AI Scorecard: 53% Accuracy for the Week of August 4th

By Yogurt · 2026-08-08 · AI Scorecard

Our weekly review of Yogurt's AI-driven earnings predictions. This week the engine verified 15 earnings reports and correctly called 8 outcomes — including ANIP, ATMU, GLP, and GTN — for a 53% accuracy rate. Full breakdown inside.

The Yogurt AI engine reviewed fifteen earnings reports for the week of August 4–8, 2026, correctly calling eight outcomes for a 53% accuracy rate. The model landed clean calls on pharmaceutical, industrial, and financial names but stumbled on several beat surprises that ran against bearish sector setups. Here is the full breakdown.

This Week's Results (August 4–8, 2026)

Here is the full breakdown of our predictions versus the actual reported outcomes:

Key Takeaways

• ANIP (ANI Pharmaceuticals): The AI correctly called a beat — ANI delivered strong specialty pharma revenues driven by its rare disease portfolio, confirming the bullish thesis around branded drug pricing power.
• ARKO (ARKO Corp): The model nailed the miss. Fuel margin compression and softening convenience store traffic weighed on ARKO's Q2 results, exactly as our bearish setup anticipated.
• ATMU (Atmus Filtration Technologies): A clean beat. Atmus benefited from robust aftermarket demand and disciplined cost control, rewarding the bullish prediction.
• CWAN (Clearwater Analytics): Correct miss call — Clearwater's revenue growth missed consensus as enterprise software deal cycles lengthened heading into Q3.
• DCH (Daiwa House Industry): The beat landed as expected, with Japanese housing and logistics development staying firm amid domestic demand tailwinds.
• CGON (Coda Octopus Group): Accurate miss prediction — the marine technology company reported weaker-than-expected project revenues as timing of contract completions slipped into H2.
• GLP (Global-e Online): Beat confirmed — cross-border e-commerce volumes accelerated in Q2 as summer travel and fashion seasons boosted international GMV.
• GTN (Gray Television): The AI correctly called the beat. Political advertising spend in advance of mid-cycle elections provided a meaningful upside surprise to local broadcasting revenues.
• ACMR (ACM Research) — MISS: The model predicted a miss but ACM Research surged past estimates, driven by robust wafer cleaning tool demand from Chinese semiconductor fabs operating at high utilization.
• ASIX (AdvanSix) — MISS: The AI called a beat but AdvanSix missed on weaker nylon and chemical pricing. Commodity input cost volatility caused margin compression the model underweighted.
• CLMT (Calumet Specialty Products) — MISS: The model expected a miss, but Calumet's renewable fuels segment delivered a surprising beat driven by higher RIN credits and improved SAF margins.
• ESNT (Essent Group) — MISS: The AI predicted a miss but the mortgage insurance company beat easily — credit quality remained pristine and new insurance written accelerated as housing activity picked up.
• AD (Ahold Delhaize) — MISS: The AI predicted an outright beat but Ahold Delhaize came in exactly at consensus — no upside surprise despite solid U.S. supermarket comparable sales. A frustrating near-miss.
• BTDR (Bitdeer Technologies) — MISS: The model called a beat but Bitdeer missed sharply as rising energy costs and Bitcoin hashrate difficulty compressed mining margins more than anticipated.
• FLR (Fluor Corporation) — MISS: The AI expected a miss but Fluor delivered a strong beat, powered by mega-project wins in energy transition infrastructure and better-than-expected project execution.

What This Week Tells Us

A 53% accuracy rate on 15 predictions is a humbling reminder that earnings season's mid-summer phase brings elevated unpredictability. The model found its footing on names with strong analyst consensus and clear sector tailwinds — pharmaceutical (ANIP), filtration (ATMU), and media (GTN) — but was wrong-footed on commodity-sensitive names (ASIX, BTDR) and several positive surprises in historically beaten-down sectors (ACMR, CLMT, ESNT, FLR).

The pattern of miss-predictions turning into beats suggests the market may be pricing in more pessimism than fundamentals justify at mid-year. That's a signal the AI is tracking — and calibrating for heading into the busy back half of earnings season.

Cumulative AI Track Record

Over all scored weeks since launch, the Yogurt AI engine has correctly called 69 out of 112 verified predictions for an all-time accuracy rate of 62%. The engine continues to refine its models as more results are verified against actual reported data.

May the Schwartz be with your portfolio. 🧘‍♂️