Three systems.
Three separate records.
These systems have unequal sample sizes and different logic. The dashboard describes what has happened; it does not rank which strategy is best.
Closed-trade PAPER PnL
* No losing closed trades yet; the sample is too small to treat this as a stable profit factor.
Why the systems coexist.
Established adaptive engine retained as the reference record.
Newer branch deployed separately so changes cannot rewrite V4 history.
Earlier learning architecture preserved as a prior experiment rather than merged into the successor.
Current ML branch retraining on incoming data and executing in PAPER mode.
Open V6 record →A higher number is not automatically better evidence.
V4 has substantially more closed trades than V6, while V4.4 has no closed observations yet. Win rate also ignores payoff size and drawdown. We therefore preserve the raw forward records rather than declaring a winner.