Glossary
Algorithmic trading glossary
Each term opens with a one-sentence definition, then the detail, and a formula or example where one helps. Every LFP figure is read from the same backtest data the console runs on.
- Backtest
A backtest runs a strategy over historical data to see what it would have done if it had been trading at the time.
A backtest is evidence, not proof. Its worth depends entirely on data quality, whether costs were charged realistically, and how much the settings were chosen on that same data.
Three questions any backtest should answer: how costs were modelled, how much of the period was seen during design, and whether the result survives real tick data.
At LFP: Every performance figure on this site is a backtest and is labelled as one. A live forward test has been running since 19 September 2026.
See also: Out-of-sample testing, Overfitting
- Balance and equity
Balance is the result of closed trades; equity is the balance plus the running profit or loss of open positions.
With no position open the two are identical. The moment a trade opens, equity moves tick by tick while balance stands still.
Any claim about risk has to say which of the two it is based on. A system that keeps losing positions open can show a very smooth balance curve and a disastrous equity curve at the same time.
See also: Drawdown
- CAGR
CAGR is the annual growth rate which, repeated every year, would take an account from its starting value to its final one.
It hides the shape of the journey. Two systems with the same CAGR can feel completely different to live with, which is why CAGR should never be read without a drawdown figure beside it.
Because growth compounds, the total figure over a long period looks far larger than simply adding the years together.
CAGR = (ending value ÷ starting value)^(1 ÷ years) − 1
- Drawdown
Drawdown is the largest distance an account falls from its own highest point, stated as a percentage.
Maximum drawdown is the worst such fall across the whole period, and it is the most common measure of a trading system’s risk. Unlike a return figure, it describes what the trader actually lives through.
Always ask which value it was measured on: balance or equity. The equity figure counts the loss on open positions too, and is always the stricter of the two.
Drawdown = (peak − subsequent trough) ÷ peak
Example: An account falling from $12,000 to $9,600 has a 20% drawdown.
At LFP: In the LFP backtest on the default Brave style, the real maximum drawdown was 20.6%.
See also: Underwater period, Balance and equity
- Expert Advisor (EA)
An Expert Advisor is a program that runs inside MetaTrader and places trades automatically according to predefined rules.
It runs on the trader’s own terminal and needs MetaTrader running and connected to work, which is why most users host it on a VPS.
An Expert Advisor has no withdrawal access to your capital. All it does is send orders through the same terminal you signed into yourself.
See also: Backtest
- Lot
A lot is the unit of trade size; one standard lot in currency pairs is 100,000 units of the base currency.
The smallest tradable size at most brokers is 0.01 lots. A system that sizes by a percentage of the account will hit that floor on a small account, and some trades simply will not be placed.
Rounding size up to clear the floor means accepting more risk than intended. The honest alternative is to skip the trade and state a minimum capital.
At LFP: LFP never rounds size up. The suggested minimum for a standard-contract account is $2,000.
See also: Pip
- Martingale and grid
Martingale means increasing size after a loss so one win recovers it; grid means adding positions in steps against the move.
Both produce very smooth equity curves with almost no losing months — until one sustained trend takes the account out in a single stretch.
A robot with a 90% win rate and no losing months is usually hiding one of the two. The tell is open positions without stop losses, and sizes that grow after every loss.
At LFP: LFP uses no martingale, no grid and no averaging down, and every trade carries a stop loss held at the broker’s server.
- Out-of-sample testing
Out-of-sample testing means measuring a strategy on a stretch of data that was never seen while its settings were being chosen.
Without it, skill cannot be separated from luck: any strategy can be tuned to look excellent on the history it was fitted to.
Its validity is single-use. The moment you look at the result and change the settings, that stretch is no longer out of sample.
At LFP: LFP’s settings were chosen on 2020–2023 only, with 2024 onward held back as the independent test.
See also: Backtest, Overfitting
- Overfitting
Overfitting happens when a strategy’s rules are chosen after seeing the result, so it learns the noise of the past instead of a real pattern.
The result is a system that is brilliant in a backtest and useless in the market. The more parameters tuned and the shorter the test period, the greater the risk.
The main defence is writing the pass criteria before seeing the result, and holding data back as an independent test.
See also: Out-of-sample testing, Backtest
- Pip
A pip is the conventional smallest unit of price movement, equal to 0.0001 in most currency pairs.
A pip is not a point. On five-digit quotes one pip equals ten points, and confusing the two means being wrong about risk by a factor of ten.
On gold (XAUUSD) a pip is $0.1, not $0.01 — one of the most common mistakes in automated stop-loss sizing.
See also: Lot
- Profit factor
Profit factor is the sum of all winning trades divided by the sum of all losing trades.
Above 1 means the system made money over that period. On real data with real costs, 1.2 to 1.6 is the ordinary range for a healthy system.
A backtest profit factor of 3 or more is usually a warning rather than a boast: either costs were not charged, or the settings were tuned too closely to that very period.
Profit factor = gross profit ÷ gross loss
At LFP: LFP’s backtest profit factor is 1.48 — with real spread and swap, across 8 markets rather than a single chosen one.
See also: Backtest
- Slippage
Slippage is the difference between the price you asked for and the price your order actually filled at.
In a fast or thin market the price moves between sending an order and filling it. Slippage can go your way, but in practice it more often does not.
No simulation reproduces real slippage fully, which is one reason a live account never matches its backtest exactly.
See also: Spread
- Spread
The spread is the gap between a symbol’s buy and sell price, and it is the cost you pay at the moment of entry.
It is not constant: it widens in thin hours, around news, and at the end of the week. A backtest assuming an average or fixed spread flatters the system.
For strategies with small stops the spread can be the difference between profit and loss, which is why markets whose spread is large relative to their daily range are poor candidates for automation.
At LFP: Every LFP backtest is run on the broker’s real recorded spread for that hour of the day, not a fixed figure.
- Swap
Swap is the interest paid or received for holding a position overnight.
The broker sets it, and it follows the interest rate difference between the two currencies. For positions held several days it can be a meaningful part of the result.
At most brokers Wednesday’s swap is charged three times, because it covers the weekend settlement.
See also: Spread
- Underwater period
The underwater period is how long an account stays below its own previous high, stated in days.
It is the companion to drawdown: drawdown says how deep the loss was, the underwater period says how long recovery took. It is usually the second one that makes a trader abandon a system.
"Underwater" is not the same as "losing". An account that went from $10,000 to $14,000 and now sits at $12,000 is in profit and underwater at once.
At LFP: The longest underwater period in the Brave-style backtest was 435 days.
See also: Drawdown
These definitions are educational, not investment advice. Performance figures are backtest results and are no guarantee of future results.