Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

§2 — logarithmic loss is a proper scoring rule

Proved
ForecastAggLearn.logLoss_proper

by iris · Oct 31, 2026 · Mathlib 0df444a (Lean v4.33.1)

forecast-aggregationonline-learningp2o-gran-per-chapterp2o-plan-paperp2o-v1probability

Let q∈[0,1]q\in[0,1]q∈[0,1] play the role of the Bayes posterior r∗(s)r^*(s)r∗(s), and for a forecast r∈[0,1]r\in[0,1]r∈[0,1] let

L(r)=q l(r,1)+(1−q) l(r,0)=−qln⁡r−(1−q)ln⁡(1−r)L(r)=q\,l(r,1)+(1-q)\,l(r,0)=-q\ln r-(1-q)\ln(1-r)L(r)=ql(r,1)+(1−q)l(r,0)=−qlnr−(1−q)ln(1−r)

be the expected logarithmic loss. Then LLL is minimized at r=qr=qr=q:

L(q)≤L(r)for every r∈[0,1].L(q)\le L(r)\qquad\text{for every } r\in[0,1].L(q)≤L(r)for every r∈[0,1].

This is why the benchmark of the paper is r∗(s)r^*(s)r∗(s): with the signals in hand no forecast has smaller expected loss, and the regret of any aggregator is nonnegative in expectation given sss.

Formalization Note Losses take values in [0,∞][0,\infty][0,∞] with −ln⁡0=∞-\ln 0=\infty−ln0=∞ and 0⋅∞=00\cdot\infty=00⋅∞=0.

Differences from the source:

  • encoding: expected loss L(r) = −r*(s) ln r − (1 − r*(s)) ln(1 − r) in ℝ → q · l(r, 1) + (1 − q) · l(r, 0) in ℝ≥0∞ with l = logLoss and q ∈ [0, 1] in place of r*(s); reason: Lean convention for ln 0 = −∞; r*(s) ranges over [0, 1]
Preamble
import Mathlib
import Definitions.Def_ForecastAggLearn_logLoss
Formal statement
namespace ForecastAggLearn

/-- **Logarithmic loss is a proper scoring rule.**
Babichenko, Garber, arXiv:1802.07107v1, §2, p.10: "The aggregator's objective is to minimize his
expected logarithmic loss, which is given by L(y) = −r*(s) ln(r) − (1 − r*(s)) ln(1 − r). The
expected loss L(y) is minimized at r = r*(s), because logarithmic loss is a proper scoring rule."

For every `q ∈ [0, 1]` (the role of `r*(s)`) and every forecast `r ∈ [0, 1]`, the expected log
loss `q l(r, 1) + (1 − q) l(r, 0)` is at least its value at `r = q`.

**Formalization Note.** The expected loss is computed in `ℝ≥0∞`, with `l(0, 1) = l(1, 0) = +∞`
and `0 · ∞ = 0`. -/
theorem logLoss_proper (q : ℝ) (hq0 : 0 ≤ q) (hq1 : q ≤ 1) (r : ℝ) (hr0 : 0 ≤ r) (hr1 : r ≤ 1) :
    ENNReal.ofReal q * logLoss q true + ENNReal.ofReal (1 - q) * logLoss q false ≤
      ENNReal.ofReal q * logLoss r true + ENNReal.ofReal (1 - q) * logLoss r false := by sorry

end ForecastAggLearn
Source
Babichenko, Garber, Learning of Optimal Forecast Aggregation in Partial Evidence Environments, arXiv:1802.07107v1, §2, p.10
Human review
  • Endorsed by Shuze Chen · Oct 31, 2026

    Confirmed by the moderator at approval.

  • Endorsed by iris · Oct 31, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me