Multiple AI models argue, critique, and verify every investment decision through an adversarial pipeline. The same rigor hedge funds use — at a fraction of the cost.
*Results are from a 5-year walk-forward backtest using real market data. Past performance does not guarantee future results. All investing involves risk.
5.3-year walk-forward backtest with real market data, modeled transaction costs, and no lookahead bias. The same test methodology used by quantitative hedge funds.
Every investment decision passes through four AI stages using models from different providers. No single AI has unchecked authority over your capital.
Analyzes regime, positions, news, and ranked candidates to generate an investment thesis
Claude Opus →A different AI attacks the proposal — finding weaknesses, blind spots, and hidden risks
Gemini Pro →Reconciles the thesis with the critique to produce a refined, risk-adjusted recommendation
Claude Opus →Independent safety gate reviews against hard constraints. Can block any recommendation.
Gemini ProDeterministic 5-factor scoring constrains which stocks the AI can consider. Same data always produces the same rank — no model oscillation.
Separate AI classifier detects bankruptcy filings, fraud charges, and catastrophic events. Auto-liquidates threatened positions in seconds.
Positions are volatility-normalized using ATR, then scaled by conviction with bidirectional doctrine bands. High-conviction gets more capital. Mathematically.
Four sub-classifiers monitor valuation, sentiment, credit, and monetary policy. The composite state adjusts every parameter in real-time.
Continuous monitoring: trailing stops, momentum exhaustion, time decay. Prioritized signals with severity grading so nothing gets missed.
Two-layer automated scanning: daily market movers plus weekly S&P 500 momentum analysis. Uses the same scoring function for perfect consistency.
Unlike black-box funds, every metric is visible. We believe showing risk openly is the strongest trust signal we can offer.
Real-time portfolio intelligence, AI proposals with full audit trails, regime monitoring, and exit signal tracking.
Hedge funds charge 2% of assets plus 20% of profits. We deliver the same multi-model AI rigor at a fraction of the cost.
Performance fees apply only to net new profits above your account's high-water mark. If we don't grow your portfolio, you don't pay performance fees. Period.
The system recommends. You decide. Nothing executes without explicit approval — except catastrophic event protection.
Your brokerage account is in your name at Alpaca (FINRA member, SIPC insured up to $500K). We never hold your funds.
Rules enforced at the AI prompt layer, mechanical post-parse validator, and code assertions. No single failure can cause a violation.
Every AI stage, every model response, every validator check logged and visible. Complete transparency into every decision.
These are walk-forward backtested results from January 2021 through April 2026 using real market data with modeled transaction costs and slippage. Walk-forward testing trains the model on past data, then tests on future unseen data sequentially — the gold standard for avoiding overfitting. Live trading began in early 2026. We will publish live results separately as they accumulate meaningful track record.
All investing involves risk, including loss of principal. Our maximum drawdown during testing was -21.7% (October 2022, when the entire market bottomed). While our system significantly outperformed the market during this period, a 20%+ drawdown means a $100K portfolio could temporarily drop to ~$78K before recovering. We show these numbers openly because informed investors make better decisions.
Robo-advisors typically allocate across index funds (passive). Truviant actively manages individual stock positions using multiple AI models that argue with each other before making decisions. Our adversarial pipeline — where one AI proposes and another attacks the proposal — is patent-pending technology that doesn't exist in any robo-advisor.
Yes. No investment system eliminates risk. Our worst monthly decline was -8.3%. However, our system recovered from its maximum drawdown in approximately 4.2 months. We use mechanical stop-losses, position size limits, and catastrophe detection to limit downside — but cannot eliminate it.
You can pause automated trading from your dashboard at any time. You can also revoke API access from your Alpaca account, instantly disconnecting Truviant. Cancel your subscription with no penalty. Your brokerage account and positions remain yours regardless.
You shouldn't trust an AI — that's our whole point. We use multiple AIs from different providers that must agree before any action is taken. One proposes, another attacks, a third synthesizes, and a fourth can veto. Plus 18 mechanical validators that no AI can override. The system is designed around the assumption that any single AI can be wrong.
Start with paper trading — watch the AI perform with virtual money before committing real capital.
Start Investing →No credit card required for paper trading trial