No backtest data yet. Click "Run backtest" above ā it takes ~15 minutes and only needs to run once a month.
šÆ Signal accuracy (all signals, traded or not)
šÆ
Not enough tracked signals yet. This builds up automatically as scans run ā checks back on every BUY/SELL/SHORT signal at 1/5/20 days to see if it was right.
š
No trade history yet. Close some positions and your personal stats will appear here.
Main paper engine plus every registered strategy family. Choose V1, V2 Stop-Capped, V3 Crash/Reclaim, V4 Global Event, CRB V2 Daily, or Learning, then select the book. Each keeps its own $25k paper bucket.
Breakout Hunter plus the V1, V2 Stop-Capped, V3 Crash/Reclaim, V4 Global Event, and Learning strategy families on volatile candidates. CRB V2 is intentionally Paper-only because it requires the completed-daily full universe.
A separate experiment: R-Multiple Swing ā its own $25k, its own entry & exit logic, run on the full scan universe. It buys strength on a shallow pullback (an AUTO_READY leader that has cooled off toward its rising 21-EMA, not one that's stretched), risks a fixed 1% per trade, and manages the exit purely in R-multiples: bank ā at +1R (stopābreakeven), ā at +2R (stopā+1R), then trail the runner. Deliberately the opposite discipline to the wide-stop "ride the blow-up" books above, so you can compare the two styles. Hypothetical ā no real money.
š§Ŗ Designed contrast: ARM (regime-aware sizing, 5% heat cap, defensive tightening) vs MPR (aggressive pyramiding control, regime-blind by design) ā with RRR riding the extensions both reject. Same $25k each; the disagreements are the experiment.
Three more experiments, each with its own $25k and its own logic, run on the full scan universe. Pick one to inspect. All hypothetical ā no real money. The point is to compare different philosophies head-to-head; judge them over a meaningful sample across market conditions, not the first few trades.
š¤ Autonomous Shadow Books
AI-created learn_ev_* candidates, each with its own independent $25,000 virtual paper account. These books can be observed and qualified in shadow, but cannot change their donor, register themselves, allocate production capital, or place broker orders.
š¤
Loading autonomous shadow booksā¦
š Trading Evidence Lab
Research-only evidence for selecting one primary and one genuinely diversified live champion. It cannot place orders or alter an existing book.
Champion candidates and book health
Loadingā¦
Advisory capital allocation
Replayāpaperāexecution parity
Gate contribution
Cross-book overlap
Opportunity funnel
Ultimate Book research specification
Decision Science ā precomputed evidence
Walk-forward experiments, statistically corrected promotion evidence, exit quality, and drawdown envelopes. Observation-only; human approval remains mandatory.
Loading performance, mover, broker, storage, and AI-shadow evidenceā¦
š§ Learning ā what the system is teaching itself
Loadingā¦
š¤ SelfAI ā multi-provider task learner
Advisory-only shadow learning across Claude, Gemini, and Perplexity. Every structured AI job is scored separately so strong and weak SelfAI skills stay visible. SelfAI cannot authorize trades or modify books.
Model & fallback-source status
Loadingā¦
Task skill ā what SelfAI can and cannot predict yet
Loadingā¦
Teacher-provider agreement
Loadingā¦
Recent SelfAI ā AI-teacher mismatches
Loadingā¦
Entry-verdict confusion / learned cues
Loadingā¦
Live portfolio risk across every paper book, computed by the streamer and pushed each cycle.