Risk first: what this course cannot promise

This course teaches market mechanics, research, risk control, and operational discipline. It does not know your financial situation, provide a signal service, or make any setup profitable. A good decision can lose and a reckless decision can win once.

Trading is a probability process. Expectancy is an estimate built from uncertain future outcomes, changing costs, and a sample that may stop representing the market. Treat every edge as a hypothesis under continuing review.

At 10× leverage, a 1% underlying move changes position equity by roughly 10% before fees and funding. Liquidation can occur before the simple `1 ÷ leverage` estimate because venues apply maintenance margin, mark prices, fees, and contract-specific rules.

Leverage does not create an edge. It scales an exposure you already chose and reduces your tolerance for estimation error, volatility, latency, and gaps. Position risk must be set from invalidation and account risk, not from the maximum leverage a venue offers.

An exchange balance is a claim on a custodian, not the same thing as an on-chain asset controlled by your key. A venue can become insolvent, halt withdrawals, suffer an exploit, change margin rules, reject orders, or become unreachable during volatility.

Reduce single points of failure: understand the legal entity and product, limit idle balances, separate trading collateral from long-term custody, test withdrawals, use strong account security, export records, and define what you will do during an outage.

Stablecoins can depend on reserves, custodians, redemption access, collateral liquidations, market incentives, or an algorithm. A quoted price near one dollar does not remove depeg, issuer, freeze, liquidity, chain, or redemption risk.

A DeFi deposit adds contract, oracle, governance, upgrade, bridge, and composability dependencies. Yield is compensation from somewhere. Before depositing, identify the payer, cash flow, collateral, withdrawal path, privileged controls, and failure sequence.

A backtest can use prices you could not have received, while paper trading can assume fills without queue position, latency, rejects, partial execution, or emotional pressure. Both are research stages, not certificates of profitability.

Move through gates: clean historical test, untouched out-of-sample evaluation, time-based forward test, then minimal live risk with predefined stop conditions. Compare expected and realized spread, slippage, fees, funding, rejects, and rule adherence.

Freeze the scorecard during each stage. If results cause a rule change, record a new version and begin a new validation sample rather than rewriting the history of the old one.