kuntal-r-d Earnings Forecast
Prompt-first, pure-prompt skill. No script — project from the supplied earnings history. Always state assumptions and a confidence band; never present a point estimate as certain.
Role & objective
Project near-term earnings (next quarter / next year) with a growth rate, an earnings-quality read, and an explicit uncertainty band. Output a normalised score for the earnings outlook.
When to use
"Forecast GP's EPS", "next-year earnings estimate", "earnings growth outlook", "is the earnings
quality good?". Feed the growth view into fundamental-analysis.
Inputs you need
fundamentals — eps_history[] (≥12 quarters ideal), revenue_history[], eps_ttm,
margins, and optional analyst estimates/earnings_surprise history.
Method (follow in order)
- Historical trend — growth rate from EPS history; detect seasonality and cyclicality.
- Projection — simple trend/exponential-smoothing projection of next quarter & year; widen the band when history is short or volatile.
- Earnings quality — accruals vs cash earnings, one-time items, core vs reported EPS.
- Growth drivers — revenue growth, margin trend, operating leverage.
- Estimate context — consensus, revision trend, surprise/beat history if supplied.
Scoring rubric
score = clamp(projected growth / 0.15, −1, 1) (≈15% growth → full positive), adjusted down for poor earnings quality. Rating: ≥0.5 strong_growth · ≥0.15 growth · −0.15..0.15 flat · ≤−0.15 declining. Confidence is low by default (forecasting is uncertain): higher with ≥12 quarters, stable margins, and consistent beats; lower with short/volatile history (state the band).
Output (emit this Thinking Card)
{ "skill": "earnings-forecast", "ticker": "..", "mode": "investment", "as_of": "..",
"score": 0.0, "confidence": 0.0, "rating": "strong_growth|growth|flat|declining",
"key_metrics": { "ttm_eps": 0, "next_year_eps": 0, "growth_rate": 0.0,
"band_low": 0, "band_high": 0, "earnings_quality": 0.0 },
"reasoning": ["assumptions + drivers"], "flags": ["short_history?", "low_earnings_quality?"],
"disclaimer": "Educational analysis only. Not financial advice. Forecasts are uncertain." }
DSE pitfalls
- DSE disclosure is often annual/semi-annual and lumpy — short history → wide band, low confidence.
- One-off gains (asset sales, revaluations) inflate reported EPS — strip them for core earnings.
- Don't extrapolate a single strong quarter; weight the trend and seasonality.
Optional precision helper
No bundled script — pure-prompt skill. Use fundamental-analysis for the valuation that
consumes this growth estimate.
Worked example
8 quarters, EPS CAGR ~16%, stable margins, 2 recent beats, clean accruals → next-year EPS +16% (band ±5%) → score ≈ +1.0 capped → strong_growth, confidence ≈ 0.6 (state the band).
References
See fundamental-analysis/references/VALUATION.md for how growth feeds fair value.
Output is educational analysis only, never financial advice.