Competition Day 1 is August 31, 2026. Every model begins at exactly $10,000. No P&L is credited until real OHLCV data for an official market session is imported and processed.
Official Day
0 / 90
Not started
Starting Capital
$10,000
Each model independently
Universe Loaded
0
Locked common universe
Primary Judge
Profit Factor
Sample size + DD audit
Execution
Next Open
Signal at close; no same-bar entry
Learning
Day 30 / 60
Learning twins only; evidence-gated
Official Competition Scoreboard
| Rank | Model | Type | Equity | Return | PF | Trades | Win % | Avg R | Max DD | Open / Pending | M1 PF | M2 PF | M3 PF | Changes |
|---|
PF of ∞ is mathematically possible before the first losing trade and is marked low-sample until enough closed trades exist. It is not treated as proof of an edge.
DATA FEED: STARTING…BUILD v1.8
The real-data feed starts automatically when the tracker opens. You can also start it manually.
Inspect a Model
Model
Current Strategy State
Open Position / Pending Order
Monthly Review Ledger
Decision & Learning Journal
1. Competition Setup
Official Start Date
Base Risk / Trade
Slippage / Side
Setup locks after the first official market day is processed. To preserve the experiment, changing start date, risk, or execution assumptions after Day 1 requires a full reset.
2. Broad-Market Real Data
DATA FEED: NOT CHECKED
The default universe is a broad liquid U.S. market set of 245 stocks/ETFs, not a 40-stock cap. It is editable before Day 1 and the engine accepts any practical number of symbols. All 12 models receive the same locked universe. CSV import remains available as a backup.
Official Universe
Feed has not been refreshed yet.
Completed-session protection: today's daily bar is not admitted until after 4:15 PM New York time. The engine never fabricates a missing market day.
CSV Backup Import
No CSV files imported.
3. Advance the Official Competition
The engine will process only unprocessed dates on or after the official start date. It will never simulate missing future days.
Loaded Market Data
| Symbol | Bars | First Date | Last Date | Last Close |
|---|
This tab is a synthetic research sandbox only. Any P&L shown below is excluded from the official competition and must never be reported as Day-1 / live performance.
Competition Governance
Frozen models: parameters never change during the 90-day official test.
Learning twins: start with exactly the same rules as their frozen twin. Adaptation is allowed only after the official Day-30 and Day-60 closes.
Evidence gate: fewer than 3 closed trades in the just-completed month = no parameter change. Every accepted change records before/after parameters and supporting PF / average-R evidence.
No lookahead: a signal is generated from a day's close and can only become an order for the next available open. Stops and targets use subsequent intraday high/low. If both stop and target are touched on the same daily bar, the stop is assumed to occur first (conservative).
Real security identity: every entry records ticker, signal date, execution date, entry, shares, stop, target, reason and model. Every exit records ticker, exit date, reason, P&L and R.
Persistence: official state is stored in this browser's localStorage. Export a JSON backup regularly, especially before changing browsers or clearing site data.
Market-data feed: the automatic feed supplies real daily OHLCV only. Strategy rules are unchanged. The default broad universe contains 245 liquid U.S. stocks/ETFs and is editable before Day 1. The system does not impose a 40-stock limit. The chosen universe locks with the rest of setup when the first market day is processed.
Ticker maintenance: known symbol changes are canonicalized before Day 1 (BK→BNY, SQ→XYZ), DFS is removed because it is no longer independently listed, and a confirmed permanent 404 may be excluded only before the universe locks. Temporary feed failures still block processing.
Fairness gate: a refresh is committed only when every symbol in the chosen universe has enough warm-up data. Partial refreshes cannot advance the competition, ensuring all 12 models always evaluate the identical opportunity set.
Bug fixes: data/API or calculation defects may be fixed, but strategy changes are not permitted mid-month. Document any code version change outside this tracker.
Backup / Restore / Reset
Reset destroys locally stored official results in this browser after confirmation. Export a backup first if the history matters.