A glossary of research and quantitative terms as used at Maple Research — from research infrastructure, simulation, and reproducibility through the standard toolkit of strategy validation: out-of-sample testing, walk-forward testing, Monte Carlo simulation, drawdown, and more.
Research infrastructure
The underlying systems — data access, modelling, testing, simulation, and validation — that let a financial question move to evidence. Maple's core thesis is that this infrastructure should be unified rather than assembled from a dozen disconnected tools.
Hypothesis-driven research
An approach that starts from an explicit question or claim to be tested, rather than starting from whichever software or dataset happens to be at hand — the organizing idea behind Maple's research workspace.
Simulation
Testing what could happen under conditions history didn't happen to provide — alternative scenarios, assumptions, or regimes — as distinct from backtesting, which replays what actually happened.
Scenario analysis
Evaluating how a model or strategy would behave under a specific, constructed set of conditions (a rate shock, a volatility spike) rather than only across the historical record.
Reproducibility
The property of a result that lets it be regenerated exactly, given the same inputs and assumptions — a foundation of Maple's deterministic engine and a prerequisite for a result being genuinely inspectable.
Backtest
A simulation of how a strategy would have performed if applied to historical market data. A deterministic backtest produces the same result every time given the same strategy and data.
Overfitting
When an idea is tuned so closely to a specific historical dataset that it captures random noise rather than a real, repeatable pattern, producing results that look strong but fail to hold up on new data.
Out-of-sample testing
Evaluating an idea on data it was not built or tuned on, to see whether its performance holds up outside the exact conditions it was designed around.
In-sample data
The portion of historical data used to build or tune a strategy, as opposed to out-of-sample data used to independently check it afterward.
Walk-forward testing
A validation method that re-tests an idea across a rolling series of time windows rather than one fixed historical period, checking whether its edge persists as conditions change.
Monte Carlo simulation
A technique that reorders or resamples a sequence of trades or events many times to see how much a result depends on the specific order things happened to occur in, rather than the underlying rule.
Drawdown
The decline in an account's value from a previous peak, usually expressed as a percentage. Maximum drawdown is one of the most important risk measures in evaluating a strategy.
Maximum drawdown
The largest peak-to-trough decline over a given test period — a key measure of how much loss a strategy could realistically require someone to tolerate.
Sharpe ratio
A measure of risk-adjusted return: the average return of a strategy relative to its volatility. A higher Sharpe ratio suggests more return was earned per unit of risk taken.
Sortino ratio
A variation on the Sharpe ratio that only penalizes downside volatility (losses), rather than all volatility, on the reasoning that upside volatility isn't the risk investors actually care about.
Win rate
The percentage of a strategy's trades that were profitable. A high win rate does not by itself indicate a good strategy — it must be considered alongside average win/loss size.
Risk-reward ratio
The ratio between a trade's potential loss and its potential gain, often used to size positions and evaluate whether a strategy's trade structure makes statistical sense.
Sample size / sample adequacy
The number of trades or data points a result is based on. A result with too few events may be statistically unreliable, even if it looks strong on its face.
Statistical significance
A measure of how likely it is that a result reflects a real effect rather than random chance, given the amount of data available.
Parameter sensitivity / parameter robustness
How much a result changes when its settings are adjusted slightly. An idea that only works with one very specific parameter combination is considered fragile, not robust.
Curve fitting
Another term for overfitting: adjusting rules or parameters until they match historical data as closely as possible, at the cost of real-world reliability.
Market regime
A distinct set of market conditions — a strong trend, a range-bound period, high volatility — that can significantly affect how a strategy or model performs.
Regime dependence / regime analysis
When an apparent edge is concentrated in one type of market condition and largely absent in others, meaning its historical average may not reflect future conditions. Regime analysis is the check that surfaces this.
Execution assumptions
The realistic constraints (costs, slippage, fill timing) applied when replaying a strategy against history, without which a result can look better than it would have been in practice.
Slippage
The difference between a trade's expected execution price and its actual execution price, often caused by market movement between the decision to trade and the order being filled.
Transaction costs
Fees, spreads, and other costs incurred when executing trades, which reduce real-world returns relative to a theoretical result.
Equity curve
A chart showing how an account's value changes over time — one of the most direct ways to see a strategy's overall behaviour and drawdown pattern.
Alpha
The portion of a strategy's return that cannot be explained by general market movement — a genuine, skill- or edge-based excess return.
Beta
A measure of how sensitive a strategy or asset's returns are to overall market movements. A beta of 1 means it tends to move in line with the market.
Volatility
A statistical measure of how much an asset's price fluctuates over time. Higher volatility generally implies higher risk and larger potential price swings.
Mean reversion
A trading approach based on the idea that prices tend to return toward an average level after moving away from it, used as the basis for many contrarian strategy designs.
Trend following
A trading approach that seeks to profit from sustained directional price movement, typically entering positions in the direction of an established trend.
Moving average
The average price of an asset over a set number of past periods, commonly used to smooth out short-term price noise and identify trend direction.
Moving average crossover
A common signal generated when a shorter-term moving average crosses above or below a longer-term moving average, often used as an entry or exit trigger.
Momentum
The tendency of an asset that has been moving in one direction to continue moving in that direction for some period, forming the basis of momentum-based strategies.
Stop-loss
A predefined price level at which a losing position is automatically closed, used to limit downside risk on an individual trade.
Take-profit
A predefined price level at which a winning position is closed to lock in gains, used alongside a stop-loss to define a trade's risk-reward structure.
Position sizing
The process of deciding how much capital to allocate to a given trade, a major factor in a strategy's overall risk profile independent of its entry/exit rules.
Historical data
Recorded past market prices and related information used as the input for testing a hypothesis against history.
Timeframe
The interval each data point in a chart or test represents — for example, 15-minute, 1-hour, or daily candles — which can significantly affect how a strategy behaves.
Candlestick
A chart element representing price movement over a given period, showing the open, high, low, and close prices for that interval.
Strategy
A defined, testable set of rules for entering and exiting positions on a given asset and timeframe — one specific form a financial hypothesis can take.
Pine Script
TradingView's scripting language for writing custom indicators and strategies, and one of the formats Maple's hypothesis-input stage is designed to read.
Confidence read
A plain-language summary of how much statistical weight a result can reasonably bear, based on how it performs under Maple's validation concepts — not a prediction of future returns.
Validator / validation concept
One of the independent checks (out-of-sample testing, parameter robustness, and others) used to assess how much confidence a result deserves.
Quantitative research
The practice of using data, statistics, and systematic testing — rather than intuition alone — to evaluate financial ideas.
Algorithmic trading
Trading based on predefined, rules-based logic rather than discretionary human decision-making at the moment of each trade.
Robustness
How consistently an idea performs across different time periods, parameter settings, and market conditions, as opposed to performing well only in one narrow, specific case.
Edge
A real, repeatable statistical advantage over random chance — the thing quantitative research is ultimately trying to identify or rule out.
Survivorship bias
A distortion that occurs when a dataset only includes assets or entities that still exist today, leaving out those that failed or were delisted, which can make historical results look better than they really were.
Look-ahead bias
A testing error where a model is accidentally given access to information that would not have been available at the time a real decision was made, inflating its apparent performance.
Data snooping
Testing many variations against the same dataset until one appears to work, without accounting for the fact that some will look good by chance alone — a major driver of overfitting.
Correlation
A statistical measure of how closely two assets or variables move in relation to one another, often used to understand diversification or shared risk.
Diversification
Spreading exposure across multiple assets or strategies so that no single position or approach dominates overall risk.
Risk management
The set of practices used to control how much a trader or strategy stands to lose, including position sizing, stop-losses, and diversification.
Deterministic (in computing)
Producing the exact same output every time given the exact same input — a property Maple's core research engine is specifically designed to have, unlike probabilistic AI-generated content.
Bull market
An extended period of generally rising asset prices, one of the market regimes a strategy's performance can depend heavily on.
Bear market
An extended period of generally falling asset prices, another distinct market regime that can significantly affect strategy performance.
Liquidity
How easily an asset can be bought or sold without significantly moving its price — low liquidity can make real-world execution meaningfully worse than a backtest suggests.