Glossary
The terms used across this site, in plain words.
- Backtest (replay)
- Running a trading rule on past prices to see what it would have done. It shows what happened, not what will happen.
- Paper trading
- Trading on live prices with pretend money. It tests the system without risking money, but fills can differ from real ones.
- Setup
- One indicator rule with specific settings, sometimes with a filter or an exit rule added. This site tested 31,407 of them.
- Bar (candle) and bar length
- A bar summarises prices over a fixed time: open, high, low and close. The bar length, or timeframe, is that time: 1 minute, 1 hour, 4 hours.
- Fee
- What the exchange charges per trade. Upbit charges 0.05% of the amount on its won market, on both buys and sells.
- Tick
- The smallest price step an exchange allows. On Upbit a coin priced between 100 and 1,000 won moves in steps of 1 won.
- Spread
- The gap between the best price buyers offer (the bid) and the best price sellers ask (the ask). A market order pays it.
- Slippage
- The difference between the price you expected and the price you got. Our backtests charge one tick per trade.
- Round trip
- A buy and the sale that closes it. It pays the fee twice and the spread at least once.
- Sea level
- On this site, break-even after costs. Results below it are drawn in blue, as water; results above it in greens and browns, as land.
- Return a year (CAGR)
- The compounded yearly growth rate that would turn the starting value into the final value over the period.
- Worst fall (maximum drawdown)
- The largest fall from a previous high before a new high was reached.
- Average return divided by how much returns swing, on a yearly scale. Higher means steadier returns; above 1 is decent, above 2 is strong.
- Win rate
- The share of trades that made money after costs. On its own it says little about profit.
- Expectancy
- The average result per trade: win rate × average win − loss rate × average loss − costs.
- Two-period test
- Our pass mark: at least +10% a year in both the first period and the second period after costs, with the second keeping at least half of the first.
- First and second period (in-sample, out-of-sample)
- The first period is where choices are made; the second is data the choices never saw. Results in the second period are closer to what live trading might show.
- Walk-forward
- Choosing settings with past data only, then moving forward in time and repeating, so no choice uses information from the future.
- Survivorship bias
- Testing only on what still exists today. Coins that collapsed or were delisted drop out, which makes backtests look better than they were.
- Look-ahead bias
- A backtest using information that was not available at the time, such as trading at a bar's close on a signal computed from that same close.
- Overfitting and multiple testing
- Tuning a rule until it fits the past, or testing so many rules that some look good by chance. Both make backtests look better than the future will be.
- Trend following
- Buying after prices have started to rise and selling after they have started to fall. Wins rarely, but wins big.
- Mean reversion
- Buying after a fall in the hope of a bounce. Wins often, but its losses can be large.
- Breakout
- A price moving above a recent high or a band, taken as the start of a move.
- Filter
- An extra condition that must hold before a signal can trade, such as a longer chart being in an uptrend.
- Stop-loss, trailing stop, take-profit
- Exit rules: sell after a fixed loss, after a fixed fall from the highest price since entry, or after a fixed gain.
- Bid-ask bounce
- Closing prices alternating between the bid and the ask. On short bars it makes prices look like they reverse, and fools fee-only backtests.
- Market order and limit order
- A market order trades immediately at the best available price and pays the spread. A limit order waits at a set price and may not fill.