Risk Disclosure
Read this document before trading. By creating a LAPOS account or using the Services, you acknowledge that you have read, understood, and accepted every risk described below. These risks are material. You can lose all of the funds you commit to prediction markets.
This Risk Disclosure forms part of the LAPOS Terms of Service by reference (Section 8). Terms not defined here have the meaning given in the Terms of Service.
1. General Risk of Loss
Prediction market trading involves a substantial risk of loss and is not suitable for every person.
- You may lose 100% of any funds you commit to a prediction market position. Losing a position that resolves against you returns zero — you do not partially recover the stake.
- Industry data consistently shows that most retail participants in prediction markets and derivatives lose money over time. This has been publicly disclosed by regulated exchanges, including Kalshi (see
https://kalshi.com/and its regulatory filings with the CFTC). There is no reason to believe LAPOS users are exempt from this base rate. - Do not trade with funds you cannot afford to lose entirely without impacting your financial wellbeing.
- Prediction markets are zero-sum before fees: for every winner there is a loser, and after fees the average participant is a net loser.
- Trading with leverage or with positions that exceed your liquid savings is especially dangerous.
2. Market Volatility Risk
Prediction market prices can move very quickly and by large amounts on unexpected news, rumors, or speculation.
- Prices can swing tens of percentage points in minutes when breaking news arrives (historical moves of 30–80 percentage points in minutes have been observed on major-news events on Polymarket).
- Even resolved outcomes can experience post-news price turbulence before the market closes.
- Gap moves are common: a price can "jump" past your intended exit point with no fills available at the old quote.
- Markets can be halted, reset, or extended by the underlying platform (Polymarket admin actions) without notice — your position may be frozen or re-priced mid-trade.
LAPOS does not shield you from volatility. Analytics like Health Badges and Lens analysis are tools, not prevention.
3. Oracle Risk (UMA Protocol)
Most Polymarket markets resolve through the UMA Optimistic Oracle. This introduces several risks unique to decentralized oracle governance:
- Resolution delays. Markets do not resolve automatically at event end. A UMA vote cycle typically takes 24-72 hours (commit + reveal phases). Complex or controversial markets can take days to weeks.
- Disputes. Any party posting a bond (currently set by UMA governance at approximately $750 USDC, subject to change) can dispute a proposed resolution, triggering a full DVM (Data Verification Mechanism) vote. Disputes can resolve against the initially-proposed outcome.
- Whale voter influence. A small number of UMA-token holders can sway DVM votes. Voters with >500K UMA stake or long-running track records have disproportionate influence. This is by design, but it means resolution is not purely deterministic.
- Manual intervention. Polymarket administrators can flag, reset, or manually resolve markets outside the UMA vote process. Historically, Polymarket has deferred to UMA consensus, but platform policy can change.
- Cost asymmetry. Mounting a dispute requires capital ($750 bond + time). A casual trader affected by a wrong resolution may not have the resources to dispute.
Your trade outcome depends on oracle resolution. If UMA resolves a market against your intuitive reading of the underlying event, you lose. There is no external appeal.
4. Admin Intervention Risk
The prediction market platform (primarily Polymarket) retains operational control over markets and can:
- Flag a market as disputed, freezing trading
- Reset a market to pre-defined state, cancelling outstanding orders
- Manually resolve a market in exceptional circumstances
- Modify resolution criteria mid-market (a "Clarification" update) — changing the rules your position was taken under
- Halt trading for maintenance or regulatory reasons
LAPOS's Rule Change Monitoring feature helps you detect changes, but detection is not prevention. You may only learn of a change after it has affected your position.
5. Liquidity and Slippage Risk
Many prediction markets have thin orderbooks.
- The "quoted price" you see may not be achievable for a meaningful size. A $1,000 sell on a market with $500 of bids at the top price will eat through the book and execute at much worse prices ("slippage").
- LAPOS's Auto-Shield includes a pre-sell slippage check (currently configured to reject sells where estimated slippage exceeds approximately 2%; threshold may be adjusted), but this means the sell simply won't execute — your position remains exposed.
- Illiquid markets can be unsellable at any reasonable price during critical moments. You may be unable to exit a losing position.
- Market makers can withdraw liquidity at any time, especially during high-stress events.
- Limit orders may sit unfilled for hours or forever.
6. Stablecoin (USDC) Risk
All trading on Polymarket uses USDC, a US-dollar-pegged stablecoin issued by Circle Internet Financial. USDC is not risk-free.
- Depeg risk. USDC has temporarily lost its $1 peg in the past. Notably, USDC traded as low as $0.87 during the March 2023 Silicon Valley Bank banking crisis (Circle held ~$3.3 billion in SVB). Recovery took days.
- Issuer risk. Circle's reserves, banking relationships, and regulatory status can change. Circle could fail, be sanctioned, or restructure.
- Freeze risk. Circle can freeze USDC at specific wallet addresses in compliance with legal orders. Your USDC could become immobilized.
- Redemption risk. Converting USDC back to USD cash depends on Circle's redemption facilities and off-ramp providers.
A USDC depeg or freeze during an active trade would cause losses independent of the prediction market outcome.
7. Smart Contract Risk
Polymarket's markets operate via smart contracts on the Polygon blockchain. Smart contracts carry inherent risks:
- Bugs and exploits. Smart contracts are code and can contain vulnerabilities. Exploited contracts can drain funds without recourse.
- Upgrade risk. Polymarket can deploy new contract versions. Legacy markets may behave differently than new ones. Migration errors can occur.
- Adapter contracts. The UMA-Polymarket adapter (
uma-ctf-adapter) is itself a smart contract bridging UMA votes to Polymarket outcomes. Adapter bugs could cause incorrect resolutions. - Immutability. Once deployed, most contract logic cannot be changed mid-transaction. A bug cannot be "patched during the trade."
LAPOS does not audit Polymarket's smart contracts. You are exposed to the full contract risk Polymarket users face.
8. Blockchain Risk
Polygon is a public blockchain network. Blockchain-specific risks include:
- Network outages and congestion. Polygon has experienced reorganizations and temporary halts. Transactions may be delayed, dropped, or reorganized.
- Gas fees. All transactions require the Polygon native gas token (POL, formerly branded MATIC) for gas. During congestion, gas prices spike; your transaction may fail or take minutes to confirm.
- Mempool exposure. Transactions sit in the public mempool before inclusion. Sophisticated actors can observe pending orders and trade against them (MEV, front-running, sandwich attacks).
- Transaction permanence. Once confirmed, a transaction is irreversible. LAPOS cannot roll back a mistakenly placed trade, a fat-fingered size, or a wrong-market click.
- Wallet loss. If you lose access to your Privy wallet (forgotten credentials, compromised recovery phrase if self-custody mode, device loss without backup), your USDC and positions may be permanently inaccessible.
- Bridge risk. If you move USDC to/from Polygon across a bridge (e.g., from Ethereum mainnet), bridge failures or exploits can cost you the funds. LAPOS does not operate a bridge.
9. Geographic Restriction Risk
Polymarket restricts access from a list of jurisdictions. The authoritative list is maintained by Polymarket at https://help.polymarket.com/ and can change at any time without notice from LAPOS. You must check the current list before each use.
At the time of drafting this Risk Disclosure (the "Effective date" at the top), jurisdictions restricted by Polymarket included (non-exhaustive and illustrative only — the Polymarket help-center list controls): United States, United Kingdom, Germany, France, Italy, Poland, Russia, Singapore, Thailand, Taiwan, Cuba, Iran, North Korea, Syria, Venezuela, and others. The count and identity of restricted jurisdictions can change at any time.
Key consequences:
- If you are located in a restricted jurisdiction, you cannot lawfully use the Services — LAPOS order routing will fail or be rejected.
- Using a VPN, proxy, or similar tool to circumvent Polymarket's geo-blocks violates Polymarket's Terms of Service Section 2.1.4. Polymarket uses detection that goes beyond IP address matching. Accounts detected circumventing restrictions may be frozen with funds locked indefinitely. Neither LAPOS nor anyone else can reverse Polymarket's enforcement.
- LAPOS cannot advise on the legal status of prediction markets in your jurisdiction. You are solely responsible for compliance with your local laws.
10. AI Analysis Risk
LAPOS uses AI systems from Anthropic (Claude model family) for Lens, Notebook Red Team, Daily Review, Auto-Shield AI enhancement, Blind Spot Report, and Rules Parser features.
AI output limitations you must internalize:
- Hallucinations. Large language models can confidently state facts that are false. A Lens analysis can cite a news event that did not happen, a Red Team critique can invent counter-evidence.
- Poor probability calibration. Academic and industry research consistently documents that LLMs produce poorly calibrated probability estimates for real-world forecasting tasks. A "73%" from the AI does not correspond to a 73% long-run frequency of occurrence.
- Training data cutoff. The model's world knowledge has a cutoff date. It may be unaware of events in the last weeks, months, or for recent model versions, longer.
- Consistency. The same question can yield different answers on different runs, even with temperature 0.
- Indirect prompt injection. Adversarial content in news articles can attempt to manipulate AI output. LAPOS employs Lakera Guard input screening on Sentinel-tier AI-driven execution as a mitigation, but defenses are imperfect. Recent academic research (Nasr et al., "The Attacker Moves Second: Stronger Adaptive Attacks Bypass Defenses Against LLM Jailbreaks and Prompt Injections," arXiv:2510.09023, October 10, 2025) documented that twelve defense techniques were bypassed with >90% success rates by adaptive attackers, indicating no single AI-safety measure is permanently reliable.
- Not financial advice. AI analysis is informational. It is not a recommendation to trade. It is not a forecast you can rely on. Your trading decisions are yours alone.
If you trade based on AI output and lose money, that is your responsibility. LAPOS is not liable for losses attributable to AI analysis, whether the analysis was accurate or not.
11. Automated Trading Risk (Auto-Shield, Conditional Orders, Auto-Forecast)
LAPOS offers features that execute trades without a per-trade human click:
- Auto-Shield (Pro and Sentinel) — sells positions on price, health, or AI-enhanced triggers
- Conditional Orders (Sentinel) — executes sells/buys on news, oracle, or price conditions
- Auto-Forecast (Sentinel) — AI-driven sell/buy/hold decision on breaking news
Automated trading amplifies risk. Historical precedents illustrate the category of harm:
- Knight Capital Group (2012): a widely-reported algorithmic trading error that caused the firm a reported loss of approximately $440 million in under an hour and effectively ended the firm as an independent entity. See SEC and public reporting archives.
- Multiple regulated financial firms (including large US robo-advisers) have faced SEC or similar regulatory enforcement actions over the past decade for automated-system defects, misstatements about automated features, or coding errors in automated advisory or trading products. See the SEC press-release archive at
https://www.sec.gov/news/pressreleasesfor primary-source information on specific cases. - Multiple crypto-native trading bots and automated-strategy products have liquidated user accounts due to flash crashes, stale price data, oracle glitches, or misfiring conditions.
Specific figures and dates for individual cases are intentionally omitted here because the authors of this document were not able to verify them against primary SEC filings at the time of drafting. The category-level risk — that automated financial systems can and do fail, sometimes catastrophically — is well-established regardless of the exact numbers in any particular case.
LAPOS safeguards. We implement multiple defenses including VWAP price validation, pre-sell slippage checks, freshness gates on price data, UMA dispute auto-pause, user-configurable daily loss caps, kill switches accessible from three distinct UI locations, and a mandatory 2-step consent flow before activation. These defenses reduce but do not eliminate automated-trading risk. Edge cases, infrastructure outages, API changes, or novel attack vectors may cause Auto-Features to fire incorrectly or fail to fire when expected.
Your responsibilities if you use Auto-Features:
- Set realistic trigger thresholds; wide bands reduce false positives.
- Set a conservative daily loss cap; use the feature rather than disabling it.
- Monitor Auto-Feature behavior actively, especially during high-volatility events.
- Be prepared for the feature to fail during Polymarket, Polygon, Anthropic, or LAPOS outages. Have a plan for manual intervention.
- Do not enable Auto-Features on positions whose unexpected liquidation would cause material hardship.
12. Systemic and Infrastructure Risk
LAPOS is a software layer built on a stack of third-party infrastructure. Failures or changes in any layer can affect your ability to trade or recover:
| Dependency | Failure mode | Impact |
|---|---|---|
| Polymarket | API changes, rate limits, restricted-country expansion, insolvency | Orders fail to route, positions become unmanageable |
| Polygon | Network halts, reorganizations, gas spikes | Trades delayed, fail, or priced unexpectedly |
| Anthropic | API outages, content filter changes, price increases | AI features unavailable or degraded |
| Privy | Authentication outage, key enclave compromise | You cannot log in or sign transactions |
| Stripe / CryptAPI | Payment processor outage | You cannot renew subscriptions or top up credits |
| Hetzner / Cloudflare | Hosting or DDoS protection failure | LAPOS website and API unavailable |
| Regulatory | CFTC, SEC, EU, or national regulator actions | Prediction markets may be restricted in more countries with little notice |
During a multi-hour LAPOS or upstream outage, your positions remain on-chain but unmanaged. Losing-positions can deteriorate without intervention. Plan accordingly.
13. Tax and Reporting Risk
Prediction market gains and losses may be taxable events in your jurisdiction.
- LAPOS does not provide tax advice.
- LAPOS does not file tax reports on your behalf.
- You are responsible for tracking your gains, losses, and any reporting obligations (including under US IRS rules for cryptocurrency or derivatives, or equivalent rules in your country).
- Depending on jurisdiction, prediction market income may be treated as gambling, short-term capital gains, ordinary income, or "other income" — consult a tax professional.
14. No Recovery
Because Polygon blockchain transactions are permanent and irreversible, there is no mechanism by which LAPOS can "undo" a trade. This applies to:
- Trades you placed and regret
- Trades Auto-Features executed on your behalf
- Trades executed at unexpected prices due to slippage, MEV, or volatility
- Trades executed based on AI analysis that turned out wrong
- Trades executed during a LAPOS bug or infrastructure issue
Our Limitation of Liability (Terms of Service Section 17.2) caps our aggregate liability at the lesser of (a) the total amount you paid LAPOS in the twelve months preceding the claim, or (b) US$1,000. The free-tier cap is US$100. If you experience a significant trading loss, your recovery from LAPOS is effectively bounded by that cap, even if you believe LAPOS was at fault. Risk Disclosure language updated 2026-04-21 to match ToS §17.2 verbatim (prior wording inverted the direction and used the wrong floor).
15. User Responsibility
By using the Services, you represent and agree that:
- You understand the risks described above in full.
- You are trading with funds you can afford to lose entirely.
- You have independently evaluated the suitability of prediction market trading for your financial situation, risk tolerance, legal jurisdiction, and tax profile.
- You do not rely on LAPOS analytics, AI output, or any tool as financial advice.
- You accept full responsibility for all trades you authorize, whether manually or via Auto-Features.
- You will not hold LAPOS responsible for trading losses, beyond the limits set by the Terms of Service and applicable law.
16. Acknowledgment
Creating a LAPOS account and using the Services constitutes your acknowledgment that you have read and understood this Risk Disclosure.
For Auto-Features specifically, the in-app 2-step consent flow constitutes an additional, feature-specific acknowledgment of the automated-trading risks described in Section 11.
If you do not accept the risks above, do not use LAPOS.
17. Contact
Questions about this Risk Disclosure: [email protected].