# AI trading claim check: returns claims vs evidence > 18 performance claims from AI trading products, each with its source, what it's based on and what anyone can verify. 6 are backtests or hypothetical. Canonical: https://tradingagentstack.com/ai-trading-claims/ 1. [Home](https://tradingagentstack.com/) 2. Claim check Claim check · as of Oct 29, 2026 # AI trading claim check AI trading products advertise big numbers. We record exactly what each one says, what it says the number is based on, and what anyone can actually check. This is a record, not an accusation: a claim with no independent evidence may still be true. **18**claims checked from 11 products **6**backtest, simulated or hypothetical **11**with no independent record **1**a third party can verify *A*Measured by us We computed the real number from public chain or venue data.*B*Venue-verified A venue or auditor reports the real account.*C*No independent record The product's own number; nothing a third party can check.*D*Backtest or hypothetical Simulated, hypothetical or backtested, by the product's own description. ### Akka Hyper AI (Hyperliquid vault) *A*Measured by us “targeting 15–20% monthly returns” Where[The vault leader's own description on Hyperliquid](https://app.hyperliquid.xyz/vaults/0x0e008684ae576f280c5426a89d3f5e1da1fc7398)SeenOct 28, 2026Claim datedNot shownBased on (their words)A target, not a resultWhat anyone can checkOur own number from Hyperliquid's public data (see below).What would verify itAlready verifiable: compare the target with the vault's measured monthly return. **Our number, from Hyperliquid's public data:** +9.8% from Nov 12, 2025 to Oct 28, 2026, about +0.8% a month, with a 5.0% max drawdown. [See it on the leaderboard](https://tradingagentstack.com/ai-trading-leaderboard/#ai-vaults). ### [LevelFields](https://tradingagentstack.com/tool/levelfields/) *D*Backtest or hypothetical “delivered 7,000% cumulative returns over the past 3 years” Where[Official Premium Alerts page (page description; the body text says 4 years)](https://levelfields.ai/premium-alerts)SeenOct 28, 2026Claim datedNot shown (closed trades since Sept 2022)Based on (their words)Hypothetical: the linked trade log calls it hypothetical P&L on a $1,000 stake per alertWhat anyone can checkA trade log hosted by the product itself, labelled hypothetical; not independent. [Link](https://tradelog-levelfields.grok.me/)What would verify itA complete timestamped alert archive with entry and exit rules; a real account record for actual returns. ### [Prospero.ai](https://tradingagentstack.com/tool/prospero-ai/) *D*Backtest or hypothetical “picks win at a 60% rate vs S&P500 and beating S&P by 81% annualized” Where[Official home page (investing newsletter)](https://www.prospero.ai/)SeenOct 28, 2026Claim datedAs of 07/15/25Based on (their words)Hypothetical: the footer says results are not actual trading resultsWhat anyone can checkHeadline statistics only; no list of picks. [Link](https://prosperoai.substack.com/)Their disclaimerHypothetical or simulated performance results have certain inherent limitations.What would verify itA dated list of every pick with entry and exit prices. ### [Prospero.ai](https://tradingagentstack.com/tool/prospero-ai/) *D*Backtest or hypothetical Paraphrase27% better returns than S&P 500 with a 54% win rate (about 5,000 picks since 2023) Where[Official home page (trading newsletter)](https://www.prospero.ai/)SeenOct 28, 2026Claim datedNot shown (since 2023)Based on (their words)Hypothetical (same footer)What anyone can checkNo complete pick log found.Their disclaimerSame footer as above.What would verify itA full pick log with timestamps. ### [Prospero.ai](https://tradingagentstack.com/tool/prospero-ai/) *D*Backtest or hypothetical “12.9x outperformance of our investment picks to the S&P 500” Where[Official home page](https://www.prospero.ai/)SeenOct 28, 2026Claim datedNot shown; no periodBased on (their words)Not stated at the claim; the site footer says results are not actual trading resultsWhat anyone can checkNone found.Their disclaimerFooter disclaimer only.What would verify itThe period, the method and the list of picks. ### Stock Prediction AI US Market (Circuit Europe OU) *D*Backtest or hypothetical “achieving over 90% accuracy in benchmark testing scenarios” Where[Apple App Store listing](https://apps.apple.com/us/app/ai-%EC%A3%BC%EC%8B%9D-%EC%98%88%EC%B8%A1/id6740812212?l=en-US)SeenOct 29, 2026Claim datedVersion note dated Sep 15 (year not shown)Based on (their words)Backtest (benchmark testing; method not given)What anyone can checkNone.What would verify itLive predictions with timestamps, scored against outcomes. ### [Trade Ideas (Holly AI)](https://tradingagentstack.com/tool/trade-ideas-holly-ai/) *D*Backtest or hypothetical “Simulated portfolio returns a scorching 54% beating S&P's performance of 3.0%” Where[Official blog post](https://www.trade-ideas.com/2018/08/03/our-a-i-managed-simulated-portfolio-smashes-market-benchmarks-for-1st-half-of-2018/)SeenOct 28, 2026Claim dated2018-08-03 (first half of 2018)Based on (their words)Simulated ("historical daily paper trades", no slippage)What anyone can checkNo trade list.Their disclaimerPast performance of the simulated portfolio does not guarantee future results.What would verify itA broker statement for a real account trading the signals. ### [Bitsgap](https://tradingagentstack.com/tool/bitsgap/) *C*No independent record Paraphrase11% average 30-day profit for the grid bot (and $203M of bot profit in a year) Where[Official home page statistics](https://bitsgap.com/)SeenOct 29, 2026Claim datedNot shownBased on (their words)Not stated (no method, date range, or gross vs net)What anyone can checkNone found.What would verify itA method that counts stopped and losing bots and fees, checked against exchange data. ### [Cryptohopper](https://tradingagentstack.com/tool/cryptohopper/) *C*No independent record “Automatic Trading Bots outperform humans” Where[Official home page](https://www.cryptohopper.com/)SeenOct 28, 2026Claim datedNot shownBased on (their words)Not stated; a screenshot on the page shows +300.57%What anyone can checkNone found.Their disclaimerPast performance is not indicative of future results; screenshots are illustrative and may be exaggerated.What would verify itA study of matched bot vs manual accounts. ### [Cryptohopper Marketplace (listing by a third-party seller)](https://tradingagentstack.com/tool/cryptohopper/) *C*No independent record “Profit 101.62%” Where[Official marketplace listing (seller Gainscrypt, APEX-R)](https://www.cryptohopper.com/marketplace)SeenOct 28, 2026Claim datedNot shown; no periodBased on (their words)Not statedWhat anyone can checkNone visible.Their disclaimerProfits shown on the marketplace are not indicative of future results.What would verify itThe period, live vs backtest, and an exchange-verified record. ### [Danelfin](https://tradingagentstack.com/tool/danelfin/) *C*No independent record “outperformed the market by +14.69% on average after 3 months (annualized alpha)” Where[Official website, home page](https://danelfin.com)SeenOct 28, 2026Claim datedNot shown (record said to start in 2017)Based on (their words)Not statedWhat anyone can checkNo independent record found.Their disclaimerNearby text: it is impossible to predict perfectly which stocks will beat the market.What would verify itA timestamped archive of every published score, with the return method. ### [Danelfin](https://tradingagentstack.com/tool/danelfin/) *C*No independent record ParaphraseBest Stocks strategy generated +191% vs +118% for the S&P 500 Where[Official website, home page](https://danelfin.com)SeenOct 28, 2026Claim datedPeriod Jan 3, 2017 – Aug 15, 2023 (not updated since)Based on (their words)Not statedWhat anyone can checkNo independent record found.What would verify itHoldings and rebalance dates as published at the time. ### [Kavout (Kai Score)](https://tradingagentstack.com/tool/kavout/) *C*No independent record ParaphraseTop Picks compound annual growth of 21.9% vs 13.3% for the S&P 500 since January 2012 Where[Official website article](https://www.kavout.com/market-lens/leveraging-alternative-sources-for-stock-picking-kai-score)SeenOct 29, 2026Claim datedNo end date shownBased on (their words)Not statedWhat anyone can checkNone found.Their disclaimerSite-wide: past performance does not guarantee future results.What would verify itWhen each score was published vs the portfolio dates, and the holdings history. ### [LevelFields](https://tradingagentstack.com/tool/levelfields/) *C*No independent record “~80% winners since 2022.” Where[Official home page description (search and social preview)](https://www.levelfields.ai/)SeenOct 29, 2026Claim datedNot shownBased on (their words)Not stated (the same text mentions 5-year backtests)What anyone can checkOnly the product-hosted log above. [Link](https://tradelog-levelfields.grok.me/)What would verify itA definition of a winner and the full alert list since 2022. ### [Tickeron](https://tradingagentstack.com/tool/tickeron/) *C*No independent record “have achieved annualized returns as high as +171% in live trading” Where[Official website article and press release](https://tickeron.com/trading-investing-101/ai-trading-agents-deliver-up-to-171-annualized-returns-in-live-trading/)SeenOct 28, 2026Claim dated2025-09-26Based on (their words)Live trading, as stated; account and broker not named; periods of 45–115 days, annualizedWhat anyone can checkNo broker-verified record found.Their disclaimerA link to a disclaimers page.What would verify itBroker-verified records per agent for 12+ months, not annualized. ### [Tickeron](https://tradingagentstack.com/tool/tickeron/) *C*No independent record “posted the highest reported annualized return at 212.87%” Where[Official website article and press release](https://tickeron.com/trading-investing-101/tickeron-launches-new-ai-trading-robots-reaching-180-return-for-retail-investors-lite/)SeenOct 28, 2026Claim dated2026-08-16Based on (their words)Not stated ("reported"; no period given)What anyone can checkNo independent record found.Their disclaimerA link to a disclaimers page.What would verify itThe period, the account type and a broker-verified record. ### [Tickeron (Trend Prediction Engine)](https://tradingagentstack.com/tool/tickeron/) *C*No independent record “forecasting stock trends with up to 70% historical accuracy” Where[Official website article and press release](https://tickeron.com/trading-investing-101/ai-trend-prediction-engine-from-tickeron-delivers-70-accuracy-for-shortterm-trades/)SeenOct 28, 2026Claim dated2025-10-21Based on (their words)Historical, method not statedWhat anyone can checkNo out-of-sample record found.Their disclaimerA link to a disclaimers page.What would verify itA definition of accuracy and forecasts published with timestamps before the outcome. ### [Trade Ideas (Holly AI)](https://tradingagentstack.com/tool/trade-ideas-holly-ai/) *C*No independent record ParaphraseThe total performance of HOLLY AI for September was -1.5% (vs SPY −5%) Where[Article by a Trade Ideas co-founder on Nasdaq.com](https://www.nasdaq.com/articles/trade-ideas-monthly-summary%3A-september-2020-2020-10-08)SeenOct 28, 2026Claim dated2020-10-08Based on (their words)Per-share profit on each signal, no stops (the article's own definition)What anyone can checkNo signal list.What would verify itAll signals with timestamps and the method used to add them up. ### Make a claim? Make it checkable Publish the trading address on your official site or docs, or connect a read-only record a venue verifies, and the number goes on the [verified leaderboard](https://tradingagentstack.com/ai-trading-leaderboard/) with our own math. [Contact us](https://tradingagentstack.com/contact/) to get a claim re-graded. ### How we collect claims Return, win-rate, accuracy and beat-the-market claims from a product's own website, app-store listing, official posts or press releases, with the link and the date we saw it. Short quotes are exact; longer claims are paraphrased and marked. Spot an error or an update? [Tell us](https://tradingagentstack.com/submit/). Not financial advice. Grades describe the evidence, not the product. Past results don't predict future results. ## For agents - [When to use this site, and the page index](https://tradingagentstack.com/llms.txt) - [Sitemap](https://tradingagentstack.com/sitemap.xml) - [Contact](https://tradingagentstack.com/contact/) - [MCP server](https://tradingagentstack.com/mcp/) - [MCP discovery](https://tradingagentstack.com/.well-known/mcp)