TL;DR
The 200-day ETH bot returned 42.43% after fees in 2022 to 2024 while holding ETH lost 10.44%, with substantially smaller drawdown. Both lost money in January 2025 through June 2026: the bot lost 22.52% versus holding ETH's 52.91%. In July through September 2026, the bot gained 24.97% versus 50.47%. Its delayed entry missed part of the rebound. These inspected historical tests include 0.7% fees per fill.
- 2022-2024 bot return
- +42.43%
- 2022-2024 holding ETH
- -10.44%
What happened to $10,000?
- 200-day ETH bot 2024-12-31, 21:00:00 UTC
- 14,242.64
- Hold ETH 2024-12-31, 21:00:00 UTC
- 8,955.62
2022-2024: the bot finished ahead
The bot finished with $14,242.64. Holding ETH finished with $8,955.62. Follow both lines: the bot still experienced a 33.14% maximum drawdown, compared with 76.28% for holding ETH.
- Bot ending value
- $14,242.64
- Hold ETH ending value
- $8,955.62
- Bot return after fees
- +42.43%
- Hold ETH return after fees
- -10.44%
This was the development period. We chose the 200-day rule after comparing it with 50-day and 100-day variants on these same dates.
- 200-day ETH bot 2026-06-30, 20:00:00 UTC
- 7,748.37
- Hold ETH 2026-06-30, 20:00:00 UTC
- 4,709.12
2025-June 2026: both lost money
The unchanged bot finished with $7,748.37, while holding ETH finished with $4,709.12. Losing less was useful in this comparison, but the bot still lost 22.52% and experienced a 40.83% maximum drawdown.
- Bot ending value
- $7,748.37
- Hold ETH ending value
- $4,709.12
- Bot return after fees
- -22.52%
- Hold ETH return after fees
- -52.91%
A trend rule can still suffer substantial losses. These later historical dates were already inspected; they are not a fresh unseen test.
- 200-day ETH bot 2026-09-30, 20:00:00 UTC
- 12,497
- Hold ETH 2026-09-30, 20:00:00 UTC
- 15,046.96
July-September 2026: holding ETH won
The bot finished with $12,497.00. Holding ETH reached $15,046.96. The bot waited before entering, then made one buy and no exit during this period.
- Bot ending value
- $12,497.00
- Hold ETH ending value
- $15,046.96
- Bot return after fees
- +24.97%
- Hold ETH return after fees
- +50.47%
The bot captured less of this rebound. Its ending value includes ETH still held, rather than a completed sale.
Three separate $10,000 backtests.
Watch the rule make an actual decision
The bot buys above the 200-day average and exits at or below it.
January 1: price above the average → buy
At 14:30 UTC, ETH was $3,709.01 and the 200-day average was $3,365.17. The portfolio held no ETH, so the bot spent $9,500 on 2.561330 ETH and paid a $66.50 fee.
- ETH price
- $3,709.01
- 200-day average
- $3,365.17
- Recorded action
- Buy 2.561330 ETH
- Buy fee
- $66.50
The rule uses 95% of available buying power for the purchase. The remaining cash after this fill was $433.50.
January 6: price below the average → exit
At 14:30 UTC, ETH was $3,389.72 and the average was $3,400.77. The bot sold all 2.561330 ETH. The $60.78 sell fee was deducted from the sale proceeds.
- ETH price
- $3,389.72
- 200-day average
- $3,400.77
- ETH after the exit
- 0 ETH
- Portfolio after the exit
- $9,054.92
The position was empty and the portfolio held $9,054.92 in cash. This first round trip lost $945.08 after fees. The full backtest finished ahead despite losses like this one.
Open these rules in Aurora
- Asset
- Spot ETH-USD
- Entry
- Above the 200-day average
- Exit
- At or below the average
Start with the exact configuration used in this study.
Choose your next test before changing the bot
For a reproduction, use the published dates, $10,000 starting capital, Day observations and 0.7% fee on every filled buy and sell.
Exact rules and test settings
Inspect all strategy settings
Enter when ETH's observed price is above its 200-calendar-day simple moving average and the portfolio holds no ETH.
| Setting | Recorded choice |
|---|---|
| Starting portfolio | $10,000; fractional spot ETH; dollar cash |
| Entry and exit | Above 200-calendar-day mean to enter; at or below to exit |
| Allocation | 95% of buying power; one ETH position |
| Fee | 0.7% of submitted notional per fill |
| Baseline | One native ETH purchase, including its 0.7% fee; no staking yield |
| Evaluation | Native Day observations; same-observation modeled market fills |
No matching rows. Clear the filter to see all records.
How the 200-day rule was selected
Show the table · 3 rows, 4 columns
Scroll sideways to compare all columns. Select a heading to sort.
| 50-day SMA / Sell100 | -27.56% | 50.48% | 45 |
| 100-day SMA / Sell100 | -23.98% | 51.69% | 30 |
| 200-day SMA / Sell100 | 42.43% | 33.14% | 14 |
No matching rows. Clear the filter to see all records.
Fresh corrected-engine runs over January 1, 2022 through December 31, 2024. All returns include the declared per-fill costs. Selection used net return; drawdown is reported separately.
The three candidates change only the moving-average window: 50, 100 or 200 calendar days.
Complete entry and exit audit
Inspect the full fill audit
The selected development run has 29 filled orders and 14 complete exits.
What this backtest includes and leaves out
Read data, costs and execution assumptions
This study trades spot ether against dollars.
Scroll sideways to compare all columns. Select a heading to sort.
| 2022-2024 development | 42.43% | -10.44% | 33.14% | 76.28% |
| 2025-June 2026 historical check | -22.52% | -52.91% | 40.83% | 68.27% |
| July-September 2026 historical check | 24.97% | 50.47% | 5.47% | 6.18% |
No matching rows. Clear the filter to see all records.
Independent $10,000 starts with matched observation timestamps and 0.7% fees on actual fills. Ending holdings are marked rather than liquidated.
Paper-test and deploy this ETH trading bot
Review broker connections and deployment checks
After reviewing the rules and historical tests, prepare a separate paper portfolio from the saved draft.
Correction recorded October 11, 2026
Research correction and provenance
Sell sizing now requests the full held quantity for a normal 100% exit and deducts its fee once from sale proceeds. The engine repair replaces the previous oversized-percentage workaround. All research results and comparisons in this revision were rerun after the repaired engine was deployed. Earlier receipts are retained as historical records.
No comments yet.