Geopolitics Sentiment Reversal Trader
This is a template.
The default signal is purely price-based with a staleness (days-to-resolution) multiplier -- no external API required. The skill discovers geopolitical markets at probability extremes, identifies likely overreactions using time-horizon analysis, and trades mean reversion.
The skill handles all the plumbing (market discovery, geopolitics filtering, trade execution, safeguards). Your agent provides the alpha.
Strategy Overview
Geopolitical prediction markets are uniquely susceptible to overreaction. When breaking news hits -- a military strike, a sanctions announcement, a ceasefire collapse -- retail traders rush to reprice markets. The result is a predictable pattern: markets overshoot to probability extremes (>92% or <8%) and then revert 30-50% of the move within 24-48 hours.
This skill systematically identifies and trades these overreactions by combining two signals:
- Probability extremes: Markets pushed beyond the reversal zones (>92% or <8%) where overreaction is statistically most likely.
- Staleness factor: Markets with long time horizons (30-180 days) at extreme prices are far more likely to be overreactions than markets with 2-3 days to resolution where the extreme price may reflect genuine resolution information.
Edge Thesis: Behavioral Finance Overreaction Bias
The overreaction hypothesis (De Bondt & Thaler, 1985) documents that markets systematically overshoot in response to dramatic news and subsequently revert. In geopolitical prediction markets, this bias is amplified by three mechanisms:
1. Availability heuristic under fear
Breaking geopolitical news (airstrikes, troop movements) activates fear responses. Retail traders anchor to the most dramatic possible outcome and price accordingly. A single airstrike pushes "Will there be a full-scale invasion?" from 40% to 93% -- even though the base rate for escalation from single strikes to full invasion is historically low.
2. Asymmetric attention
The news that pushes a market to 95% gets wall-to-wall coverage. The slow de-escalation that brings it back to 70% happens quietly over days. Retail attention is front-loaded on the shock, creating a predictable overshoot-and-decay pattern.
3. Time-horizon mispricing
A market at 92% with 180 days to resolve embeds enormous uncertainty that retail ignores in the heat of the moment. The staleness factor captures this: longer horizons mean more room for reversion, more intervening events, and more time for the initial overreaction to wash out.
Signal Logic
Step 1 -- Market Discovery
Keyword sweep across geopolitical topics: war, ceasefire, military, strike, Iran, Israel, Gaza, Lebanon, sanctions, nuclear, troops, conflict, attack, bomb, invasion, Hezbollah, Hamas.
A regex filter then confirms the market question is genuinely geopolitical (avoids false positives from keywords appearing in non-geopolitical contexts).
Step 2 -- Signal Gates
- Spread <= MAX_SPREAD (wide spreads eat the reversal edge)
- Days to resolution >= MIN_DAYS (standard minimum)
- Days to resolution >= MIN_DAYS_FOR_REVERSAL (need enough runway for reversion)
Step 3 -- Signal Direction
| Condition | Trade | Rationale |
|---|
| p >= REVERSAL_ZONE_HIGH (92%) and p >= NO_THRESHOLD | Buy NO | Market is "too certain" -- expect reversion DOWN |
| p <= REVERSAL_ZONE_LOW (8%) and p <= YES_THRESHOLD | Buy YES | Market is "too pessimistic" -- expect reversion UP |
| Between zones | Skip | Not at an overreaction extreme |
Step 4 -- Staleness Factor
The staleness factor is the key innovation. It scales from 0.0 to 1.0 based on days to resolution:
| Days to resolution | Staleness factor | Interpretation |
|---|
| 7 (minimum) | 0.0 | Near-term -- extreme price may be correct |
| 30 | 0.28 | Some room for reversion |
| 60 | 0.64 | Substantial overreaction likely |
| 90+ | 1.00 | Maximum -- extreme is almost certainly an overreaction |
Formula: staleness = min(1.0, (days - MIN_DAYS_FOR_REVERSAL) / (90 - MIN_DAYS_FOR_REVERSAL))
Step 5 -- Conviction Sizing
For NO (high extreme reversion):
raw_conviction = (p - REVERSAL_ZONE_HIGH) / (1 - REVERSAL_ZONE_HIGH)
conviction = staleness * raw_conviction
size = max(MIN_TRADE, round(conviction * MAX_POSITION, 2))
For YES (low extreme reversion):
raw_conviction = (REVERSAL_ZONE_LOW - p) / REVERSAL_ZONE_LOW
conviction = staleness * raw_conviction
size = max(MIN_TRADE, round(conviction * MAX_POSITION, 2))
How Sizing Works
With defaults (REVERSAL_ZONE_HIGH=92%, REVERSAL_ZONE_LOW=8%, MIN_TRADE=$5, MAX_POSITION=$40):
High extreme reversion (buy NO):
| Market price p | Days left | Staleness | Raw conviction | Final conviction | Size |
|---|
| 93% | 14d | 0.08 | 12.5% | 1.0% | $5 (floor) |
| 95% | 30d | 0.28 | 37.5% | 10.5% | $5 (floor) |
| 95% | 90d | 1.00 | 37.5% | 37.5% | $15 |
| 98% | 60d | 0.64 | 75.0% | 48.0% | $19 |
| 98% | 120d | 1.00 | 75.0% | 75.0% | $30 |
Low extreme reversion (buy YES):
| Market price p | Days left | Staleness | Raw conviction | Final conviction | Size |
|---|
| 7% | 14d | 0.08 | 12.5% | 1.0% | $5 (floor) |
| 5% | 30d | 0.28 | 37.5% | 10.5% | $5 (floor) |
| 3% | 90d | 1.00 | 62.5% | 62.5% | $25 |
| 1% | 120d | 1.00 | 87.5% | 87.5% | $35 |
Remix Ideas
- GDELT event velocity: Wire in GDELT's real-time event stream to measure the rate of new events. A single spike followed by silence is the classic overreaction pattern. Sustained event velocity means the market move may be justified -- reduce conviction.
- Twitter/X sentiment decay: Track sentiment volume on geopolitical hashtags. If sentiment peaked 12-24h ago and is declining, the overreaction is already unwinding -- increase conviction for the reversal trade.
- Historical reversion rate by event type: Airstrikes revert faster than sanctions announcements. Build a lookup table of event-type-specific reversion rates to weight conviction.
- Cross-market correlation: If multiple geopolitical markets spike simultaneously on the same news event, the overreaction is likely correlated. Trade the most extreme one and skip the rest to avoid correlated risk.
- Volatility regime detection: During periods of sustained geopolitical escalation (war phases), extreme prices may be correct. Add a regime filter that reduces conviction when multiple markets have been at extremes for >7 days (sustained crisis vs. flash overreaction).
Safety & Execution Mode
The skill defaults to paper trading (venue="sim"). Real trades only with --live flag.
| Scenario | Mode | Financial risk |
|---|
python trader.py | Paper (sim) | None |
| Cron / automaton | Paper (sim) | None |
python trader.py --live | Live (polymarket) | Real USDC |
autostart: false and cron: null -- nothing runs automatically until you configure it in Simmer UI.
Required Credentials
| Variable | Required | Notes |
|---|
SIMMER_API_KEY | Yes | Trading authority. Treat as high-value credential. |
Tunables (Risk Parameters)
All declared as tunables in clawhub.json and adjustable from the Simmer UI.
| Variable | Default | Purpose |
|---|
SIMMER_MAX_POSITION | 40 | Max USDC per trade (reached at 100% conviction) |
SIMMER_MIN_VOLUME | 15000 | Min market volume filter |
SIMMER_MAX_SPREAD | 0.08 | Max bid-ask spread (8%) |
SIMMER_MIN_DAYS | 3 | Min days until resolution (standard gate) |
SIMMER_MAX_POSITIONS | 6 | Max concurrent open positions |
SIMMER_MIN_TRADE | 5 | Floor for any trade (min USDC regardless of conviction) |
SIMMER_YES_THRESHOLD | 0.38 | Standard YES gate -- only buy YES below this |
SIMMER_NO_THRESHOLD | 0.62 | Standard NO gate -- only buy NO above this |
SIMMER_REVERSAL_ZONE_HIGH | 0.92 | Above this probability, market is in overreaction zone (high) |
SIMMER_REVERSAL_ZONE_LOW | 0.08 | Below this probability, market is in overreaction zone (low) |
SIMMER_MIN_DAYS_REVERSAL | 7 | Min days to resolution for reversal trades (need runway for reversion) |
Dependency
simmer-sdk by Simmer Markets (SpartanLabsXyz)
