Research tracks

Research Summary

High-level conclusions from the trading lab, organized around the current operating branch, paused detector work, and active exit optimization.

Baseline paper return

86.59%

Current operating branch

Exit variant return

124.42%

Current comparison branch

Baseline max drawdown

-12.22%

Operational risk path

Exit variant max drawdown

-14.93%

Comparison branch risk path

Core conclusions

Most durable conclusions

Regime-native logic beat one universal strategy. Matching tactics to market state is still the main edge hypothesis.
Bull conditions: buying meaningful dips remains the strongest practical idea.
Bear conditions: fading strong rallies remains the strongest practical idea.
Sideways conditions: current sideways trading logic remains weak and should not be a focus until something cleaner appears.
Main bottleneck: live state detection and operational validation, not basic strategy discovery.

Experiment 1

The main operating branch is stable enough to keep running. It moved from manual benchmark to live detector to hardened recurring paper operations.

Experiment 2

This branch found a structurally calmer detector, but the dry validation underperformed badly enough that it should stay research-only for now.

Experiment 3

This is the active improvement branch. It found an exits-only variant strong enough to justify a live paper comparison against the baseline.

Next steps