Backtesting Arena holds more than 25 strategies. The obvious question — "which is best?" — is the wrong one. None beats all the others on every asset and every timeframe. The right question is: which strategy fits which market, in which phase? The library is a toolbox, not a leaderboard.
Here's a tour — and four honest truths that run through all of it.
Trend: fast vs slow
Golden Cross (50/200 SMA) is the famous classic: slow, stable. EMA Cross (9/21) is the fast cousin — reacts earlier, trades more, but produces more whipsaws.
On BNB, daily, over eight years, the gap was dramatic: EMA Cross +92% CAGR (130 trades), Golden Cross +43.6% (24 trades). But careful — that's one coin with a monster run. Across hundreds of runs, EMA Cross's lead over Buy & Hold shrinks to a few points. A single coin flatters; the average tells the truth.
Momentum: and why volume can lie
MACD Cross is the momentum classic. On 2- and 3-day candles it shows an aggregate lead of over 30 points against B&H — and on monthly candles it loses to B&H. One indicator, five timeframes, opposite verdicts.
OBV-MACD applies the same formula to volume instead of price. Clever — but with a dangerous assumption: that the volume is real. In crypto it often isn't. Much of Solana's DEX volume has been bot and wash activity. An indicator is only as clean as its input.
Volatility: the ATR family and the cost of smoothing
Keltner Channel Breakout and Supertrend both use ATR bands, with different discipline: Keltner breaks out directly, Supertrend ratchets its band up as a trailing stop. Keltner grabs more of the expansion; Supertrend gives back less on reversals — but enters later.
Smoothed Heiken Ashi smooths the candles to make trends read cleanly. On 2-day candles that shows a big lead — on weekly candles it loses to just holding. The same smoothing that filters noise lags too much on higher timeframes.
The meta-strategy: trade the alt, but read BTC
BTC Signal computes RSI/SMA on Bitcoin and uses that to decide whether to hold the altcoin. The thesis: if Bitcoin isn't trending, a weaker alt has even less reason to rise. On daily it's barely usable (Score 27); on weekly it's one of the platform's highest-rated (Score 86). Same mechanism — the timeframe decides everything.
Four truths running through the whole library
1. No timeframe king. Almost every strategy behaves completely differently on D, 2D, 3D, W and M. Anyone selling you a strategy without a timeframe is hiding half the story.
2. "Beats B&H" often means: loses less. In the brutal long-tail of alts, Average Buy & Hold is deeply negative. RSI/SMA does −4% on weekly — and still beats B&H by 20 points, because B&H sits at −24.5%. Capital preservation is a strategy. We say it out loud.
3. Win rate isn't profit. EMA Cross wins only 34% of its trades on BNB — and still makes +92%. Trend-following lives on a few big winners, not a high hit rate.
4. Fit beats breadth. Our Wuguan Master strategy shows a 35-point lead on crypto — and "insufficient data" on stocks. That's not a bug. A strategy that "works everywhere" usually works nowhere in particular.
And one more: sample size counts
Every number in the library carries a run count. A lead over 2,600 runs is not the same as one over 50. Small samples get flagged, not hidden — treat them as direction, not proof.
The toolbox is open. Don't hunt for the "best" strategy — hunt for the one that fits your asset, your timeframe, and the current market phase. That's exactly what the library is built for.
Study the Past — Improve your Future. 🥋