Dispute Risk Intelligence

Stop trading into disputes

OracleMangle scores every Polymarket and Kalshi market for resolution dispute risk — before you enter. Markets we flag get disputed at 8x the baseline rate.

Start Free on Telegram See the API

Free tier available. No credit card required.

164K+
Markets scored
8.2x
Dispute detection
in top risk bucket
0.80
AUC-ROC
accuracy
1,654
Actual disputes
in training data
The Problem

Your bot checks price, volume, and close date.
It doesn't check for disputes.

Prediction market disputes lock up capital for weeks. A trader with $200K across 40 positions who gets $50K frozen in 2 disputed markets isn't just losing money — they're losing every trade they can't make with that capital.

The risk nobody prices

Every trader on Polymarket sees the same odds. None of them see the resolution risk. Questions with subjective criteria, vague thresholds, or unreliable sources get disputed — and your capital gets frozen while the UMA oracle deliberates.

One signal changes everything

OracleMangle analyses question wording, resolution source reliability, and historical precedent across 164K+ markets. One API call or Telegram alert tells you: trade it or skip it. Before the dispute happens.

Proof It Works

We caught these disputes before they happened

Real alerts from real data. Every market below was flagged as high risk by our model before the dispute was filed.

Disputed "Biden senile during the debate?"
Our model scored this 75% dispute risk. Resolution required that Biden "exhibits clear signs of senility" — a subjective medical judgment with no defined criteria. No agreed-upon threshold. The UMA oracle couldn't resolve it cleanly. Money was frozen during the dispute.
Risk: 75% Outcome: Disputed Platform: Polymarket
Disputed 2x "Will Gaza ceasefire deal be reached before Dec 31 2024?"
Scored at 75% risk. "Ceasefire deal" has no universally agreed definition. Does a temporary pause count? A partial agreement? Two separate disputes were filed. Capital was frozen twice.
Risk: 75% Outcome: Disputed 2x Platform: Polymarket
High Risk "Will the Iranian regime fall before 2027?"
$12.4M in volume. What counts as "fall"? A coup? A revolution? A change in supreme leader? The resolution criteria require subjective interpretation of regime change. Markets like this get disputed at 8x the baseline rate.
Risk: 55% Volume: $12.4M Platform: Polymarket
Clean Pick "Will France win the 2026 FIFA World Cup?"
Risk score: 5%. Resolution is binary — FIFA officially declares the winner. No ambiguity in outcome, source, or timeframe. $373K volume. Safe to trade without dispute risk concerns.
Risk: 5% Volume: $373K Platform: Polymarket

Model calibration

Dispute rate increases monotonically with our risk score. Higher score = more disputes. Every bucket.

Risk Bucket Markets Actually Disputed Dispute Rate vs Baseline
Clean (0-10%) 9,493 141 1.5% 0.6x
Medium (10-25%) 4,130 125 3.0% 1.3x
High (25-50%) 957 72 7.5% 3.2x
Extreme (50%+) 51 10 19.6% 8.2x
How It Works

Three steps to dispute-proof your portfolio

1

We scan every market

Continuous scraping of Polymarket and Kalshi. Every open market is analysed for resolution ambiguity, source reliability, and historical precedent from 1,654 actual disputes.

2

AI scores dispute risk

Our model (AUC 0.80, benchmarked against 4 LLMs) assigns a 0-100% dispute risk score. RAG-enhanced analysis compares each market to similar historical precedents.

3

You get the signal

Via Telegram alerts for humans or a JSON API for bots. Risk spikes, clean picks, and plain-English explanations of exactly why a market is risky. Before the crowd knows.

For Bot Operators

One line of code. One fewer dispute.

Your bot manages hundreds of positions. Add a dispute risk check before every entry. If risk is above your threshold, skip it.

# Add to your Polymarket bot
import requests

# Check dispute risk before entering
risk = requests.get(
  f"https://api.oraclemangle.com/v1/markets/{market_id}",
  headers={"X-API-Key": API_KEY}
).json()

if risk["dispute_risk"] > 0.25:
  skip_market() # High risk — don't enter
elif risk["safe_to_trade"]:
  enter_position() # Clean — go

Your bot doesn't check for disputes. Now it does.

The API returns JSON with a dispute risk score, trend analysis, and plain-English explanation. Integrate in minutes, not days.

  • 200ms average response time
  • Batch queries for portfolio-wide screening
  • Webhook push for real-time risk spikes
  • Trained on 164K+ historical markets
Pricing

Less than one frozen position costs you

One avoided dispute saves more than a year of Pro. Cancel anytime.

Free

See if it works for you
$0
No credit card needed
  • 3 alerts per week
  • 1-hour delay vs Pro
  • Risk scores only
  • No AI explanations
  • No API access
  • No watchlist
Start Free
Most Popular

Pro

For active traders
$49 /mo
7-day free trial
  • Unlimited real-time alerts
  • AI explanations with precedent analysis
  • Risk spike detection
  • Cross-platform divergence alerts
  • 50-market watchlist
  • Weekly calibration report
Start 7-Day Trial

Institutional

For bots and desks
$299 /mo
30-day free trial
  • Everything in Pro
  • REST API access (JSON)
  • Batch queries and webhooks
  • Unlimited watchlists
  • Priority scoring for new markets
  • Direct support channel
Contact for Access
FAQ

Common questions

How accurate is the model?
AUC-ROC of 0.80, benchmarked blind against Gemini Flash, Claude Sonnet, Kimi K2, and Llama 4. Markets in our highest risk bucket (50%+) get disputed at 19.6% vs a 2.4% baseline — that's 8.2x more likely. The calibration is monotonic: higher score = more disputes, in every single bucket.
What data do you use?
164K+ scored markets across Polymarket and Kalshi, 1,654 actual disputes as ground truth, UMA governance events, and a RAG pipeline that retrieves similar historical precedents for every market scored. The model analyses question wording ambiguity, resolution source reliability, and historical dispute patterns.
Can I integrate this with my trading bot?
Yes. The Institutional tier includes REST API access with JSON responses, batch queries, and webhook push. One GET request returns dispute risk, trend analysis, and explanation. Average response time is 200ms. Add 3 lines of code to your bot.
What if Polymarket changes their oracle?
Our scoring is based on question text ambiguity, not the specific oracle mechanism. If Polymarket switches from UMA to another oracle, ambiguous questions will still get disputed. The signals we detect — subjective criteria, missing sources, vague thresholds — are universal to any resolution mechanism.
Why should I pay when I can check markets manually?
You check price, volume, and close date. Do you also read every word of the resolution criteria and compare it against 1,654 historical disputes? We do, at machine speed, for every market. If you trade more than 10 markets at a time, the time savings alone justify the cost.
What's the difference between Free and Pro?
Free gets you 3 alerts per week with a 1-hour delay and a risk score only. Pro removes the delay, makes it unlimited, and adds AI-generated explanations that tell you exactly why a market is risky — with historical precedent analysis. In a fast-moving dispute, 1 hour is the difference between exiting your position and having capital frozen.

Know which markets will blow up
before they do.

Join the Telegram bot. See your first risk score in 30 seconds.

Start Free on Telegram