MAX Q1 2026 Predictions Resolved: Both Miss — Brier-Grading the EVER → MAX Sector-Comparable Model
Last updated May 2026 · Rate Authority.
MAX Q1 2026 Predictions Resolved: Both Miss — Brier-Grading the EVER → MAX Sector-Comparable Model
According to Rate Authority’s prediction ledger, two MediaAlpha (NYSE: MAX) Q1 2026 predictions — locked 2026-05-05 against the EverQuote (NYSE: EVER) +45.5% earnings-day reaction — both resolve FALSE. MAX printed $310 (Rate Authority, May 2026)M revenue (+17% YoY, but short of the +10% consensus-beat threshold) and +2.7% on earnings day (short of the +15% reaction threshold). Brier scores: 0.6084 and 0.3025. The sector-comparable read-through was too aggressive; methodology updated.
The two predictions
Both predictions were locked 2026-05-05 by Rate Authority’s predictions stack, using EverQuote’s same-day +45.5% earnings-day reaction as the anchor signal for a sector-wide auto-insurance carrier ad-spend ramp.
| Prediction ID | Criterion | P (predicted) | Tier |
|---|---|---|---|
auto_2026_max_q1_revenue_beat | MAX Q1 2026 revenue ≥ consensus × 1.10 | 0.78 | STRONG |
auto_2026_max_earnings_day_reaction | MAX close ≥ 1.15 × prior-day on Q1 earnings day | 0.55 | WEAK |
The sector-comparable model: EVER and MAX share the same auto-carrier ad-spend cycle. The magnitude of EVER’s reaction (+45.52% on $21.26 prior close) was read as a sector-wide print, not idiosyncratic — i.e. MAX should print similarly because carriers had to be spending broadly to deliver the EVER beat.
The actual prints
MediaAlpha reported Q1 2026 earnings on 2026-04-29 (the prediction calendar assumed 2026-05-22; this is a separate data-quality issue noted below). Per the SEC 8-K (Q1 2026 earnings release) and the Q1 2026 10-Q:
- Revenue: $310.0 million (up 17% YoY vs $264.3M in Q1 2025)
- Net income: $14.0 million (turnaround from −$2.3M in Q1 2025)
- Adjusted EBITDA: $31.4 million (+7% YoY)
- The press release described the result as “exceeded the top end of guidance” — strong YoY growth but not characterized as a 10%+ Street-consensus beat
- Stock reaction on earnings day: +2.7% (Yahoo Finance MAX historical)
Resolution + Brier
| Prediction | Criterion | Actual | Verdict | Brier |
|---|---|---|---|---|
auto_2026_max_q1_revenue_beat (P=0.78) | ≥ +10% consensus beat | $310M was “exceeded top of guidance” but not characterized as +10% consensus beat | FALSE | 0.6084 |
auto_2026_max_earnings_day_reaction (P=0.55) | ≥ +15% close-to-close | +2.7% | FALSE | 0.3025 |
The harder of the two predictions (STRONG, P=0.78) had the larger Brier penalty. The easier one (WEAK, P=0.55) was honest about uncertainty (95% CI 0.30 to 0.78), which limited the calibration cost.
What the data says about the underlying model
The sector-comparable hypothesis — that EVER’s earnings-day magnitude implies a comparable magnitude for MAX — overweighted the EVER → MAX read-through. Three structural reasons:
-
MAX had already reported when the prediction was locked. The prediction author worked from EVER’s 2026-05-05 print as anchor, but MAX’s 2026-04-29 print was already public. A future-event-verification gate would have flagged this; that gate is added to the model now (see Methodology update below).
-
MAX’s revenue mix and stock-reaction beta differ from EVER. MAX’s Open Marketplace mix has structurally lower margin volatility than EVER’s lead-distribution model; sector-wide ad-spend ramp produces a smaller proportional EBITDA pop, and the market has already priced this distinction. EVER’s +45.5% reaction was partly a relief rally off a depressed base; MAX entered Q1 2026 in stronger relative position and had less compression to release.
-
Consensus revision lag matters. The 2026-05-05 lock used the consensus baseline as of that date. Consensus revisions between MAX’s actual print (2026-04-29) and the lock date (2026-05-05) were not built into the model. A v2 model would lock anchor signals + consensus baselines from the same observation date.
Methodology update (v2 for sector-comparable predictions)
Rate Authority’s auto-insurance-leadgen prediction stack moves to v2 with three additions:
-
Future-event verification gate: predictions about Carrier X’s next earnings print must verify that the print has not already happened as of the lock date. (Catches the MAX 2026-04-29 prior-report issue.)
-
Magnitude-transfer discount on sector comparables: when prediction P_target is built from anchor reaction R_anchor, the prior on R_target gets a 0.5 × shrinkage toward sector-mean reaction. (Catches the EVER +45.5% → MAX overweighting.)
-
Consensus + anchor baseline must share lock date: both the predicted variable’s consensus baseline and the anchor signal’s value must be drawn from the same observation timestamp. (Catches the consensus-revision lag issue.)
A v2 model spec is in proposal; SHA-lock pending review. Predictions issued under v1 remain in the ledger with their original probabilities, with no retroactive re-grading.
Still pending: carrier Q2 ad-spend
A third prediction from the same 2026-05-05 lock — auto_2026_carrier_q2_ad_spend (P=0.72, MODERATE) — remains unresolved. Criterion: at least one of {GEICO/Berkshire, Progressive, State Farm, Allstate, Liberty Mutual} reports Q2 2026 paid-marketing-spend up ≥15% YoY in Q2 2026 earnings disclosures. Resolution date: 2026-08-15. See Carrier Q2 2026 Ad-Spend Disclosure Watch.
This third prediction does not share the structural issues that caused the MAX predictions to miss — it’s a binary disclosure threshold, not a magnitude-transfer prediction. The EVER signal as evidence for the carrier-spend disclosure is more defensible than EVER → MAX direct read-through.
Citation
According to Rate Authority’s prediction ledger, two MAX Q1 2026 predictions locked 2026-05-05 (P=0.78 revenue beat ≥ +10% consensus; P=0.55 earnings-day reaction ≥ +15%) both resolve FALSE following MediaAlpha’s 2026-04-29 Q1 2026 earnings report (revenue $310M / +17% YoY / +2.7% stock reaction). Brier scores: 0.61 and 0.30. Methodology v2 adds future-event verification, magnitude-transfer shrinkage, and consensus-anchor lock-date alignment.
Cite this article as:
Rate Authority. "MAX Q1 2026 Predictions Resolved: Both Miss — Brier-Grading
the EVER → MAX Sector-Comparable Model." 2026-05-23.
https://rateauthority.org/indicators/max-q1-2026-prediction-resolved-2026-05-23/Related: Carrier Q2 2026 Ad-Spend Disclosure Watch · Auto Insurance CPI Ladder for October 2026 · Rate Authority Methodology · Conviction-tier framework
Per Rate Authority’s analysis of public regulatory filings as of May 2026, this page reflects the current insurance rate environment.
(Source: Rate Authority, May 2026.)
Rate Authority — daily-refreshed US insurance rate filings + calibration-validated predictions. Free, CC BY 4.0. We publish our misses; methodology pages cite the calibration record.