Polymarket Bot Tutorial · Chapter 26 of 32
Election and politics market bots on Polymarket: US 2024 lessons, 2028 setup, EU and UK markets, Trump-related markets, polling-vs-market spread, long-hold sizing, and code for monitoring multiple election markets.
What this chapter covers
Politics is the highest-volume Polymarket category. Strategies range from multi-month hold positions on national elections to headline-arb on individual political news. This chapter is the honest separation of which approaches survive and which lose.
- Why politics is the highest-volume Polymarket category
- Long-hold vs short-hold political bots
- Polling-vs-market spread
- News flow and headline arb
- International elections (EU, UK, India)
- Risk: black-swan polls
- Code: scan politics markets daily
Why politics is the highest-volume Polymarket category
Politics is Polymarket's flagship category. The 2024 US presidential cycle pushed total Polymarket volume from $200M lifetime in early 2024 to $9B+ by November. Lifetime through 2026 sits at $63B+.
Why: the audience overlaps with crypto traders, the event horizons are long enough that price discovery has many cycles, and the media coverage drives continuous order flow. Other markets compete for attention; politics defines the platform.
Bot economics: even with high competition, the political markets are deep enough that a $1k position doesn't significantly move the price. Plenty of room for retail bots if the edge is real.
Long-hold vs short-hold political bots
Two strategy axes for politics.
- Long-hold (weeks to months): take a position based on a fundamental view, hold through news cycles, exit near resolution. Edge: model accuracy + risk tolerance.
- Short-hold (minutes to hours): react to specific news events (debate moments, court rulings, polling drops). Edge: latency + interpretation speed.
Most retail bots try short-hold first and fail because the competition is fierce. Long-hold is less crowded but requires the operational discipline to ignore daily noise. The honest assessment: long-hold is harder psychologically but easier mechanically. Pick based on your own behavior.
Polling-vs-market spread
A common trade is shorting the candidate whose poll average leads the market price implied probability - or vice versa.
Example: 538's election forecast says 65% for Candidate A on Oct 15; Polymarket trades A at 55%. The 10c spread implies the market thinks 538 is wrong, OR there is information the market knows that 538 has not absorbed yet.
The trade rule: if the spread persists for 7+ days with no news that should explain it, the market is wrong; take the long side at market price. Wait at least a week before acting on a new gap; news that closes the gap arrives daily.
News flow and headline arb
Specific news types that consistently move political markets enough to bot.
- Polling drops by NYT, Reuters, Bloomberg - major sources move the market 1-3c on release.
- Court rulings on candidate eligibility, ballot access, indictments. Move 3-8c instantly.
- Debate moments - gaffes, strong performances. Move 2-5c during the debate itself.
- Health events - anything that creates real candidate-replacement risk. Move 10-20c on confirmation.
Bot pattern: subscribe to a curated list of news sources, classify event type via keyword, place sized FOK on the predicted direction. Chapter 14 covers the general pattern; political markets are the most rewarding application of it.
International elections (EU, UK, India)
Polymarket lists international elections with varying volume. UK and EU elections typically attract $1-10M each; Indian elections (which interest the much larger user base) can attract $50M+. Latin American elections are typically thinner.
Edge profile: international markets are less efficient than US markets because the trader population is more US-centric. A bot with specific country expertise (a German political analyst, an Indian local) has a real edge in those markets that they don't have in US politics.
If you're not from the country or don't have specific expertise, international markets are not bot territory - the lack of inefficiency relative to US is not the same as more inefficiency. Domain knowledge dominates.
Risk: black-swan polls
The catastrophic political-bot failure mode is a black swan poll or event that moves the market 15-25c in 24 hours. A bot heavily positioned on one side gets crushed.
Examples:
- 2024 Biden debate performance → Biden replacement implied probability moved from 8% to 65% inside two weeks.
- Major candidate health events historically move markets 15-30c within hours.
- Indictment or major scandal can move 10-20c within a day.
Defenses: never have more than 20% of capital on a single political position; use stop-loss orders even though they're imperfect on prediction markets; halt new entries when implied volatility exceeds 2x baseline.
Code: scan politics markets daily
Reference: daily scan of high-volume politics markets, alert on significant moves.
def daily_politics_scan():
events = gamma_events(tag_id=2, limit=100, order="volume24hr")
for ev in events:
for m in ev["markets"]:
prev = load_last_snapshot(m["slug"])
curr = float(json.loads(m["outcomePrices"])[0])
if abs(curr - prev) > 0.05:
alert(f"big move on {m['slug']}: {prev:.2f} → {curr:.2f}")
save_snapshot(m["slug"], curr)The alert is the trigger for human review (or for downstream bots that subscribe to your alert feed). Don't auto-trade on big moves - political moves are usually news-driven and the bot needs context the alert doesn't carry.
