Here's what the data shows.
Asset class breakdown — and what it reveals
| Asset class | Share | Count |
|---|---|---|
| Crypto | 70.7% | 7,180 |
| Stocks | 16.1% | 1,637 |
| ETFs | 5.9% | 599 |
| Commodities | 4.9% | 493 |
| Forex | 2.4% | 245 |
Crypto dominates — no surprise there. But here's the interesting part: almost one in three backtests runs on something else. Stocks (1,600+ runs), ETFs, commodities, and forex combined.
We originally built the Arena as a crypto-first tool. Users had different ideas.
Notably: among ETFs, GLD, SLV, and TLT are at the top. Traders test strategies on gold and bonds, not on SPY. That doesn't match the typical tutorial universe. It matches people who actually think.
What gets tested — and what doesn't
Top strategies by run count:
rsi_sma 631 ← volume monster
rsi_ob_os 69
golden_cross 68
fg_cadence 57
bnh_fixed 46
dca_reference 44
stoch_rsi_sma 22
ema_trend_bias 21
RSI/SMA Cross is by far the most-tested strategy. Nearly 10x more runs than #2. That genuinely surprises us: RSI/SMA is the simplest classic we offer. Pine Script schools dismiss it, "pros" turn up their noses.
But: it's the first one beginners test. And it's the one experienced traders start with when probing a new asset. There's a reason simple indicators get the attention — they're the baseline everything else has to beat.
What we didn't expect: 317 pair/strategy combinations were tested exactly once. A long list of curiosity. TAOUSDT + RSI OB/OS. AWEUSDT + Golden Cross. SUSHIUSDT + RSI/SMA. Nobody twice. That's the tail of the long tail — and exactly what a backtesting platform should deliver: answer every curious question without making anyone write Pine Script.
Win-rate distribution — the inconvenient truth
This is where it gets interesting. When trading Twitter promises you a 70% win rate, look at where our backtests actually land:
| Win rate | Backtest count |
|---|---|
| <30% | 2,123 |
| 30–50% | 4,029 |
| 50–60% | 702 |
| 60–70% | 302 |
| 70–80% | 155 |
| >80% | 188 |
Over 60% of all backtests come in under 50% win rate. Only 3.4% break above 70%.
This doesn't mean those strategies are bad. An RSI/SMA strategy with 30% win rate can still be CAGR-positive if its winners are significantly larger than its losers. Win rate in isolation says nothing — which is why we optimize for "Beat B&H by CAGR + Drawdown," not for a single quote.
Anyone promising you a strategy that reliably hits 70%+ win rate: ask about sample size, out-of-sample validation, and survivorship bias. We wrote about why win rate alone is dangerous in detail.
Beat-vs-Buy-and-Hold: 64.3%
Of all backtests with a meaningful B&H comparison (≥5 trades, comparable period), 643 out of 1,000 (64.3%) beat Buy-and-Hold.
That sounds like a strong rate for active strategies. But: selection bias. Users test strategies because they want to beat B&H — and they keep the runs that do. Catastrophic failures get deleted or ignored.
Real insight: if your strategy beats B&H, it needs to do so consistently across different market regimes. A single run riding the 2017–2018 bull market proves nothing. We show Avg B&H (average of all possible entry points with 20% minimum remaining duration) as the benchmark — it's significantly harder to beat.
Anecdotes we couldn't have made up
16.3 years on TSLA. The longest single run goes back to 2010: Tesla with EMA Trend Bias on monthly bars, 32.7% CAGR. Few trades, long holding periods — the patient long-game trader.
A Chinese-named memecoin at 197,923% CAGR. We're not naming it. What we are saying: B&H on that coin was 2,239% — the strategy beat Buy-and-Hold by a factor of 88x. How? It caught the right pumps and skipped the dumps. Survivorship luck or actual edge? That's the question our Strategy Insights tool is systematically answering.
A2Z on Binance (USDC and USDT). Both variants with RSI/SMA, both spanning 8 years, both CAGRs around −95%. Win rates 11–18%. The asset got traded into the ground; no strategy could salvage it. Sometimes the asset is the problem, not the system.
What's still missing — and where we're headed
Filter adoption (Pro+) is low: 5.6% use the 200-WMA filter, 9.2% the Altcoin Season filter, 8.6% the ATR Volatility filter. The Bullmarket Ampel is freshly rolled out — adoption will show in the coming weeks.
Our hypothesis: many users don't yet know what the filters do. We're building a compact "Filter Explainer" flow directly into the backtest setup, so the cognitive load is smaller.
Strategy Insights — our next major feature — will systematically scan pair/strategy/filter combinations and surface which setups actually outperform robustly versus those that just shine once. Stay tuned.
About the next 10,000: we're shipping an API (api.tradingstrategies.work) so you can run backtests programmatically. Plus conversion-funnel insights so we can understand the most common drop-off points.
If you haven't run a backtest yet: the first three are free. And if you already have one — tell us what you learned. The best insights in this collection don't come from us. They come from you.
— Backtesting Arena
Methodology note: all backtests run on historical data, no look-ahead bias, with the trading-cost model chosen in the backtest setup. Aggregates shown are platform-wide (user runs + admin bulk runs) because data is data. Survivorship bias on top-CAGR anecdotes is explicitly flagged. More in our backtest docs.