call it msMAPE; Smyl (2025) gives that name to a different measure, so this study follows Kolassa (2026) and calls it the Suilin loss. Version 2 of this paper called it msMAPE.",
   "source": "B14",
   "symbol": "$\\tilde s(y,f)$"
  },
  {
   "id": "D33",
   "term": "msMAPE term",
   "kind": "variable",
   "definition": "The contribution of one actual $y \\ge 0$ and one forecast $f \\ge 0$ to the sparse-proof sMAPE (msMAPE) of Smyl (2025), as Kolassa (2026, Definition 4) states it, for a small constant $\\nu > 0$ (Kolassa writes it $m$, which is the median here):\n$$s_\\nu(y,f) = \\frac{|f-y|}{f+\\max\\{y, \\nu\\}},$$\nwhich is 0 when $y = f = 0$; msMAPE is 200 times its mean over a test set (Kolassa writes the factor 2 into the term). It replaces the actual in the denominator of the [[D1:sMAPE term]] by $\\nu$ when the actual is below $\\nu$.",
   "source": "B13",
   "symbol": "$s_\\nu(y,f)$"
  },
  {
   "id": "D8",
   "term": "sMAPE minimiser of a test set",
   "kind": "variable",
   "definition": "The forecast that minimises the sMAPE of one test set $y_1, \\dots, y_n$, $\\hat f_n = \\arg\\min_f \\frac1n\\sum_t s(y_t, f)$, found numerically (the sample counterpart of the [[D6:sMAPE target]]).",
   "symbol": "$\\hat f_n$"
  },
  {
   "id": "D9",
   "term": "relative efficiency",
   "kind": "variable",
   "definition": "For a test-set size $n$ and an outcome distribution where sMAPE and MAE both target the median: $\\mathrm{RE}(n) = \\operatorname{Var}(\\ln \\hat m_n) / \\operatorname{Var}(\\ln \\hat f_n)$ over test sets, where $\\hat m_n$ is the sample median (the MAE minimiser) and $\\hat f_n$ the sMAPE minimiser of the same test set. Above 1, sMAPE's minimiser varies less; below 1, MAE's does.",
   "symbol": "$\\mathrm{RE}(n)$"
  },
  {
   "id": "D32",
   "term": "signal-to-noise ratio of a comparison",
   "kind": "variable",
   "definition": "For a measure with term $L(y,f)$, two forecasts $f_1$ and $f_2$ and a distribution of the actual $y$: $|\\mathrm E[D]|/\\mathrm{sd}(D)$, where $D = L(y,f_2) - L(y,f_1)$ is the difference of the two forecasts' scores on one actual. Over a test set of $n$ independent actuals the mean of $D$ is $\\sqrt n$ times this many of its standard errors from zero: the larger the ratio, the smaller the test set that tells the two forecasts apart."
  },
  {
   "id": "D31",
   "term": "formula checks",
   "kind": "method",
   "definition": "Numerical checks of the formulas of the formal claims (claimset-code/smape-target/check_formulas.py, seed 6; check_jump.py for C11): identities and inequalities on 200000 random pairs, derivatives against finite differences, minimisers, balances and thresholds against quadrature and bisection on every setting of the selection design, and the large-spread expansion against its bound. For the claims new or reworded in version 3 (C3, C4, C12, C17 to C20): check_v3.py (seed 8), check_h39.py and check_corrected.py (seed 8), in the same way."
  }
 ],
 "evidence": [
  {
   "id": "E1",
   "profile": "quant",
   "design": "other",
   "description": "Numerical checks of the formulas of the formal claims ([[D31:formula checks]]), lab run RN8 under the formal plan that decided them; every check is within the plan's tolerances. They check the formulas on concrete numbers; the proofs are the evidence for the claims. The rows are labelled by the hypotheses of the record's first plan, whose wording kept the factor 200: H1 checks C1; H2 and H10 check C2; H5 checks C5; H7 checks C6; H8 checks C7; H12 checks C8; H13 checks C9; H14 checks C10. H3 and H9 checked C3, and H4 and H6 checked C4, as version 2 worded them; C3, C4 and C12 as now worded are checked in E31. The two results E1.R1 and E1.R2 compare a simulated variance with the large-sample variance that C19 now proves.",
   "availability": {
    "level": "public"
   },
   "lab_runs": [
    "RN8"
   ],
   "results": [
    {
     "id": "R1",
     "variable": "simulated n Var(ln sMAPE minimiser) / large-sample value, lognormal sigma=1",
     "value": 0.9944,
     "interval": {
      "level": 0.99,
      "lower": 0.9353981515994695,
      "upper": 1.0598078456840363,
      "kind": "confidence"
     },
     "conditions": {
      "hypothesis": "H11",
      "test-set size": 1000,
      "test sets": 4000
     }
    },
    {
     "id": "R2",
     "variable": "simulated n Var(ln sMAPE minimiser) / large-sample value, lognormal sigma=1",
     "value": 1.022,
     "interval": {
      "level": 0.99,
      "lower": 0.9396347019298308,
      "upper": 1.0981626206556903,
      "kind": "confidence"
     },
     "conditions": {
      "hypothesis": "H11",
      "test-set size": 10000,
      "test sets": 2000
     }
    },
    {
     "id": "R3",
     "variable": "maximum absolute difference from the ratio form",
     "value": 8.527e-14,
     "conditions": {
      "hypothesis": "H1"
     }
    },
    {
     "id": "R4",
     "variable": "maximum absolute difference between the ratio and its reciprocal",
     "value": 8.527e-14,
     "conditions": {
      "hypothesis": "H2"
     }
    },
    {
     "id": "R5",
     "variable": "maximum absolute difference from 200 tanh(|z|/2)",
     "value": 8.527e-14,
     "conditions": {
      "hypothesis": "H3"
     }
    },
    {
     "id": "R6",
     "variable": "violations of monotonicity, concavity or the bound 200 on 0 <= |z| <= 12",
     "value": 0,
     "conditions": {
      "hypothesis": "H3"
     }
    },
    {
     "id": "R7",
     "variable": "violations of 0 <= 100|z| - s <= 25|z|^3/3",
     "value": 0,
     "conditions": {
      "hypothesis": "H9"
     }
    },
    {
     "id": "R8",
     "variable": "maximum of |s - 100|z| + 25|z|^3/3| / |z|^5 for 0.05 < |z| < 0.5",
     "value": 0.8331,
     "conditions": {
      "hypothesis": "H9"
     }
    },
    {
     "id": "R9",
     "variable": "violations of s(y, y+e) < s(y, y-e)",
     "value": 0,
     "conditions": {
      "hypothesis": "H10"
     }
    },
    {
     "id": "R10",
     "variable": "maximum absolute difference between s(y, ky) and s(y, y/k)",
     "value": 1.137e-13,
     "conditions": {
      "hypothesis": "H10"
     }
    },
    {
     "id": "R11",
     "variable": "maximum relative difference between the formula for S' and a numerical derivative",
     "value": 4.823e-08,
     "conditions": {
      "hypothesis": "H4"
     }
    },
    {
     "id": "R12",
     "variable": "maximum distance in ln f between the minimiser and the median",
     "value": 8.641e-08,
     "conditions": {
      "hypothesis": "H5"
     }
    },
    {
     "id": "R13",
     "variable": "number of local minima, lognormal sigma=1 and log-Laplace b=1",
     "value": 1,
     "conditions": {
      "hypothesis": "H5"
     }
    },
    {
     "id": "R14",
     "variable": "maximum balance residual at the minimiser (pull below minus pull above)",
     "value": 8.633e-09,
     "conditions": {
      "hypothesis": "H6"
     }
    },
    {
     "id": "R15",
     "variable": "settings with one local minimum where the sign of D misplaces the target",
     "value": 0,
     "conditions": {
      "hypothesis": "H6"
     }
    },
    {
     "id": "R16",
     "variable": "maximum distance between the one-step and the exact quantile level",
     "value": 0.002514,
     "conditions": {
      "hypothesis": "H6"
     }
    },
    {
     "id": "R17",
     "variable": "2 g(m) - E[sech^2(Z/2) tanh(|Z|/2)] for the equal-mode bimodal setting (negative: median a local maximum)",
     "value": -0.3475,
     "conditions": {
      "hypothesis": "H7"
     }
    },
    {
     "id": "R18",
     "variable": "number of local minima of the equal-mode bimodal setting",
     "value": 2,
     "conditions": {
      "hypothesis": "H7"
     }
    },
    {
     "id": "R19",
     "variable": "difference between the two minimal expected sMAPEs, equal-mode bimodal setting",
     "value": 1.421e-14,
     "conditions": {
      "hypothesis": "H7"
     }
    },
    {
     "id": "R20",
     "variable": "maximum distance between the formula and the bisected threshold",
     "value": 5.551e-17,
     "conditions": {
      "hypothesis": "H8"
     }
    },
    {
     "id": "R21",
     "variable": "opposite-sign or one-zero pairs not scoring 200",
     "value": 0,
     "conditions": {
      "hypothesis": "H8"
     }
    },
    {
     "id": "R22",
     "variable": "relative distance between the 1/y-weighted median and the minimiser of expected MAPE",
     "value": 6.745e-05,
     "conditions": {
      "hypothesis": "H12"
     }
    },
    {
     "id": "R23",
     "variable": "difference of expected squared errors minus (f1-f2)(f1+f2-2E[y])",
     "value": 0.0,
     "conditions": {
      "hypothesis": "H12"
     }
    },
    {
     "id": "R24",
     "variable": "sMAPE terms outside [0, 200]",
     "value": 0,
     "conditions": {
      "hypothesis": "H13"
     }
    },
    {
     "id": "R25",
     "variable": "expected sMAPE at f = 1e-6 and f = 1e10 (both near 200)",
     "value": 200.0,
     "conditions": {
      "hypothesis": "H13"
     }
    },
    {
     "id": "R26",
     "variable": "the proof's bound on the error times sigma^3, 600 zeta(3) max|g''|, log-Gumbel",
     "value": 309.9,
     "conditions": {
      "hypothesis": "H14"
     }
    },
    {
     "id": "R27",
     "variable": "limit of sigma^3 times the relative efficiency for normal log-outcomes, from the proof",
     "value": 7.226,
     "conditions": {
      "hypothesis": "H14"
     }
    }
   ],
   "tables": [
    {
     "id": "T1",
     "description": "zero threshold by setting; hypothesis: H8 (lab runs RN8)",
     "conditions": {
      "hypothesis": "H8"
     },
     "row_labels": [
      "setting"
     ],
     "columns": [
      "zero threshold"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "lognormal sigma=0.3"
       ],
       "values": [
        0.4687
       ]
      },
      {
       "id": "r2",
       "label": [
        "lognormal sigma=0.5"
       ],
       "values": [
        0.4469
       ]
      },
      {
       "id": "r3",
       "label": [
        "lognormal sigma=1"
       ],
       "values": [
        0.394
       ]
      },
      {
       "id": "r4",
       "label": [
        "lognormal sigma=2"
       ],
       "values": [
        0.3075
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T2",
     "description": "maximum |S - (200 - 800 ln2 h)| times sigma^3, log-Gumbel right, sigma^3 times the asymptotic relative efficiency, lognormal by sigma; hypothesis: H14 (lab runs RN8)",
     "conditions": {
      "hypothesis": "H14"
     },
     "row_labels": [
      "sigma"
     ],
     "columns": [
      "maximum |S - (200 - 800 ln2 h)| times sigma^3, log-Gumbel right",
      "sigma^3 times the asymptotic relative efficiency, lognormal"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        5.0
       ],
       "values": [
        264.5,
        5.613
       ]
      },
      {
       "id": "r2",
       "label": [
        10.0
       ],
       "values": [
        292.3,
        6.743
       ]
      },
      {
       "id": "r3",
       "label": [
        20.0
       ],
       "values": [
        300.1,
        7.099
       ]
      },
      {
       "id": "r4",
       "label": [
        40.0
       ],
       "values": [
        302.1,
        7.193
       ]
      }
     ],
     "artefact": "A6"
    }
   ],
   "artefacts": [
    "A6"
   ],
   "execution": {
    "code_version": "fa0ce626f18bd1ca22076d62452d2b68c484a297",
    "software": {
     "python": "3.13.18",
     "platform": "Linux-6.18.44-fc-v553-x86_64-with-glibc2.39"
    }
   },
   "setup": {
    "studied": [
     "D1",
     "D3",
     "D4",
     "D5",
     "D6"
    ],
    "procedure": [
     "D31"
    ]
   }
  },
  {
   "id": "E31",
   "profile": "quant",
   "design": "other",
   "description": "Numerical checks of the formulas of the claims new or reworded in version 3 ([[D31:formula checks]]), lab run RN13 (check_v3.py, seed 8) under the formal plan that decided C3, C4, C12, C17 and C20; every check is within the plan's tolerances. They check the formulas on concrete numbers; the proofs are the evidence for the claims. The tables are labelled by the hypotheses the run's script names: H3 (T1 and T2) checks C3; H4 (T3) checks C4; H34 (T4 and T5) checks C12; H35 (T6 and T7) checks C17; H42 (T8 and T9) checks C20. The tables of H39 (T10 to T12) and H40 (T13 to T17) check the formulas of C18 and C19, whose wording was decided after later checks (E32, E33).",
   "availability": {
    "level": "public"
   },
   "lab_runs": [
    "RN13"
   ],
   "tables": [
    {
     "id": "T1",
     "description": "violations of the bound on the log-MAE of the sMAPE minimiser, largest excess of the minimiser's log-MAE over the median's, minus the bound; hypothesis: H3, test_sets: 2000 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H3",
      "test_sets": 2000
     },
     "row_labels": [],
     "columns": [
      "violations of the bound on the log-MAE of the sMAPE minimiser",
      "largest excess of the minimiser's log-MAE over the median's, minus the bound"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        0,
        -1.263e-06
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "description": "sum of sMAPE terms, forecast at the median, actuals 1, 1, e, e^4, e^4, sum of sMAPE terms, forecast at 1, actuals 1, 1, e, e^4, e^4; hypothesis: H3 (lab runs RN13)",
     "row_labels": [],
     "columns": [
      "sum of sMAPE terms, forecast at the median, actuals 1, 1, e, e^4, e^4",
      "sum of sMAPE terms, forecast at 1, actuals 1, 1, e, e^4, e^4"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        2.735,
        2.39
       ]
      }
     ],
     "conditions": {
      "hypothesis": "H3"
     },
     "artefact": "A6",
     "id": "T2"
    },
    {
     "description": "largest difference of the weights below and above the target, over their sum; hypothesis: H4, distributions: 6 (lab runs RN13)",
     "row_labels": [],
     "columns": [
      "largest difference of the weights below and above the target, over their sum"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        2.089e-08
       ]
      }
     ],
     "conditions": {
      "hypothesis": "H4",
      "distributions": 6
     },
     "artefact": "A6",
     "id": "T3"
    },
    {
     "description": "distributions with more than one local minimum, distributions where the target is not on the side of the median that carries more weight; hypothesis: H34, distributions: 6 (lab runs RN13)",
     "row_labels": [],
     "columns": [
      "distributions with more than one local minimum",
      "distributions where the target is not on the side of the median that carries more weight"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        0,
        0
       ]
      }
     ],
     "conditions": {
      "hypothesis": "H34",
      "distributions": 6
     },
     "artefact": "A6",
     "id": "T4"
    },
    {
     "description": "largest distance in logs between the target and the median, lognormal and log-Laplace; hypothesis: H34, distributions: 16 (lab runs RN13)",
     "row_labels": [],
     "columns": [
      "largest distance in logs between the target and the median, lognormal and log-Laplace"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        1.753e-07
       ]
      }
     ],
     "conditions": {
      "hypothesis": "H34",
      "distributions": 16
     },
     "artefact": "A6",
     "id": "T5"
    },
    {
     "description": "Suilin term of a zero forecast of a zero actual; hypothesis: H35 (lab runs RN13)",
     "row_labels": [],
     "columns": [
      "Suilin term of a zero forecast of a zero actual"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        0.0
       ]
      }
     ],
     "conditions": {
      "hypothesis": "H35"
     },
     "artefact": "A6",
     "id": "T6"
    },
    {
     "id": "T7",
     "description": "largest error of the identity of (a), largest error of the identity of (b), values of the Suilin term outside [0, 1); hypothesis: H35, pairs: 200000 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H35",
      "pairs": 200000
     },
     "row_labels": [],
     "columns": [
      "largest error of the identity of (a)",
      "largest error of the identity of (b)",
      "values of the Suilin term outside [0, 1)"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        1.11e-16,
        0.0,
        0
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T8",
     "description": "largest error of the identity of (a), largest error of the identity of (b), values of the msMAPE term outside [0, 1], values of the msMAPE term not below 1 for actuals below nu, pairs with a count of at least 1 where the msMAPE and sMAPE terms differ by nu; hypothesis: H42, pairs: 200000 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H42",
      "pairs": 200000
     },
     "row_labels": [
      "nu"
     ],
     "columns": [
      "largest error of the identity of (a)",
      "largest error of the identity of (b)",
      "values of the msMAPE term outside [0, 1]",
      "values of the msMAPE term not below 1 for actuals below nu",
      "pairs with a count of at least 1 where the msMAPE and sMAPE terms differ"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        0.1
       ],
       "values": [
        0.0,
        0.0,
        0,
        0,
        0
       ]
      },
      {
       "id": "r2",
       "label": [
        0.5
       ],
       "values": [
        0.0,
        0.0,
        0,
        0,
        0
       ]
      },
      {
       "id": "r3",
       "label": [
        1.0
       ],
       "values": [
        0.0,
        0.0,
        0,
        0,
        0
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T9",
     "description": "msMAPE term of a forecast of 1e-9 for a zero actual by nu; hypothesis: H42 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H42"
     },
     "row_labels": [
      "nu"
     ],
     "columns": [
      "msMAPE term of a forecast of 1e-9 for a zero actual"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        0.1
       ],
       "values": [
        1e-08
       ]
      },
      {
       "id": "r2",
       "label": [
        0.5
       ],
       "values": [
        2e-09
       ]
      },
      {
       "id": "r3",
       "label": [
        1.0
       ],
       "values": [
        1e-09
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T10",
     "description": "signal-to-noise ratio of sMAPE over its large-spread formula, signal-to-noise ratio of the absolute error over its limit by error_distribution, c1, c2; hypothesis: H39, sigma: 200 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H39",
      "sigma": 200
     },
     "row_labels": [
      "error_distribution",
      "c1",
      "c2"
     ],
     "columns": [
      "signal-to-noise ratio of sMAPE over its large-spread formula",
      "signal-to-noise ratio of the absolute error over its limit"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "normal",
        0.0,
        0.5
       ],
       "values": [
        0.9999,
        0.9913
       ]
      },
      {
       "id": "r2",
       "label": [
        "normal",
        0.0,
        1.0
       ],
       "values": [
        1.0,
        0.9956
       ]
      },
      {
       "id": "r3",
       "label": [
        "normal",
        -0.5,
        1.5
       ],
       "values": [
        1.001,
        0.9966
       ]
      },
      {
       "id": "r4",
       "label": [
        "Gumbel",
        0.0,
        0.5
       ],
       "values": [
        0.9998,
        0.9647
       ]
      },
      {
       "id": "r5",
       "label": [
        "Gumbel",
        0.0,
        1.0
       ],
       "values": [
        1.0,
        0.9937
       ]
      },
      {
       "id": "r6",
       "label": [
        "Gumbel",
        -0.5,
        1.5
       ],
       "values": [
        1.0,
        0.9967
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T11",
     "description": "relative spread of the absolute log error's signal-to-noise ratio over sigma = 2, 10, 50 by error_distribution, c1, c2; hypothesis: H39 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H39"
     },
     "row_labels": [
      "error_distribution",
      "c1",
      "c2"
     ],
     "columns": [
      "relative spread of the absolute log error's signal-to-noise ratio over sigma = 2, 10, 50"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "normal",
        0.0,
        0.5
       ],
       "values": [
        1.686e-15
       ]
      },
      {
       "id": "r2",
       "label": [
        "normal",
        0.0,
        1.0
       ],
       "values": [
        8.388e-16
       ]
      },
      {
       "id": "r3",
       "label": [
        "normal",
        -0.5,
        1.5
       ],
       "values": [
        4.51e-16
       ]
      },
      {
       "id": "r4",
       "label": [
        "Gumbel",
        0.0,
        0.5
       ],
       "values": [
        4.186e-15
       ]
      },
      {
       "id": "r5",
       "label": [
        "Gumbel",
        0.0,
        1.0
       ],
       "values": [
        4.885e-15
       ]
      },
      {
       "id": "r6",
       "label": [
        "Gumbel",
        -0.5,
        1.5
       ],
       "values": [
        1.356e-14
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T12",
     "description": "signal-to-noise ratio of the squared error over its formula, lognormal, exp(sigma^2 / 2) times the squared error's ratio, lognormal, formula by c1, c2, sigma; hypothesis: H39 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H39"
     },
     "row_labels": [
      "c1",
      "c2",
      "sigma"
     ],
     "columns": [
      "signal-to-noise ratio of the squared error over its formula, lognormal",
      "exp(sigma^2 / 2) times the squared error's ratio, lognormal, formula"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        0.0,
        0.5,
        0.5
       ],
       "values": [
        1.0,
        null
       ]
      },
      {
       "id": "r2",
       "label": [
        0.0,
        1.0,
        0.5
       ],
       "values": [
        1.0,
        null
       ]
      },
      {
       "id": "r3",
       "label": [
        -0.5,
        1.5,
        0.5
       ],
       "values": [
        1.0,
        null
       ]
      },
      {
       "id": "r4",
       "label": [
        0.0,
        0.5,
        1.0
       ],
       "values": [
        1.0,
        null
       ]
      },
      {
       "id": "r5",
       "label": [
        0.0,
        1.0,
        1.0
       ],
       "values": [
        1.0,
        null
       ]
      },
      {
       "id": "r6",
       "label": [
        -0.5,
        1.5,
        1.0
       ],
       "values": [
        1.0,
        null
       ]
      },
      {
       "id": "r7",
       "label": [
        0.0,
        0.5,
        2.0
       ],
       "values": [
        1.0,
        null
       ]
      },
      {
       "id": "r8",
       "label": [
        0.0,
        1.0,
        2.0
       ],
       "values": [
        1.0,
        null
       ]
      },
      {
       "id": "r9",
       "label": [
        -0.5,
        1.5,
        2.0
       ],
       "values": [
        1.0,
        null
       ]
      },
      {
       "id": "r10",
       "label": [
        0.0,
        1.0,
        4.0
       ],
       "values": [
        null,
        0.9907
       ]
      },
      {
       "id": "r11",
       "label": [
        0.0,
        1.0,
        6.0
       ],
       "values": [
        null,
        1.0
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "description": "distributions compared with the asymptotic values recorded by RN7; hypothesis: H40 (lab runs RN13)",
     "row_labels": [],
     "columns": [
      "distributions compared with the asymptotic values recorded by RN7"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        16
       ]
      }
     ],
     "conditions": {
      "hypothesis": "H40"
     },
     "artefact": "A6",
     "id": "T13"
    },
    {
     "description": "distributions where A is not below 2 g at the median, largest difference from the asymptotic values recorded by RN7; hypothesis: H40, distributions: 16 (lab runs RN13)",
     "row_labels": [],
     "columns": [
      "distributions where A is not below 2 g at the median",
      "largest difference from the asymptotic values recorded by RN7"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        0,
        0.00014
       ]
      }
     ],
     "conditions": {
      "hypothesis": "H40",
      "distributions": 16
     },
     "artefact": "A6",
     "id": "T14"
    },
    {
     "id": "T15",
     "description": "large-sample relative efficiency, formula by family, spread; hypothesis: H40 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H40"
     },
     "row_labels": [
      "family",
      "spread"
     ],
     "columns": [
      "large-sample relative efficiency, formula"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "lognormal",
        0.25
       ],
       "values": [
        0.9696
       ]
      },
      {
       "id": "r2",
       "label": [
        "lognormal",
        0.5
       ],
       "values": [
        0.8881
       ]
      },
      {
       "id": "r3",
       "label": [
        "lognormal",
        1.0
       ],
       "values": [
        0.6601
       ]
      },
      {
       "id": "r4",
       "label": [
        "lognormal",
        1.5
       ],
       "values": [
        0.454
       ]
      },
      {
       "id": "r5",
       "label": [
        "lognormal",
        2.0
       ],
       "values": [
        0.3073
       ]
      },
      {
       "id": "r6",
       "label": [
        "lognormal",
        3.0
       ],
       "values": [
        0.1474
       ]
      },
      {
       "id": "r7",
       "label": [
        "lognormal",
        4.0
       ],
       "values": [
        0.07778
       ]
      },
      {
       "id": "r8",
       "label": [
        "lognormal",
        6.0
       ],
       "values": [
        0.0279
       ]
      },
      {
       "id": "r9",
       "label": [
        "log-Laplace",
        0.25
       ],
       "values": [
        0.9969
       ]
      },
      {
       "id": "r10",
       "label": [
        "log-Laplace",
        0.5
       ],
       "values": [
        0.9765
       ]
      },
      {
       "id": "r11",
       "label": [
        "log-Laplace",
        1.0
       ],
       "values": [
        0.8953
       ]
      },
      {
       "id": "r12",
       "label": [
        "log-Laplace",
        1.5
       ],
       "values": [
        0.8036
       ]
      },
      {
       "id": "r13",
       "label": [
        "log-Laplace",
        2.0
       ],
       "values": [
        0.7207
       ]
      },
      {
       "id": "r14",
       "label": [
        "log-Laplace",
        3.0
       ],
       "values": [
        0.5903
       ]
      },
      {
       "id": "r15",
       "label": [
        "log-Laplace",
        4.0
       ],
       "values": [
        0.4969
       ]
      },
      {
       "id": "r16",
       "label": [
        "log-Laplace",
        6.0
       ],
       "values": [
        0.3755
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T16",
     "description": "distance of the large-sample relative efficiency from 1 at spread 0.01 by family; hypothesis: H40 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H40"
     },
     "row_labels": [
      "family"
     ],
     "columns": [
      "distance of the large-sample relative efficiency from 1 at spread 0.01"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "lognormal"
       ],
       "values": [
        5e-05
       ]
      },
      {
       "id": "r2",
       "label": [
        "log-Laplace"
       ],
       "values": [
        1.249e-08
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T17",
     "description": "large-sample relative efficiency times sigma^3 over 6 (ln 2)^2 sqrt(2 pi), lognormal, large-sample relative efficiency times b over 3, log-Laplace; hypothesis: H40, spread: 200 (lab runs RN13)",
     "conditions": {
      "hypothesis": "H40",
      "spread": 200
     },
     "row_labels": [],
     "columns": [
      "large-sample relative efficiency times sigma^3 over 6 (ln 2)^2 sqrt(2 pi), lognormal",
      "large-sample relative efficiency times b over 3, log-Laplace"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        0.9998,
        0.9906
       ]
      }
     ],
     "artefact": "A6"
    }
   ],
   "artefacts": [
    "A6"
   ],
   "setup": {
    "studied": [
     "D1",
     "D28",
     "D33",
     "D6",
     "D8"
    ],
    "lens": [
     "D32",
     "D9",
     "D5",
     "D4",
     "D19",
     "D11",
     "D10"
    ],
    "procedure": [
     "D31"
    ]
   },
   "execution": {
    "code_version": "2f0d2fca48b049d8004fec7228e47b8857f6ace4",
    "software": {
     "python": "3.13.18",
     "platform": "Linux-6.18.44-fc-v595-x86_64-with-glibc2.39"
    }
   }
  },
  {
   "id": "E32",
   "profile": "quant",
   "design": "other",
   "description": "Numerical checks of the formulas of C18 ([[D31:formula checks]]), lab run RN18 (check_h39.py) under the formal plan that decided C18; every check is within the plan's tolerances. They check the formulas on concrete numbers; the proof is the evidence for the claim. The run's script labels the checks H39, the record's first wording of C18; the formulas are those of C18.",
   "availability": {
    "level": "public"
   },
   "lab_runs": [
    "RN18"
   ],
   "tables": [
    {
     "id": "T1",
     "description": "distance of sMAPE's signal-to-noise ratio over its large-spread formula from 1, distance of the absolute error's signal-to-noise ratio over its limit from 1 by error_distribution, c1, c2, sigma; hypothesis: H39 (lab runs RN18)",
     "conditions": {
      "hypothesis": "H39"
     },
     "row_labels": [
      "error_distribution",
      "c1",
      "c2",
      "sigma"
     ],
     "columns": [
      "distance of sMAPE's signal-to-noise ratio over its large-spread formula from 1",
      "distance of the absolute error's signal-to-noise ratio over its limit from 1"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "normal",
        0.0,
        0.5,
        200
       ],
       "values": [
        5.266e-05,
        0.008727
       ]
      },
      {
       "id": "r2",
       "label": [
        "normal",
        0.0,
        0.5,
        2000
       ],
       "values": [
        2.749e-06,
        0.0008713
       ]
      },
      {
       "id": "r3",
       "label": [
        "normal",
        0.0,
        1.0,
        200
       ],
       "values": [
        0.0003989,
        0.004379
       ]
      },
      {
       "id": "r4",
       "label": [
        "normal",
        0.0,
        1.0,
        2000
       ],
       "values": [
        4.703e-05,
        0.0004373
       ]
      },
      {
       "id": "r5",
       "label": [
        "normal",
        -0.5,
        1.5,
        200
       ],
       "values": [
        0.00122,
        0.003402
       ]
      },
      {
       "id": "r6",
       "label": [
        "normal",
        -0.5,
        1.5,
        2000
       ],
       "values": [
        0.0001273,
        0.0003398
       ]
      },
      {
       "id": "r7",
       "label": [
        "Gumbel",
        0.0,
        0.5,
        200
       ],
       "values": [
        0.0001627,
        0.03529
       ]
      },
      {
       "id": "r8",
       "label": [
        "Gumbel",
        0.0,
        0.5,
        2000
       ],
       "values": [
        5.872e-07,
        0.00352
       ]
      },
      {
       "id": "r9",
       "label": [
        "Gumbel",
        0.0,
        1.0,
        200
       ],
       "values": [
        0.0001512,
        0.006295
       ]
      },
      {
       "id": "r10",
       "label": [
        "Gumbel",
        0.0,
        1.0,
        2000
       ],
       "values": [
        2.457e-05,
        0.0006281
       ]
      },
      {
       "id": "r11",
       "label": [
        "Gumbel",
        -0.5,
        1.5,
        200
       ],
       "values": [
        0.0003779,
        0.00326
       ]
      },
      {
       "id": "r12",
       "label": [
        "Gumbel",
        -0.5,
        1.5,
        2000
       ],
       "values": [
        4.713e-05,
        0.0003255
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T2",
     "description": "relative spread of the absolute log error's signal-to-noise ratio over sigma = 2, 10, 50 by error_distribution, c1, c2; hypothesis: H39 (lab runs RN18)",
     "conditions": {
      "hypothesis": "H39"
     },
     "row_labels": [
      "error_distribution",
      "c1",
      "c2"
     ],
     "columns": [
      "relative spread of the absolute log error's signal-to-noise ratio over sigma = 2, 10, 50"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "normal",
        0.0,
        0.5
       ],
       "values": [
        3.761e-15
       ]
      },
      {
       "id": "r2",
       "label": [
        "normal",
        0.0,
        1.0
       ],
       "values": [
        2.397e-16
       ]
      },
      {
       "id": "r3",
       "label": [
        "normal",
        -0.5,
        1.5
       ],
       "values": [
        1.579e-15
       ]
      },
      {
       "id": "r4",
       "label": [
        "Gumbel",
        0.0,
        0.5
       ],
       "values": [
        4.243e-13
       ]
      },
      {
       "id": "r5",
       "label": [
        "Gumbel",
        0.0,
        1.0
       ],
       "values": [
        7.932e-13
       ]
      },
      {
       "id": "r6",
       "label": [
        "Gumbel",
        -0.5,
        1.5
       ],
       "values": [
        1.43e-12
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T3",
     "description": "distance of the squared error's signal-to-noise ratio over its formula from 1, lognormal, distance of exp(sigma^2 / 2) times the squared error's ratio from 1, lognormal, formula by c1, c2, sigma; hypothesis: H39 (lab runs RN18)",
     "conditions": {
      "hypothesis": "H39"
     },
     "row_labels": [
      "c1",
      "c2",
      "sigma"
     ],
     "columns": [
      "distance of the squared error's signal-to-noise ratio over its formula from 1, lognormal",
      "distance of exp(sigma^2 / 2) times the squared error's ratio from 1, lognormal, formula"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        0.0,
        0.5,
        0.5
       ],
       "values": [
        3.109e-15,
        null
       ]
      },
      {
       "id": "r2",
       "label": [
        0.0,
        1.0,
        0.5
       ],
       "values": [
        6.661e-16,
        null
       ]
      },
      {
       "id": "r3",
       "label": [
        -0.5,
        1.5,
        0.5
       ],
       "values": [
        1.11e-15,
        null
       ]
      },
      {
       "id": "r4",
       "label": [
        0.0,
        0.5,
        1.0
       ],
       "values": [
        1.554e-15,
        null
       ]
      },
      {
       "id": "r5",
       "label": [
        0.0,
        1.0,
        1.0
       ],
       "values": [
        8.882e-16,
        null
       ]
      },
      {
       "id": "r6",
       "label": [
        -0.5,
        1.5,
        1.0
       ],
       "values": [
        5.551e-16,
        null
       ]
      },
      {
       "id": "r7",
       "label": [
        0.0,
        0.5,
        2.0
       ],
       "values": [
        1.409e-13,
        null
       ]
      },
      {
       "id": "r8",
       "label": [
        0.0,
        1.0,
        2.0
       ],
       "values": [
        7.55e-15,
        null
       ]
      },
      {
       "id": "r9",
       "label": [
        -0.5,
        1.5,
        2.0
       ],
       "values": [
        1.887e-15,
        null
       ]
      },
      {
       "id": "r10",
       "label": [
        0.0,
        1.0,
        4.0
       ],
       "values": [
        null,
        0.009325
       ]
      },
      {
       "id": "r11",
       "label": [
        0.0,
        1.0,
        6.0
       ],
       "values": [
        null,
        3.08e-06
       ]
      },
      {
       "id": "r12",
       "label": [
        0.0,
        1.0,
        8.0
       ],
       "values": [
        null,
        1.888e-11
       ]
      }
     ],
     "artefact": "A6"
    }
   ],
   "artefacts": [
    "A6"
   ],
   "setup": {
    "studied": [
     "D1"
    ],
    "lens": [
     "D32",
     "D19",
     "D11",
     "D10"
    ],
    "procedure": [
     "D31"
    ]
   },
   "execution": {
    "code_version": "6aa7a7ab87bf5c4e40d704a7492cd9af9ef481db",
    "software": {
     "python": "3.13.18",
     "platform": "Linux-6.18.44-fc-v595-x86_64-with-glibc2.39"
    }
   }
  },
  {
   "id": "E33",
   "profile": "quant",
   "design": "other",
   "description": "Numerical checks of the formulas of C19 ([[D31:formula checks]]), lab run RN16 (check_corrected.py, seed 8) under the formal plan that decided C19; every check is within the plan's tolerances. They check the formulas on concrete numbers; the proof is the evidence for the claim. The run's script labels the checks by the record's first wordings: H40 (T4 to T7 and the result R1) checks C19; H39 (T1 to T3) checks the formulas of C18, which the same plan tested in an earlier wording.",
   "availability": {
    "level": "public"
   },
   "lab_runs": [
    "RN16"
   ],
   "results": [
    {
     "id": "R1",
     "variable": "distributions compared with the asymptotic values recorded by RN7",
     "value": 16,
     "conditions": {
      "hypothesis": "H40"
     }
    }
   ],
   "tables": [
    {
     "id": "T1",
     "description": "distance of sMAPE's signal-to-noise ratio over its large-spread formula from 1, distance of the absolute error's signal-to-noise ratio over its limit from 1 by error_distribution, c1, c2, sigma; hypothesis: H39 (lab runs RN16)",
     "conditions": {
      "hypothesis": "H39"
     },
     "row_labels": [
      "error_distribution",
      "c1",
      "c2",
      "sigma"
     ],
     "columns": [
      "distance of sMAPE's signal-to-noise ratio over its large-spread formula from 1",
      "distance of the absolute error's signal-to-noise ratio over its limit from 1"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "normal",
        0.0,
        0.5,
        200
       ],
       "values": [
        5.266e-05,
        0.008727
       ]
      },
      {
       "id": "r2",
       "label": [
        "normal",
        0.0,
        0.5,
        2000
       ],
       "values": [
        2.749e-06,
        0.0008713
       ]
      },
      {
       "id": "r3",
       "label": [
        "normal",
        0.0,
        1.0,
        200
       ],
       "values": [
        0.0003989,
        0.004379
       ]
      },
      {
       "id": "r4",
       "label": [
        "normal",
        0.0,
        1.0,
        2000
       ],
       "values": [
        4.703e-05,
        0.0004373
       ]
      },
      {
       "id": "r5",
       "label": [
        "normal",
        -0.5,
        1.5,
        200
       ],
       "values": [
        0.00122,
        0.003402
       ]
      },
      {
       "id": "r6",
       "label": [
        "normal",
        -0.5,
        1.5,
        2000
       ],
       "values": [
        0.0001273,
        0.0003398
       ]
      },
      {
       "id": "r7",
       "label": [
        "Gumbel",
        0.0,
        0.5,
        200
       ],
       "values": [
        0.0001627,
        0.03529
       ]
      },
      {
       "id": "r8",
       "label": [
        "Gumbel",
        0.0,
        0.5,
        2000
       ],
       "values": [
        5.872e-07,
        0.00352
       ]
      },
      {
       "id": "r9",
       "label": [
        "Gumbel",
        0.0,
        1.0,
        200
       ],
       "values": [
        0.0001512,
        0.006295
       ]
      },
      {
       "id": "r10",
       "label": [
        "Gumbel",
        0.0,
        1.0,
        2000
       ],
       "values": [
        2.457e-05,
        0.0006281
       ]
      },
      {
       "id": "r11",
       "label": [
        "Gumbel",
        -0.5,
        1.5,
        200
       ],
       "values": [
        0.0003779,
        0.00326
       ]
      },
      {
       "id": "r12",
       "label": [
        "Gumbel",
        -0.5,
        1.5,
        2000
       ],
       "values": [
        4.713e-05,
        0.0003255
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T2",
     "description": "relative spread of the absolute log error's signal-to-noise ratio over sigma = 2, 10, 50 by error_distribution, c1, c2; hypothesis: H39 (lab runs RN16)",
     "conditions": {
      "hypothesis": "H39"
     },
     "row_labels": [
      "error_distribution",
      "c1",
      "c2"
     ],
     "columns": [
      "relative spread of the absolute log error's signal-to-noise ratio over sigma = 2, 10, 50"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "normal",
        0.0,
        0.5
       ],
       "values": [
        3.761e-15
       ]
      },
      {
       "id": "r2",
       "label": [
        "normal",
        0.0,
        1.0
       ],
       "values": [
        2.397e-16
       ]
      },
      {
       "id": "r3",
       "label": [
        "normal",
        -0.5,
        1.5
       ],
       "values": [
        1.579e-15
       ]
      },
      {
       "id": "r4",
       "label": [
        "Gumbel",
        0.0,
        0.5
       ],
       "values": [
        4.243e-13
       ]
      },
      {
       "id": "r5",
       "label": [
        "Gumbel",
        0.0,
        1.0
       ],
       "values": [
        7.932e-13
       ]
      },
      {
       "id": "r6",
       "label": [
        "Gumbel",
        -0.5,
        1.5
       ],
       "values": [
        1.43e-12
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T3",
     "description": "distance of the squared error's signal-to-noise ratio over its formula from 1, lognormal, distance of exp(sigma^2 / 2) times the squared error's ratio from 1, lognormal, formula by c1, c2, sigma; hypothesis: H39 (lab runs RN16)",
     "conditions": {
      "hypothesis": "H39"
     },
     "row_labels": [
      "c1",
      "c2",
      "sigma"
     ],
     "columns": [
      "distance of the squared error's signal-to-noise ratio over its formula from 1, lognormal",
      "distance of exp(sigma^2 / 2) times the squared error's ratio from 1, lognormal, formula"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        0.0,
        0.5,
        0.5
       ],
       "values": [
        3.109e-15,
        null
       ]
      },
      {
       "id": "r2",
       "label": [
        0.0,
        1.0,
        0.5
       ],
       "values": [
        6.661e-16,
        null
       ]
      },
      {
       "id": "r3",
       "label": [
        -0.5,
        1.5,
        0.5
       ],
       "values": [
        1.11e-15,
        null
       ]
      },
      {
       "id": "r4",
       "label": [
        0.0,
        0.5,
        1.0
       ],
       "values": [
        1.554e-15,
        null
       ]
      },
      {
       "id": "r5",
       "label": [
        0.0,
        1.0,
        1.0
       ],
       "values": [
        8.882e-16,
        null
       ]
      },
      {
       "id": "r6",
       "label": [
        -0.5,
        1.5,
        1.0
       ],
       "values": [
        5.551e-16,
        null
       ]
      },
      {
       "id": "r7",
       "label": [
        0.0,
        0.5,
        2.0
       ],
       "values": [
        1.409e-13,
        null
       ]
      },
      {
       "id": "r8",
       "label": [
        0.0,
        1.0,
        2.0
       ],
       "values": [
        7.55e-15,
        null
       ]
      },
      {
       "id": "r9",
       "label": [
        -0.5,
        1.5,
        2.0
       ],
       "values": [
        1.887e-15,
        null
       ]
      },
      {
       "id": "r10",
       "label": [
        0.0,
        1.0,
        4.0
       ],
       "values": [
        null,
        0.009325
       ]
      },
      {
       "id": "r11",
       "label": [
        0.0,
        1.0,
        6.0
       ],
       "values": [
        null,
        3.08e-06
       ]
      },
      {
       "id": "r12",
       "label": [
        0.0,
        1.0,
        8.0
       ],
       "values": [
        null,
        1.888e-11
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T4",
     "description": "large-sample relative efficiency, formula by family, spread; hypothesis: H40 (lab runs RN16)",
     "conditions": {
      "hypothesis": "H40"
     },
     "row_labels": [
      "family",
      "spread"
     ],
     "columns": [
      "large-sample relative efficiency, formula"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "lognormal",
        0.25
       ],
       "values": [
        0.9696
       ]
      },
      {
       "id": "r2",
       "label": [
        "lognormal",
        0.5
       ],
       "values": [
        0.8881
       ]
      },
      {
       "id": "r3",
       "label": [
        "lognormal",
        1.0
       ],
       "values": [
        0.6601
       ]
      },
      {
       "id": "r4",
       "label": [
        "lognormal",
        1.5
       ],
       "values": [
        0.454
       ]
      },
      {
       "id": "r5",
       "label": [
        "lognormal",
        2.0
       ],
       "values": [
        0.3073
       ]
      },
      {
       "id": "r6",
       "label": [
        "lognormal",
        3.0
       ],
       "values": [
        0.1474
       ]
      },
      {
       "id": "r7",
       "label": [
        "lognormal",
        4.0
       ],
       "values": [
        0.07778
       ]
      },
      {
       "id": "r8",
       "label": [
        "lognormal",
        6.0
       ],
       "values": [
        0.0279
       ]
      },
      {
       "id": "r9",
       "label": [
        "log-Laplace",
        0.25
       ],
       "values": [
        0.9969
       ]
      },
      {
       "id": "r10",
       "label": [
        "log-Laplace",
        0.5
       ],
       "values": [
        0.9765
       ]
      },
      {
       "id": "r11",
       "label": [
        "log-Laplace",
        1.0
       ],
       "values": [
        0.8953
       ]
      },
      {
       "id": "r12",
       "label": [
        "log-Laplace",
        1.5
       ],
       "values": [
        0.8036
       ]
      },
      {
       "id": "r13",
       "label": [
        "log-Laplace",
        2.0
       ],
       "values": [
        0.7207
       ]
      },
      {
       "id": "r14",
       "label": [
        "log-Laplace",
        3.0
       ],
       "values": [
        0.5903
       ]
      },
      {
       "id": "r15",
       "label": [
        "log-Laplace",
        4.0
       ],
       "values": [
        0.4969
       ]
      },
      {
       "id": "r16",
       "label": [
        "log-Laplace",
        6.0
       ],
       "values": [
        0.3755
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T5",
     "description": "distance of the large-sample relative efficiency from 1 at spread 0.01 by family; hypothesis: H40 (lab runs RN16)",
     "conditions": {
      "hypothesis": "H40"
     },
     "row_labels": [
      "family"
     ],
     "columns": [
      "distance of the large-sample relative efficiency from 1 at spread 0.01"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [
        "lognormal"
       ],
       "values": [
        5e-05
       ]
      },
      {
       "id": "r2",
       "label": [
        "log-Laplace"
       ],
       "values": [
        1.249e-08
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T6",
     "description": "distributions where A is not below 2 g at the median, largest difference from the asymptotic values recorded by RN7; hypothesis: H40, distributions: 16 (lab runs RN16)",
     "conditions": {
      "hypothesis": "H40",
      "distributions": 16
     },
     "row_labels": [],
     "columns": [
      "distributions where A is not below 2 g at the median",
      "largest difference from the asymptotic values recorded by RN7"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        0,
        0.00014
       ]
      }
     ],
     "artefact": "A6"
    },
    {
     "id": "T7",
     "description": "large-sample relative efficiency times sigma^3 over 6 (ln 2)^2 sqrt(2 pi), lognormal, large-sample relative efficiency times b over 3, log-Laplace; hypothesis: H40, spread: 200 (lab runs RN16)",
     "conditions": {
      "hypothesis": "H40",
      "spread": 200
     },
     "row_labels": [],
     "columns": [
      "large-sample relative efficiency times sigma^3 over 6 (ln 2)^2 sqrt(2 pi), lognormal",
      "large-sample relative efficiency times b over 3, log-Laplace"
     ],
     "rows": [
      {
       "id": "r1",
       "label": [],
       "values": [
        0.9998,
        0.9906
       ]
      }
     ],
     "artefact": "A6"
    }
   ],
   "artefacts": [
    "A6"
   ],
   "setup": {
    "studied": [
     "D1",
     "D8"
    ],
    "lens": [
     "D32",
     "D9",
     "D19",
     "D11",
     "D10"
    ],
    "procedure": [
     "D31"
    ]
   },
   "execution": {
    "code_version": "6aa7a7ab87bf5c4e40d704a7492cd9af9ef481db",
    "software": {
     "python": "3.13.18",
     "platform": "Linux-6.18.44-fc-v595-x86_64-with-glibc2.39"
    }
   }
  },
  {
   "id": "E18",
   "profile": "formal",
   "description": "Proof of C1 (section H1 of the proof document A1, rendered in A2). Only the ratio of the actual to the forecast matters, not their size.",
   "foundation": "Kolmogorov probability axioms and real analysis",
   "steps": [
    {
     "id": "S1",
     "statement": "Since $y$ and $f$ are positive, $|y| + |f| = y + f$. So $s = |y - f|/(y + f)$.",
     "rule": "algebra",
     "by": [
      "D1"
     ]
    },
    {
     "id": "S2",
     "statement": "Dividing the numerator and the denominator by $f > 0$ gives $\\dfrac{|y-f|/f}{(y+f)/f}$. Since $f > 0$, $|y - f|/f = |(y - f)/f| = |r - 1|$. And $(y+f)/f = r + 1$.",
     "rule": "algebra",
     "by": [
      "S1"
     ]
    },
    {
     "id": "S3",
     "statement": "So $s = |r - 1|/(r + 1)$. The contribution depends only on the ratio of actual to forecast.",
     "rule": "conclusion",
     "by": [
      "S2"
     ]
    }
   ],
   "artefacts": [
    "A1",
    "A2"
   ],
   "availability": {
    "level": "public"
   }
  },
  {
   "id": "E20",
   "profile": "formal",
   "description": "Proof of C3 (section H3 of the proof document A1, rendered in A2). sMAPE measures the error on the log scale. For small errors it is half the absolute log error; for large errors it levels off.",
   "foundation": "Kolmogorov probability axioms and real analysis",
   "steps": [
    {
     "id": "S1",
     "statement": "Write $z = \\ln y - \\ln f$, so that $r = y/f = e^z$. By C1, $s = \\dfrac{|e^z - 1|}{e^z + 1}$.",
     "rule": "definition",
     "by": [
      "C1"
     ]
    },
    {
     "id": "S2",
     "statement": "Multiplying the numerator and the denominator of $\\dfrac{e^z - 1}{e^z + 1}$ by $e^{-z/2}$ gives $\\dfrac{e^{z/2} - e^{-z/2}}{e^{z/2} + e^{-z/2}} = \\tanh(z/2)$. The denominator is positive, so $s = |\\tanh(z/2)|$.",
     "rule": "algebra",
     "by": [
      "S1"
     ]
    },
    {
     "id": "S3",
     "statement": "$\\tanh$ changes only its sign when its argument changes sign: $\\tanh(-x) = -\\tanh(x)$, as the definition shows. So $|\\tanh(z/2)| = \\tanh(|z|/2)$, which is the stated form.",
     "rule": "algebra",
     "by": [
      "S2"
     ]
    },
    {
     "id": "S4",
     "statement": "For (a): $\\tanh$ is 0 at 0. On $[0, \\infty)$ it is strictly increasing and concave, and it approaches 1 as its argument grows; its derivative $1 - \\tanh^2 x$ is positive and decreasing for $x > 0$. With the argument $|\\ln y - \\ln f|/2$, $s$ has these properties in $|\\ln y - \\ln f|$.",
     "rule": "monotonicity",
     "by": [
      "S3"
     ]
    },
    {
     "id": "S5",
     "statement": "For (b): for every $x \\ge 0$, $x - x^3/3 \\le \\tanh x \\le x$. The upper bound holds because $\\tanh$ has slope at most 1; the lower one because the derivative of $\\tanh x - x + x^3/3$ is $x^2 - \\tanh^2 x \\ge 0$, and the difference is 0 at $x = 0$. Putting $x = |\\ln y - \\ln f|/2$ gives the bounds.",
     "rule": "algebra",
     "by": [
      "S4"
     ]
    },
    {
     "id": "S6",
     "statement": "For (c), the median: the mean of $|\\ln y_t - \\ln f|$ over a test set is minimised when $\\ln f$ is a median of the $\\ln y_t$. Since the logarithm is increasing, this is when $f$ is a median $m$ of the $y_t$.",
     "rule": "monotonicity",
     "by": [
      "S5"
     ]
    },
    {
     "id": "S7",
     "statement": "For (c), the bound: with $z_t = \\ln y_t - \\ln f$ the log errors of the test set, write $\\overline{|z|}(f) = \\frac1n\\sum_t|\\ln y_t - \\ln f|$, $\\overline{|z|^3}(f) = \\frac1n\\sum_t|\\ln y_t - \\ln f|^3$ and $\\bar s(f)$ for the mean of the sMAPE terms of the test set. Averaging the bounds of step 5 over the test set gives $\\bar s(f) \\le \\tfrac12\\overline{|z|}(f) \\le \\bar s(f) + \\tfrac1{24}\\overline{|z|^3}(f)$ for every $f > 0$. Let $f^*$ minimise $\\bar s$, so that $\\bar s(f^*) \\le \\bar s(m)$. Then\n$$\\tfrac12\\overline{|z|}(f^*) \\le \\bar s(f^*) + \\tfrac1{24}\\overline{|z|^3}(f^*) \\le \\bar s(m) + \\tfrac1{24}\\overline{|z|^3}(f^*) \\le \\tfrac12\\overline{|z|}(m) + \\tfrac1{24}\\overline{|z|^3}(f^*).$$\nMultiplying by 2 gives $\\overline{|z|}(f^*) - \\overline{|z|}(m) \\le \\tfrac1{12}\\overline{|z|^3}(f^*)$, the stated inequality.",
     "rule": "algebra",
     "by": [
      "S5"
     ]
    },
    {
     "id": "S8",
     "statement": "For (c), not the median in general: take the five actuals $1, 1, e, e^4, e^4$, whose median is $e$. With the forecast at $e$ the log errors are $1, 1, 0, 3, 3$ in size and the sMAPE terms sum to $2\\tanh(0.5) + 2\\tanh(1.5) = 2.734$. With the forecast at 1 the log errors are $0, 0, 1, 4, 4$ and the terms sum to $\\tanh(0.5) + 2\\tanh(2) = 2.390$, which is lower. So the median does not minimise the sMAPE of this test set, although it minimises its log-MAE (step 6).",
     "rule": "algebra",
     "by": [
      "S6"
     ]
    },
    {
     "id": "S9",
     "statement": "So $s = \\tanh(|\\ln y - \\ln f|/2)$. It rises from 0 like half the absolute log error and levels off at 1. For small errors sMAPE is close to the log-MAE, and its minimiser is close to the median without in general being the median.",
     "rule": "conclusion",
     "by": [
      "S8"
     ]
    }
   ],
   "artefacts": [
    "A1",
    "A2"
   ],
   "availability": {
    "level": "public"
   }
  },
  {
   "id": "E28",
   "profile": "formal",
   "description": "Proof of C10 (section H11 of the proof document A1, rendered in A2). For widely spread actuals, every forecast scores close to the worst value. The differences between forecasts shrink, and the target moves from the median to the mode of $\\ln y$.",
   "foundation": "Kolmogorov probability axioms and real analysis",
   "steps": [
    {
     "id": "S1",
     "statement": "Rewrite the expectation. By C3, and since $\\tanh(|x|/2) = 1 - \\dfrac{2}{1 + e^{|x|}}$, $\\mathrm E[s(y,f)] = 1 - \\mathrm E\\big[2/(1 + e^{|\\ln y - \\ln f|})\\big]$. The density of $\\ln y$ at $x$ is $h\\big((x - \\mu)/\\sigma\\big)/\\sigma$. Substituting $x = \\ln f + v$ in the integral gives\n$$\\mathrm E[s(y,f)] = 1 - \\frac1\\sigma\\int \\frac{2}{1 + e^{|v|}}\\,h\\Big(c + \\frac v\\sigma\\Big)\\,dv.$$",
     "rule": "change of variables",
     "by": [
      "C3"
     ]
    },
    {
     "id": "S2",
     "statement": "Expand $h$. By Taylor's theorem, $h(c + v/\\sigma) = h(c) + h'(c)\\,v/\\sigma + \\tfrac12 h''(\\xi)\\,v^2/\\sigma^2$ for some $\\xi$ between $c$ and $c + v/\\sigma$.",
     "rule": "Taylor's theorem",
     "by": [
      "S1"
     ]
    },
    {
     "id": "S3",
     "statement": "The first term integrates to $4\\ln 2\\,h(c)$, since $\\int \\frac{2}{1+e^{|v|}}\\,dv = 4\\ln 2$. Over $v > 0$, $\\int 2/(1 + e^v)\\,dv = 2\\ln 2$.",
     "rule": "algebra",
     "by": [
      "S2"
     ]
    },
    {
     "id": "S4",
     "statement": "The second term integrates to 0: its factor $v$ changes sign with $v$, while the rest does not.",
     "rule": "algebra",
     "by": [
      "S3"
     ]
    },
    {
     "id": "S5",
     "statement": "The third term is at most $\\tfrac12\\max|h''|$ times $\\int \\frac{2v^2}{1+e^{|v|}}\\,dv = 6\\zeta(3) \\approx 7.21$, divided by $\\sigma^2$. This integral is a standard Fermi–Dirac integral. With the factor $1/\\sigma$ of step 1, this gives the expansion.",
     "rule": "standard integral",
     "by": [
      "S1"
     ]
    },
    {
     "id": "S6",
     "statement": "(a). The expansion tends to 1, so the expected sMAPE tends to 200. Two forecasts at fixed quantile levels have fixed positions $c_1$ and $c_2$. So their difference is $4\\ln 2\\,[h(c_2) - h(c_1)]/\\sigma + O(\\sigma^{-3})$.",
     "rule": "algebra",
     "by": [
      "S5"
     ]
    },
    {
     "id": "S7",
     "statement": "(b). Minimising the expansion over $c$ is maximising $h(c)$, up to a term that shrinks like $\\sigma^{-2}$ for every $c$ at once. A density with a bounded derivative tends to 0 far out. So its single highest point $c_0$ stands clear of every other point, and the minimisers' positions tend to $c_0$. So do their quantile levels.",
     "rule": "algebra",
     "by": [
      "S6"
     ]
    },
    {
     "id": "S8",
     "statement": "(c). By C3 (b), $\\tfrac12\\sigma|\\varepsilon - c| - s$ lies between 0 and $\\tfrac1{24}\\sigma^3|\\varepsilon - c|^3$. So for $c$ in any bounded range, $\\mathrm E[s]/\\sigma = \\tfrac12\\mathrm E|\\varepsilon - c| + O(\\sigma^2)$. This uses $\\mathrm E|\\varepsilon|^3 < \\infty$. And $\\mathrm E|\\varepsilon - c|$ is smallest at the median of $\\varepsilon$.",
     "rule": "earlier result",
     "by": [
      "C3"
     ]
    },
    {
     "id": "S9",
     "statement": "The minimisers also stay in a bounded range. For large $|c|$, most actuals have $\\varepsilon$ at least $|c|/2$ from $c$. And $\\tanh x \\ge x\\tanh 1$ for $0 \\le x \\le 1$. So $\\mathrm E[s]/\\sigma$ grows with $|c|$, or exceeds a multiple of $1/\\sigma$, beyond its value at the median. So the minimisers' positions tend to the median of $\\varepsilon$, whose quantile level is $1/2$.",
     "rule": "algebra",
     "by": [
      "S8"
     ]
    },
    {
     "id": "S10",
     "statement": "So as the actuals spread out, the expected sMAPE flattens towards its bound. Differences between forecasts vanish like $1/\\sigma$. The target moves from the median to the mode of $\\ln y$.",
     "rule": "conclusion",
     "by": [
      "S9"
     ]
    }
   ],
   "artefacts": [
    "A1",
    "A2"
   ],
   "availability": {
    "level": "public"
   }
  }
 ],
 "artefacts": [
  {
   "id": "A1",
   "kind": "proof",
   "uri": "https://entrylib.org/p/2026.000013/v3/files/files/proofs.md",
   "availability": {
    "level": "public"
   },
   "version": "sha256:90c89f3395e068a7a2762e2839e86894e4efe488dd09ca3fa988e7ce0cc1df51",
   "licence": "CC-BY-4.0",
   "description": "Proofs of the formal claims, version 3, in Markdown with LaTeX: section H3 with part (c); H5 shortened to the weighted median; new sections H12 (the ordinary median, with the checker's example), H13 (signal-to-noise ratios) and H14 (large-sample efficiency). Checked independently by AC6 (CK27 to CK31, CK34 to CK37) and corrected as it proposed; after its checks the notation was made consistent (no second meaning for k, a, G), which AC6 read and found to change nothing. Sections H1 to H14 prove C1, C2, C3, C9, C4, C5, C6, C7, C11, C8, C10, C12, C18 and C19, in that order. Sections H1 to H11 were also checked independently by lab actor AC4 (proof-check-report.md) and by the supervisor."
  },
  {
   "id": "A2",
   "kind": "proof",
   "uri": "https://entrylib.org/p/2026.000013/v3/files/files/proofs.pdf",
   "availability": {
    "level": "public"
   },
   "version": "sha256:767603774315d5d392a7fed0c7fb35531c52a4f68f836b0d944aa5c83d5cdc35",
   "licence": "CC-BY-4.0",
   "description": "The same proofs, version 3, rendered as PDF."
  },
  {
   "id": "A6",
   "kind": "data",
   "uri": "https://entrylib.org/p/2026.000013/v3/files/files/smape-target.results.csv",
   "version": "sha256:dbcda16de8d13b8449906fa1142665fc1efc4b472ebe082233f3e457c2eb65eb",
   "availability": {
    "level": "public"
   },
   "description": "Every result of the lab record (17182 results of 17 runs) at full precision: the lab result's ID, its run, the quantity as the run's script named it, its conditions and its value (both as JSON). The evidence's tables give the values to four significant digits.",
   "columns": [
    {
     "name": "lab_result",
     "dimension": "the result event of the lab record"
    },
    {
     "name": "run",
     "dimension": "the lab run that produced it"
    },
    {
     "name": "quantity",
     "dimension": "the quantity, as the run's script named it"
    },
    {
     "name": "conditions",
     "dimension": "the result's conditions, as JSON"
    },
    {
     "name": "value",
     "variable": "the value of the quantity, as JSON, at full precision"
    }
   ],
   "licence": "CC-BY-4.0"
  }
 ],
 "references": [
  {
   "id": "B8",
   "target": "10.1016/0169-2070(93)90079-3",
   "id_type": "doi"
  },
  {
   "id": "B13",
   "target": "10.3390/forecast8020024",
   "id_type": "doi"
  },
  {
   "id": "B14",
   "target": "10.1016/j.ijforecast.2020.06.008",
   "id_type": "doi"
  }
 ]
}
