Learn · Glossary
The vocabulary, in plain language.
Finance hides behind its vocabulary. These definitions are written to be used, not to impress — each one tells you what the term means, why it matters, and where beginners get burned by it. Terms are grouped by what you are trying to do.
Measuring performance
Return
How much money a strategy made or lost, usually expressed as a percentage of the capital it used. A return means nothing without its denominator, its time period, and its risk — '+$500' is not information; '+2% on a $25,000 account in one month, with a 10% worst dip' is.
Sharpe ratio
Return divided by the volatility it took to earn it — reward per unit of risk. A strategy that makes 10% smoothly beats one that makes 12% with stomach-dropping swings. Rule of thumb: below 0.5 is noise territory, 1.0 is respectable, and anything above ~2 in a backtest deserves suspicion before celebration.
Deflated Sharpe ratio (DSR)
A Sharpe ratio adjusted for how many things you tried before finding it. If you test 100 strategy variations, the best one will look great by pure luck. DSR charges you for every variation tried, shrinking the score toward what remains after luck is paid for. It is the difference between 'this worked' and 'this worked after accounting for how hard I searched'.
Drawdown
The drop from a strategy's highest point to its lowest point after — the pain measurement. Maximum drawdown is the worst such drop ever recorded. A strategy returning 20% annually with a 40% max drawdown will bankrupt most people psychologically before it bankrupts them financially.
Win rate
The percentage of trades that make money. Widely worshipped, mostly meaningless alone: a strategy can win 96% of the time and still lose everything if the 4% of losers are large enough. Win rate only means something next to the average size of wins versus losses.
Expectancy
The average amount you expect to make or lose per trade, counting both winners and losers: (win rate × average win) − (loss rate × average loss). If it is not positive after costs, nothing else about the strategy matters.
Volatility
How much a price or an account swings around, usually measured as standard deviation of returns. Not the same as risk, but the most convenient proxy for it. Volatility is also a thing you can effectively buy and sell through options — which is why the word appears twice in every options conversation.
Benchmark
The boring alternative your strategy must beat to justify existing — usually 'just buy the index and hold it'. A strategy that returns 8% a year while the S&P returns 10% did not make you 8%; it cost you 2% plus your time.
Testing honestly
Backtest
Running a strategy's rules against historical data to see how it would have done. The most abused tool in trading: done carelessly it is a machine for generating false confidence. A backtest is an indictment, never a verdict — the forward test is the jury.
In-sample / out-of-sample
In-sample is the data you used to build and tune the strategy; out-of-sample is data it has never seen. Performance on in-sample data is nearly worthless — of course it fits; you fit it. Only out-of-sample results count as evidence.
Overfitting
Tuning a strategy so precisely to past data that it has memorized history's noise instead of learning a real pattern. The overfit strategy backtests beautifully and dies immediately in live trading. It is the default outcome of strategy research, not the exception.
Walk-forward validation
Repeatedly training on one window of history and testing on the next, sliding forward through time — simulating what you would actually have experienced running the strategy live. Far harder to fool than a single backtest, which is exactly the point.
Cross-validation (CPCV)
Testing a strategy across many different train/test splits of the data rather than one, so a single lucky period cannot carry the result. Combinatorial purged cross-validation (CPCV) is the finance-grade version, which also removes overlapping data that would let information leak from training into testing.
Probability of backtest overfitting (PBO)
A direct estimate of the chance that your 'best' strategy configuration is best only in the backtest — that its ranking would collapse on new data. If the top-ranked configurations keep swapping places across data splits, the ranking is noise and PBO says so.
Lookahead bias
Accidentally letting the strategy use information that would not have existed at the moment of the trade — closing prices before the close, earnings numbers before the announcement, revised data instead of what was originally published. One of the two great backtest killers.
Survivorship bias
Testing only on assets that still exist today, which quietly excludes everything that crashed, delisted, or went bankrupt — the exact events your strategy needed to be tested against. The other great backtest killer.
Data snooping / multiple testing
The statistical sin of testing many ideas on the same data and reporting the winner as if it were the only thing you tried. With enough attempts, something always 'works' by chance. Honest research counts every attempt and discounts the winner accordingly.
Pre-registration
Writing down your hypothesis, your test, and what result would falsify it before running anything. Borrowed from clinical science. It costs five minutes and removes the temptation to move the goalposts after seeing the data — the cheapest overfitting defense that exists.
Paper trading
Running a strategy with fake money against real live prices. The proving ground between backtest and live: it catches execution problems, data problems, and logic bugs a backtest cannot, while a mistake costs nothing but pride.
Regime
The market's prevailing character — trending or choppy, calm or violent. Strategies are usually creatures of a regime: a trend-follower starves in a sideways market. The hard part is that regimes are obvious in hindsight and nearly invisible in real time, so any claim to detect them deserves its own validation.
Risk and sizing
Position sizing
Deciding how much to risk on each trade — the decision that determines survival. Most blown-up accounts had entry rules that were fine and sizing that was fatal. Serious sizing keys off a percentage of current account value, so it shrinks automatically when you are losing.
Risk of ruin
The probability that a strategy, sized the way you are sizing it, eventually digs a hole too deep to climb out of. Any positive-expectancy strategy can still ruin you if sized too aggressively — bet enough of your account per trade and a normal losing streak becomes fatal.
Stop loss
A standing order to exit a position once it loses a set amount — the seatbelt. The catch beginners miss: a stop order becomes a market order when triggered, so in a fast market you can fill well past your stop price. A stop limits loss on most days, not the worst day.
Leverage
Trading with more exposure than the cash you hold, via margin or derivatives. It multiplies gains and losses identically, but its real danger is subtler: leverage shortens the losing streak you can survive. Forex and futures are leveraged by construction.
Diversification
Spreading risk across positions that do not fail together. The key word is 'together': ten tech stocks are one bet wearing ten costumes. Real diversification is measured by correlation, not by counting tickers — and correlations have a habit of converging toward 1 exactly when markets panic.
Kelly criterion
A formula for the bet size that maximizes long-run growth given your edge and odds. Mathematically elegant, practically violent — full Kelly produces drawdowns most humans cannot sit through, and it assumes you know your edge precisely, which you never do. Practitioners who use it at all use a fraction of it.
Execution and market structure
Slippage
The gap between the price your backtest assumed and the price the market actually gave you. Comes from spreads, market impact, and delay. Small per trade, lethal in aggregate — slippage and costs are the most common reason a 'profitable' backtest loses real money.
Bid-ask spread
The gap between the highest price buyers will pay (bid) and the lowest price sellers will accept (ask). Crossing it is the toll every trade pays before it can be profitable. For a fast-trading strategy, the spread is usually the whole ballgame.
Liquidity
How much you can trade without moving the price against yourself. Deep liquidity means you get filled near the quoted price; thin liquidity means your own order is the market event. Every strategy has a capacity — a size beyond which its own trading destroys its edge.
Market order / limit order
A market order says 'fill me now at whatever the price is' — certain execution, uncertain price. A limit order says 'fill me at this price or better' — certain price, uncertain execution. Every execution decision is a trade-off between those two uncertainties.
Fill
The actual execution of your order: the real price, size, and time. Backtests imagine fills; brokers report them. A live system must treat every order as unconfirmed until the fill (or rejection) is verified — assuming an order 'went through' is how accounts quietly diverge from what their operators believe.
Broker API
The programmatic interface a broker exposes so software can place orders, check positions, and pull account state — the doorway between your code and the market. The unglamorous truth of systematic trading is that most of the engineering lives here, in authentication, error handling, and verifying what actually happened.
Instruments
Derivative
A contract whose value derives from something else — a stock, an index, a currency, a barrel of oil. Options and futures are the common ones. Derivatives let you shape exposure precisely, and they let you lose money with equal precision.
Option
A contract giving the right, but not the obligation, to buy (call) or sell (put) an asset at a set price by a set date. The buyer pays a premium for that right; the seller collects it and takes on the obligation. Options are three-dimensional — price, time, and volatility all move them — which is why they punish one-dimensional thinking.
Futures contract
An agreement to buy or sell an asset at a set price on a set future date, standardized and exchange-traded. Leveraged by construction and marked to market daily, meaning gains and losses settle in cash every day — there is no 'waiting for it to come back'.
Volatility risk premium (VRP)
The persistent tendency of options to price in more future volatility than actually arrives — insurance is systematically sold for more than it pays out, because sellers must be paid for carrying tail risk. Harvesting it means collecting steady premiums and occasionally paying out catastrophically; whether that trade survives honest validation depends entirely on the sizing and the tail.
Spread (options)
Combining options bought and sold at once so the position's maximum loss is fixed and known at entry — defined risk. A put spread, for example, sells one put and buys a cheaper one below it as a hard floor. You surrender some premium to make your worst-case a number instead of a hope.
Carry
The return earned simply for holding a position, independent of price movement — interest-rate differentials in currencies, dividends in stocks, the shape of the futures curve in commodities. Carry strategies collect steadily and then occasionally give it back all at once, which is exactly the kind of return profile that fools a short backtest.
Missing a term you ran into? The curriculum builds all of these from first principles, in order, and the research graveyard shows most of them doing real work on real verdicts.
About this curriculum: everything in the Learn section is for educational and informational purposes only. It does not constitute personalized investment advice, a recommendation to buy or sell any security or derivative, or an offer of advisory services. Crash Money Analytics does not know your individual financial situation, and nothing here should be treated as tailored to it. Trading involves substantial risk of loss. Consult a licensed financial advisor before making investment decisions. See our Risk Disclosures.