Crypto Research · satgas.ai.id
basket_mom validation (E06)
One strategy passed the initial gate. This page re-tests it against a stricter
standard: whether any future information leaked, whether the results hold up after
costs are raised, whether the parameters were chosen too cleverly, and whether the numbers still make
sense after correction because we have tried many combinations.
1. Look-ahead test (prefix)
Signals are recalculated on data truncated at 40 random points. If the signal on the
truncated data differs from the full-data signal, it means future information leaked.
| Points tested | 40 |
| Signals differ | 0 |
| Verdict | no leakage |
2. Cost sensitivity
| Cost per transaction | Sharpe |
Return | CAGR | Max DD |
|---|
| 0 bps | 1.71 | 6,527.2% | 100.5% | -37.7% |
| 5 bps | 1.68 | 6,000.4% | 97.7% | -38.8% |
| 10 bps | 1.66 | 5,515.4% | 95.0% | -39.8% |
| 15 bps | 1.63 | 5,068.8% | 92.4% | -40.9% |
| 25 bps | 1.57 | 4,279.0% | 87.2% | -42.9% |
| 40 bps | 1.48 | 3,314.3% | 79.6% | -45.8% |
The 15 bps cost is the baseline assumption (10 fee + 5 slippage). If Sharpe collapses
at 25 bps, this strategy is fragile under real execution conditions.
3. All parameters, not just the winner
| Parameter | Sharpe | Return |
CAGR | Max DD | Exposure |
|---|
| {"lookback": 30} | 1.63 | 5,068.8% | 92.4% | -40.9% | 65% |
| {"lookback": 60} | 1.33 | 2,536.4% | 72.0% | -58.7% | 65% |
| {"lookback": 90} | 0.95 | 691.8% | 40.9% | -63.9% | 62% |
| {"lookback": 180} | 0.65 | 223.1% | 21.5% | -54.4% | 65% |
If only one value is good and the rest are bad, that is a sign of overfit.
A clean spread across several values is more convincing.
4. Last 12 months holdout test
| Parameters frozen from data before 2025-10-01 | {"lookback": 30} |
| Test period | 389 days |
| Strategy return | -5.3% |
| BTC return over the same period | -31.8% |
| Strategy vs BTC Sharpe | -0.03 vs -0.60 |
| Strategy drawdown | -31.1% |
Parameters were chosen only from data before the cutoff date, so the numbers here
have never been used to choose anything. This is the test closest to real conditions.
5. Data frequency (weekly bars)
| Weekly Sharpe | 1.57 |
| Return | 4,886.1% |
| Drawdown | -39.0% |
| Number of bars | 315 |
If the effect appears only on daily bars but disappears on weekly bars,
it is most likely micro noise, not an economic edge.
6. Universe dependence
| Universe | Sharpe | Return |
Max DD |
|---|
| BTC, ETH | 1.18 | 514.2% | -29.8% |
| BTC, ETH, SOL | 1.63 | 5,068.8% | -40.9% |
The universe uses coins that still exist today. Dead coins have no
exchange data, so the results contain survivorship bias and must be read as
an upper bound, not an achievable number.
7. Null test with two block lengths
| Block | Original Sharpe |
Null median | p-value |
|---|
| 10 days | 1.63 |
0.91 | 0.012 |
| 30 days | 1.63 |
1.10 | 0.012 |
Two block lengths are used to ensure the conclusion does not depend on
a single test parameter choice. A p below 0.05 means it is hard to explain with random data.
8. Deflated Sharpe Ratio
| Daily Sharpe | 0.0851 |
| Sharpe expected from luck (SR0) | 0.0977 |
| DSR (probability of real edge) | 0.278 |
| Return skew / kurtosis | 0.28 / 12.11 |
DSR corrects Sharpe because we have tested dozens of parameter combinations.
A value below 0.95 means the claim of edge is not strong enough to be considered certain.
9. Paper trading notes
| Start | 2026-10-25 |
| Days recorded | 1 |
| Simulated equity | $10,000 |
| Paper return | 0.00% |
| BTC return over the same period | 0.00% |
| Last position | {} |
The number of days is still small. Paper trading cannot be used to conclude
anything until at least 30 days and after passing one period of market stress.
Conclusion
Look-ahead: clean. At 15 bps cost, Sharpe 1.63 and CAGR 92.4%. At 25 bps cost, Sharpe 1.57. 12-month holdout: strategy -5.3% vs BTC -31.8%. Null test significant for 2 of 2 block lengths. DSR 0.28 (needs >= 0.95 for a strong claim). Paper trading 1 day, +0.00% vs BTC +0.00%. Conclusion: the evidence is not yet sufficient to consider this a persistent edge.
Updated 25 Oct 2026 08:53 UTC+7. All numbers are calculated
from data stored in research/db. There is no real money, no orders.