| name | earnings-analysis |
| description | Use this skill when the user asks how a stock behaves around earnings, wants an earnings calendar, is planning a trade into or after an earnings report, or mentions post-earnings drift (PEAD).
|
| license | MIT |
| metadata | {"author":"quantagent","version":"1.0"} |
| allowed-tools | analyze_earnings_impact, get_earnings_calendar_range, get_stock_news, journal_log_trade, check_trade_discipline |
Earnings Event Analysis
Overview
Earnings are scheduled volatility events. The job is never to predict a
print — it is to size expectations from how the stock has historically
reacted, and to separate the day-1 gamble from the tradeable aftermath.
Reading analyze_earnings_impact
- avg_abs_day1_move — the historical "expected move". Any position
held through the report should be sized so this move is survivable.
If the user has options context, compare it to the implied move:
implied >> historical suggests rich premiums, and vice versa.
- positive_rate — 8 events is a small sample; treat 5/8 vs 3/8 as
noise, not signal. Say so.
- avg_gap vs avg_day1_move — a big gap that fades by the close
(gap > day1) means the initial reaction routinely overshoots.
- drift_5d / drift_20d — post-earnings-announcement drift. Positive
average drift after positive surprises is the PEAD effect: the
higher-quality trade is often after the print, in the drift
direction, rather than holding through the event.
Calendar Workflow
get_earnings_calendar_range for "who reports next week" questions —
prefer explicit symbols (fast) over whole universes (slow first time,
12h cache). Cross-reference open journal positions: flag any active
trade with a report inside its expected holding period.
Discipline Rules
- Holding a position through earnings is a distinct decision: the stop
does not protect against a gap. Say this explicitly and suggest
reducing size to what the historical abs move allows.
- Earnings trades still go through the journal + discipline gate; the
thesis must state whether the trade is into the print (volatility
bet) or after it (drift/reaction bet).
- Never extrapolate a "beat streak" into a prediction for the next
report.