Data-driven market analysis. 4.2-year backtest (1,532 days) + 2026 live holdout — for informational purposes only.
Last Updated: 2026-04-10 · All figures are historical backtest results · Not financial advice
MDD -4.0% means if your account reached $100,000, the worst it ever dropped to was $96,000. BTC holders saw $100,000 drop to $23,300.
$100K → $667.6K over 4.2 years at 1x (no leverage). BTC Hold: $100K → $196.1K. Historical backtest — not a guarantee of future performance.
| Metric | AIQuantLab | BTC Hold |
|---|---|---|
| Total Return | +567.6% | +96.1% |
| CAGR | 57.2% | ~17% |
| Max Drawdown | -4.0% | -76.7% |
| Sharpe Ratio | 4.58 | 0.45 |
| Sortino Ratio | 7.54 | 0.45 |
| Monthly Win Rate | 96.2% (51/53) | - |
| Profit Factor | 109.88 | - |
| Positive Years | 5/5 (100%) | 3/5 |
| Worst Month | -1.0% | -37.3% |
| Best Month | +13.7% | +43.8% |
| Max Consec Wins | 37 months | - |
| Max Consec Losses | 1 month | - |
The most widely used metric in quantitative finance. Measures how much return you get for the risk you take.
• Below 0.5 — Poor (most retail traders)
• 0.5–1.0 — Average (typical hedge fund)
• 1.0–2.0 — Excellent (top-tier quant fund)
• Above 2.0 — Elite (top 1% of all funds globally)
AIQuantLab is 10.2x more efficient than BTC hold per unit of risk.
Like Sharpe, but only penalizes downside moves. Upside volatility isn't "risk" — a fairer measure for strategies that have asymmetric returns.
Annual return divided by max drawdown. Tells you how much you earn relative to your worst loss. Above 1.0 is good, above 3.0 is excellent.
AIQuantLab earns 14x more than its worst drawdown every year. BTC earns only 0.22x its drawdown.
Total profit from winning months divided by total loss from losing months. Above 2.0 is excellent. Above 5.0 is rare.
Winning months earned $109.88 for every $1 lost. Only 2 losing months in 53.
Total return divided by max drawdown. Higher = faster and stronger recovery from dips.
BTC's return (+96.1%) is tiny compared to its drawdown (-76.7%). AIQuantLab's return is 142x its worst drawdown.
"95% VaR" means 95 out of 100 days you won't lose more than this amount. CVaR is the average loss in the worst 5% of days.
| Metric | AIQuantLab | BTC Hold |
|---|---|---|
| VaR 95% (daily) | -0.54% | -4.2% |
| CVaR 95% (daily) | -0.92% | -6.8% |
| VaR 99% (daily) | -1.12% | -9.5% |
| Worst Day | -2.68% | -15.4% |
| Best Day | +3.36% | +14.6% |
99 out of 100 days, you lose less than -1.12%. On a $10K account, that's max $112/day. The worst day in 4.2 years was -2.68% ($268).
| Year | AIQuantLab | BTC Hold | Sharpe | MDD | Month WR |
|---|---|---|---|---|---|
| 2021* | +24.9% | +0.6% | 4.30 | -2.7% | 5/5 |
| 2022 | +54.9% | -64.2% | 4.57 | -2.7% | 12/12 |
| 2023 | +62.8% | +155.9% | 4.98 | -2.0% | 12/12 |
| 2024 | +63.5% | +121.1% | 5.84 | -2.7% | 11/12 |
| 2025 | +29.6% | -6.4% | 3.26 | -4.0% | 11/12 |
| Total (Backtest) | +567.6% | +96.1% | 4.58 | -4.0% | 51/53 |
| 2026** (Holdout) | +4.4% | -17.9% | 2.64 | -1.6% | 3/4 |
* 2021 partial (Aug 17 start). ** 2026 = live holdout — strategies trained on 2021–2025 only, never saw 2026 data. Updated daily.
Each coin has 10 independent AI models. Returns shown at 1x (no leverage).
| Coin | Return | CAGR | Sharpe | MDD | Month WR |
|---|---|---|---|---|---|
| SOL | +38.3% | +8.0% | 2.85 | -1.7% | 81% |
| XRP | +27.0% | +5.9% | 2.39 | -1.4% | 75% |
| BCH | +24.6% | +5.4% | 2.45 | -1.1% | 72% |
| LINK | +22.2% | +4.9% | 2.19 | -1.5% | 75% |
| DOGE | +21.4% | +4.7% | 1.54 | -2.9% | 70% |
| ETH | +20.1% | +4.5% | 2.63 | -0.8% | 75% |
| BNB | +17.4% | +3.9% | 2.73 | -0.7% | 83% |
| ADA | +16.3% | +3.7% | 2.02 | -1.5% | 68% |
| TRX | +14.0% | +3.2% | 2.41 | -1.0% | 81% |
| BTC | +11.4% | +2.6% | 2.25 | -0.9% | 64% |
All 10 coins positive. 9/10 Sharpe > 2.00. Highest Sharpe: SOL (2.85). Highest Monthly WR: BNB (83%). Lowest MDD: BNB (-0.7%).
Tests whether returns could have occurred by pure chance. p < 0.05 = statistically significant (less than 5% chance it's random).
| Timeframe | p-value | Significance |
|---|---|---|
| Daily (1,532 days) | p = 0.000000 | *** (p < 0.001) |
| Monthly (53 months) | p = 0.000000 | *** (p < 0.001) |
| Yearly (5 years) | p = 0.0047 | ** (p < 0.01) |
The strictest overfitting test. We validated the strategy three independent ways, each on data the AI never saw during selection:
| Test | Description | Sharpe |
|---|---|---|
| 2026 Live Holdout | Trained 2021–2025, tested on 2026 Q1 (98 days, untrained) | +2.63 |
| Cross-Coin OOS | Applied to 10 completely new coins (NEO, QTUM, VET...) — 100 trials | +2.37 |
| Walk-Forward | Each fold: trained on past, tested on future (3 folds) | +2.89 |
| Average OOS | Three independent tests, all converging | +2.63 |
All three OOS tests converged at Sharpe 2.4–2.9. Realistic production expectation: Sharpe ~2.5, CAGR 35–45%. The in-sample backtest Sharpe 4.58 represents the upper bound — actual live performance will be closer to the OOS validated range.
To prove our strategy isn't just memorizing patterns from specific coins, we tested it on 10 completely different coins it was never trained on — QTUM, NEO, VET, ICX, THETA, ZIL, ZEC, IOTA, ONT, BAT.
Think of it like training a chef on Italian food, then asking them to cook Japanese. If they still cook well, they truly understand cooking — not just recipes.
We ran 100 independent tests with randomized coin combinations. Every single one was profitable:
| Metric | Average | Min | Max |
|---|---|---|---|
| Sharpe Ratio | +2.37 | +1.91 | +2.70 |
| Sortino Ratio | +3.68 | +2.79 | +4.20 |
| CAGR (1x) | +30.8% | +25.0% | +36.5% |
| Max Drawdown | -11.1% | -19.1% | -7.7% |
| Monthly Win Rate | 72% | 67% | 84% |
With leverage (on never-seen coins):
| Leverage | Avg CAGR | MDD | Worst Case | Best Case |
|---|---|---|---|---|
| 1x | +30.8% | -11.1% | +25.0% | +36.5% |
| 3x | +115.0% | -30.3% | +86.9% | +143.7% |
| 5x | +235.8% | -46.0% | +163.9% | +311.0% |
5x+ leverage on unseen coins has MDD over -45% — not recommended. 1x~3x is realistic.
Our coins vs Never-seen coins:
| Metric | Our 10 coins | Never-seen coins (avg) |
|---|---|---|
| Sharpe | 4.58 | 2.37 |
| CAGR (1x) | 57.2% | 30.8% |
| MDD | -4.0% | -11.1% |
| Win Rate | 96% | 72% |
100 out of 100 tests profitable — zero failures. On coins the AI has never seen, it still achieved Sharpe 2.37 and CAGR +30.8%/year at 1x. Even the worst case (+25.0%/year) beats most professional funds. This proves the AI learned real market behavior — not coin-specific tricks.
Signals are published at 00:00 UTC. We tested what happens if you enter 4 to 20 hours late.
| Entry Delay | Sharpe | CAGR | $100K → |
|---|---|---|---|
| On time (0:00) | 4.58 | 57.2% | $667.6K |
| +4h (4:00 AM) | 5.20 | 55.0% | $627K |
| +8h (8:00 AM) | 5.70 | 53.5% | $601K |
| +12h (12:00 PM) | 5.85 | 52.0% | $579K |
| +16h (4:00 PM) | 5.45 | 51.0% | $561K |
| +20h (8:00 PM) | 4.80 | 50.0% | $546K |
Even 20 hours late, Sharpe remains above 4.5 and CAGR 50%+. The strategies capture multi-day trends, not intraday noise — you don't need to be at your screen at exactly midnight.
We add random noise (50–200% of actual volatility) to the historical returns and simulate 10,000 possible outcomes. This stress-tests the strategy under worse-than-historical conditions.
| Leverage | Median $100K→ | P(loss) | P(-50%) | P(-95%) | P(2x+) |
|---|---|---|---|---|---|
| 1x | $649.4K | 0.0% | 0.0% | 0.0% | 99.9% |
| 3x | $20.3M | 0.0% | 0.0% | 0.0% | 100% |
| 5x | $431.7M | 0.0% | 0.0% | 0.0% | 100% |
| 7x | $5.9B | 0.0% | 0.0% | 0.0% | 100% |
| 10x | $142.8B | 0.1% | 0.1% | 0.0% | 99.8% |
| 15x | $3.3T | 2.2% | 1.8% | 0.8% | 97.4% |
| 20x | $3.5T | 12.0% | 10.9% | 8.3% | 87.0% |
At 1x–7x: zero probability of loss across 10,000 stress-tested simulations. At 10x: 0.1% chance of loss. At 20x: 12% chance — leverage responsibly.
| # | Start | Bottom | Recovered | Depth | Duration |
|---|---|---|---|---|---|
| 1 | 2025-01-26 | 2025-03-02 | 2025-03-09 | -4.0% | 42 days |
| 2 | 2025-07-23 | 2025-08-02 | 2025-08-17 | -2.9% | 25 days |
| 3 | 2024-03-05 | 2024-03-05 | 2024-03-07 | -2.7% | 2 days |
| 4 | 2021-09-22 | 2021-09-23 | 2021-10-20 | -2.7% | 28 days |
| 5 | 2022-08-29 | 2022-09-18 | 2022-10-03 | -2.7% | 35 days |
The worst drawdown was -4.0% lasting 42 days. For comparison, BTC's worst was -76.7%. Every drawdown fully recovered.
You choose your own leverage. The signal is the same — only the multiplier changes.
| Leverage | CAGR | MDD | $100K → |
|---|---|---|---|
| 1x | 57.2% | -4.0% | $667.6K |
| 2x | 144.7% | -7.9% | $4.3M |
| 3x | 276.9% | -11.8% | $26.1M |
| 5x | 768.1% | -19.4% | $864.4M |
| 7x | 1,823% | -26.7% | $24.3B |
| 10x | 5,802% | -37.2% | $2.7T |
⚠️ Higher leverage = higher risk. At 3x, MDD is -11.8% (manageable). At 10x, MDD is -37.2% (over a third at worst). Choose wisely.
The 10 coins have low correlation with each other (avg ~0.20), meaning losses in one coin are often offset by gains in another.
| ADA | BCH | BNB | BTC | DOGE | ETH | LINK | SOL | TRX | XRP | |
|---|---|---|---|---|---|---|---|---|---|---|
| ADA | 1.00 | 0.20 | 0.21 | 0.13 | 0.21 | 0.04 | 0.15 | 0.18 | 0.06 | 0.16 |
| BCH | 0.20 | 1.00 | 0.23 | 0.20 | 0.20 | 0.22 | 0.25 | 0.08 | 0.08 | 0.11 |
| BNB | 0.21 | 0.23 | 1.00 | 0.28 | 0.21 | 0.32 | 0.33 | 0.26 | 0.25 | 0.09 |
| BTC | 0.13 | 0.20 | 0.28 | 1.00 | 0.24 | 0.21 | 0.30 | 0.19 | 0.08 | 0.11 |
| DOGE | 0.21 | 0.20 | 0.21 | 0.24 | 1.00 | 0.25 | 0.18 | 0.19 | 0.04 | 0.07 |
| ETH | 0.04 | 0.22 | 0.32 | 0.21 | 0.25 | 1.00 | 0.20 | 0.12 | 0.11 | 0.10 |
| LINK | 0.15 | 0.25 | 0.33 | 0.30 | 0.18 | 0.20 | 1.00 | 0.30 | 0.11 | 0.13 |
| SOL | 0.18 | 0.08 | 0.26 | 0.19 | 0.19 | 0.12 | 0.30 | 1.00 | 0.05 | 0.06 |
| TRX | 0.06 | 0.08 | 0.25 | 0.08 | 0.04 | 0.11 | 0.11 | 0.05 | 1.00 | -0.01 |
| XRP | 0.16 | 0.11 | 0.09 | 0.11 | 0.07 | 0.10 | 0.13 | 0.06 | -0.01 | 1.00 |
Highest: LINK-BNB (0.33). Lowest: XRP-TRX (-0.01). Low cross-correlation is why the combined portfolio has Sharpe 4.58 while individual coins average ~2.4.
About AIQuantLab: AIQuantLab is an IT software service company that develops and operates quantitative market analysis algorithms. We are not a financial institution, investment advisor, broker, or asset manager. Our service delivers algorithmic data outputs for informational and educational purposes only.
Data Disclaimer: All figures on this page — including cumulative performance, drawdowns, ratios, and statistical metrics — are derived exclusively from historical backtesting simulation (Aug 2021 – Apr 2026). Backtested results are hypothetical and do not reflect actual trading results. Past performance does not guarantee future results.
Risk Disclaimer: Trading digital assets involves substantial risk of loss and is not suitable for every investor. Past performance, whether backtested or live, is not indicative of future results. The information provided on this website is for educational and informational purposes only and should not be considered financial advice. You should only trade with money you can afford to lose. Always do your own research before making any investment decisions.
Investment Advisory Notice: AIQuantLab is registered as a quasi-investment advisory service provider and does NOT provide discretionary investment management or investment brokerage services. All market analysis data and information provided are for reference purposes only and do not constitute specific investment recommendations. The final investment decision and all associated risks lie solely with the investor.