If any trading word on this site felt unclear, look it up here. No jargon, no YouTube-guru word-salad, no "as you well know." Bookmark and refer back.
Browse by category
Core concepts every crypto reader should know.
HODL means simply buying and holding — no trading. Someone who HODLs BTC buys once and never sells. Our strategies must beat this baseline, otherwise why bother with the complexity?
Crypto slang originally from a misspelling of 'hold' on Bitcointalk forum in 2013. It stuck as an ironic badge of honour: 'I'm not trading, I'm just holding through everything.' For us, HODL is the passive benchmark every active strategy must beat after fees — otherwise the strategy's complexity is pure waste.
Instead of putting $12'000 in at once, you put $1'000/month for 12 months. Reduces regret-risk from bad timing. Mathematically slightly worse than lump-sum on average — but emotionally much easier to sustain.
Mathematically, lump-sum investing beats DCA about 66% of the time over long horizons because markets trend up. But the 34% of the time DCA wins, it tends to win BIG. And the psychological benefit — never feeling 'I bought at the top' — often makes DCA the superior real-world choice.
How we test strategies and avoid lying to ourselves.
Running a strategy's rules on historical price data to see how it would have performed. Useful but dangerous — a backtest can easily look amazing and still fail live. See also: Out-of-sample.
A backtest method: we split the historical data into 3 independent time windows and check whether the strategy beats HODL in each one separately. Catches strategies that only work in one specific market regime.
Walk-forward is stricter than a single full-period backtest. If a strategy beats HODL in only 1 of 3 windows, it's regime-dependent (maybe it only works in bull runs). If it beats HODL in all 3, it's a genuine edge. We require 2/3 minimum for any live-tier bot; 3/3 is the gold standard (e.g., The Watchdog, Rotator, Hedge Hopper).
'Out-of-sample' means: the strategy is tested on data that was NOT seen during development. This catches overfitting. 'In-sample' = data the strategy was built on. Out-of-sample performance is the more meaningful measure.
A strategy is parameter-robust when it works across a range of settings — not just at one magic value. Example: works at 20, 25, 30, 35, 40-day lookbacks (robust) vs. only at exactly 22 days (fragile, likely a coincidence).
After our Sharpshooter post-mortem (a backtest that showed +2'053% at exactly 14-day lookback but fell apart at 13 or 15) we require every new strategy to pass a parameter-neighbourhood sweep before deployment. Real edges are robust; lucky backtest spikes are not.
Numbers that describe how much a strategy can hurt you.
The worst loss from a peak to the subsequent trough. If the bot falls from $20'000 to $11'600 that's a -42% drawdown. Note: not the loss at the end, but how deep it went in between.
Drawdown matters more than total return for real-world traders. Most people quit a strategy during the bottom of its drawdown, turning a paper loss into a realized one. A strategy with 15%/yr return and -30% max drawdown is often better in practice than 25%/yr with -70% drawdown — because the second one shakes you out.
Measures how evenly returns come in over time. High Sharpe = many small, stable gains. Low Sharpe = wild swings even if total return is high. Above 1.0 is solid, above 2.0 is excellent.
Annual return divided by the worst interim loss. Higher = better balance between gains and pain endured. Example: 80% annual return with -40% drawdown = Calmar 2.0.
The difference between the price your order intended to fill at vs. the price it actually filled at. On liquid majors it's ~0.05%. On thin memecoins it can be 5-15% per trade.
When a leveraged position's margin is wiped out and the exchange forcibly closes your position. You lose 100% of that trade's collateral. Much more common than most retail traders admit.
Cognitive and statistical traps that wreck backtests + decisions.
The urge to buy an asset that's already up 30-50% in a week, because you're afraid to miss the next 30%. Historical base rate: FOMO-day entries usually mark the local top.
Studying only the survivors of a process, ignoring the ones that failed. Classic example: copy-trading leaderboards only show traders who haven't blown up yet — invisible graveyard of losers.
Using information in a backtest that wouldn't have been available at the time. Example: using end-of-day close to trigger a signal that supposedly fires mid-day.
Testing so many strategy variants that one of them looks statistically significant by pure chance. If you test 20 random strategies, 1 will look 'significant' at the 5% level — and be useless.
Tuning a strategy so perfectly to past data that it breaks on new data. A strategy that captured every single trade in the backtest is almost certainly overfit.
Overweighting what just happened. Most traders build strategies that would have worked in the last 3 months of data — ignoring that the last 3 months are a tiny, atypical slice of history.
How our bots actually operate under the hood.
A filter that detects whether the market is in a bull or bear phase. When the filter says 'bear', the bot sits in cash and waits. When it says 'bull', the bot buys. Dramatically reduces the worst drawdowns.
Share of trades that closed profitably. Example: 60% = 6 out of 10 trades made money. Careful: a high win rate ≠ high profit (many small wins can be wiped out by a few large losses).
A regime (or 'market phase') is a longer stretch of time with similar market behavior — e.g. bull market (rising), bear market (falling), sideways. A good strategy works in multiple regimes, not just one.
The tendency of prices that have been going up to keep going up (and vice versa). A 30-day momentum = compare today's price to the price 30 days ago. Positive = uptrend.
A bot that places a ladder of buy/sell orders at evenly-spaced price levels. Harvests volatility when price wiggles inside the ladder. Breaks when price leaves the range.
You pick a 'lead trader' from a platform leaderboard, allocate capital, and the platform automatically mirrors their trades into your account. Lead trader takes a share of profits as fee.
If you hit a word on the site that isn't in this glossary, reply to any newsletter and I'll add it. This page is maintained as a living reference.
Subscribe & suggest a term