Risk Disclosure

Arizmic is quantitative research software. Its website, Documentation, Education content, models, simulations, and AI-assisted features are provided for research and informational purposes. They are not investment advice, a recommendation, a brokerage service, or a promise of performance.

Past and hypothetical performance

Past performance is not indicative of future results. Backtests, walk-forward studies, paper trading, and other hypothetical results do not represent actual trading and can differ materially from live outcomes. Hypothetical work benefits from hindsight and cannot fully reproduce the financial and psychological effects of risking capital.

A favorable result may depend on a particular period, market regime, universe, cost model, parameter choice, or data revision. No metric or validation procedure proves that a strategy will remain profitable.

Data and research-design risk

Results can be distorted by missing or incorrect observations, timestamp errors, corporate-action handling, symbol changes, stale quotes, bad depth reconstruction, survivorship bias, lookahead bias, universe selection, or unrepresentative coverage. A provider's right to display data does not automatically grant every research or redistribution right.

Reproducibility also depends on the exact dataset version, preprocessing, assumptions, software version, seeds, parameters, and evaluation path. A result that cannot be reproduced should not be treated as reliable merely because it looks plausible.

Models, signals, optimization, and AI

Every model and signal simplifies the market. It may omit material variables, encode unstable relationships, or behave differently after a regime change. Parameter searches and optimizers can select noise, especially when many alternatives are tried and only favorable results are retained.

AI output can be incorrect, incomplete, insecure, or fabricated. Generated code, explanations, research ideas, and diagnoses require human review and independent validation. Native AI integration does not transfer responsibility for a research or deployment decision to Arizmic or the model provider.

Simulation fidelity

Vector, bar, trade, quote, and supported depth simulations answer different questions and use different assumptions. Faster evaluation can support broad exploration without reproducing the ordering, queue, liquidity, or timing behavior available to a more detailed event simulation.

Access to the complete fidelity range supported by Arizmic does not mean every dataset contains trades, quotes, or market depth, or that every strategy can be evaluated at every fidelity. The selected data and evaluator determine what can actually be modeled.

Execution and market risk

Modeled fills may differ materially from executable fills. Queue position, spread, liquidity, latency, slippage, fees, partial fills, rejects, market impact, halts, auctions, order types, and broker rules can change an outcome. A simulation cannot guarantee that an order would have been accepted or filled at the modeled time or price.

Markets can move rapidly and losses can exceed historical estimates or expectations. Leverage, derivatives, short positions, concentration, and gaps can increase losses and, in some circumstances, create obligations beyond the initial capital allocated.

Provider, broker, and operational risk

Data, AI, broker, network, operating-system, and hardware services can be delayed, unavailable, inconsistent, compromised, or changed without notice. Credentials can expire or be revoked. Broker and local state can diverge, and reconciliation may require operator intervention.

A supported connection is not a guarantee that every account, instrument, order type, market, or jurisdiction is available. Users must review the exact provider support matrix, entitlements, account permissions, and release status.

Supervised live deployment

Any live deployment must remain supervised. The user is responsible for confirming the environment, account, capital, strategy version, permissions, risk limits, open orders, positions, fills, and reconciliation state. A readiness check or risk control can reduce a class of error but cannot prevent every loss, outage, or misuse.

Research results should be treated as inputs to a decision, not authorization to deploy capital. Users should be prepared to stop a workflow, cancel orders, reconcile state, and manage positions outside Arizmic when conditions require it.

Legal, tax, and jurisdictional responsibility

The user is responsible for determining whether a data source, model, strategy, instrument, broker connection, or trading activity is lawful and suitable in the user's jurisdiction and circumstances. Laws, taxes, exchange rules, market-data terms, professional obligations, and broker requirements vary.

Arizmic does not provide legal, tax, accounting, fiduciary, or regulatory advice. Consult appropriately qualified professionals before acting where those matters are material.