Orbit

Every Algo Court bot goes through the same 10 steps. Each one shows the real number behind it.

1. 01

   Rules coded exactly as taught

   Each member started as a strategy taught in a video or book. We coded the taught rules exactly, before changing anything. Every rule is in its strategy review.

   5strategies, each with its own strategy review
2. 02

   Every trade re-counted

   We rebuilt each strategy from scratch and compared it with its published results.

   5 of 5rebuilds matched. 1,857 of 1,857 published member trades re-counted, profit factors to the published decimals.
3. 03

   Costs on every trade

   Commission, spread and slippage are charged on every single trade.

   1,817trades in the results, all counted after costs (2019–2023 and 2024–2026).
4. 04

   Designed on 2019–2023. Judged on 2024–2026.

   We built the bot on the first five years. Then we ran it on 2024–26 data.

   1.62→1.55profit factor, design → test (1.00 is break-even)
5. 05

   Members don't move together

   The strategies win and lose on different days. That smooths the ride.

   0.07highest correlation between any two members (Golden Cross × Opening Gap Fill). 0 is unrelated, 1 is identical.
6. 06

   Risk set on design data only

   Total risk was sized so the deepest drop in 2019–2023 was 8% of the account. The test years played no part.

   −4.27%worst day in 2019–2023. Deepest drop in that period: −8.0%.
7. 07

   5,000 shuffled-history stress tests

   Stress test: we shuffled the order of the test-period trades 5,000 times to see how bad a bad run can get.

   13.9%is the 95th-percentile drawdown: 1 run in 20 was worse. The typical run was 7.7%.

   Max drawdown of each shuffled run. Runs worse than 18% are counted in the last bar. Design-period 95th percentile: 12.8%.

8. 08

   Prop-firm style limits (5% daily / 10% total)

   We checked every day, and every rolling 30 days, against a 5% daily and 10% total loss limit.

   	Design 2019–2023	Test 2024–2026
   Days past the 5% daily limit	0	0
   10% total limit hit	No	No
   30-day windows that breached	0.0% 0 of 1,557	1.4% 11 of 789

9. 09

   Same logic on cTrader and MetaTrader 5

   One bot, two platforms. Both versions compile with no errors and no warnings.

   cTrader0 errors, 0 warnings

   MT50 errors, 0 warnings
10. 10

    Forward test on a demo account

    Next: a public forward test on a demo account. Results will be published here as they come in. No backtest can replace this step.

    STARTING SOONon a demo account

Each member has its own strategy review. The bot runs them together, each on its own market and timeframe.

77% of months were positive in the test period, 77% in the design period.

A preset scales the risk on every trade. The rules and markets stay the same.

• Standard. The tested design.
• Conservative. Smaller trades, so the worst day in the design years stays at or under 2%.
• Prop-firm safe. Smaller still, so the 95th-percentile stress-test drawdown stays at or under 6%. That 6% target was set using the test years, as a safety margin, not as a result.

On prop-firm accounts use Prop-firm safe. At Standard, 1 in 20 stress-test runs fell more than 13.9%, past the 10% total-loss limit.