No strategy promoted. Faster scalp increased turnover, but lost more net. Hold-only is a research hypothesis, not a verified edge.
The research archive could not be loaded. Your current replays are still available.
Mean net return—Not a portfolio return
Mean turnover—Entry + exit notional
Mean net P&L—After fees and funding
Worst marked drawdown—Worst selected case
No cases match these filters
Choose another market or reset the filters to the matched SOL study.
Does more turnover earn more?
Each point is one simulated account. Select a point or case to inspect it.
Use a point’s keyboard focus or pointer to see its measurements.
Strategy comparison for the selected evidence
Variant
Cases
Net return, mean
Turnover, mean
Net / $10k turnover
Trades / day, mean
Inspect a replay
What closed the trades?
Signal decisions by bar
Modeled trades; holds use bar timestamps, not fill-clock timing. Entry allocation uses $800 reference capital, not measured pre-entry equity.
Side
Entry UTC
Hold
Entry notional
% of $800
Exit reason
Net P&L
What do costs do to this case?
Reprice the same fixed trades. No new entries, exits, fills or strategy backtest. Existing modeled slippage stays in gross P&L.
Use finite, non-negative values within the input ranges.
Your next hypothesis
Notes stay in this browser. They are never sent to Cloudflare or the bot.
Saved locally as you type.
Options are not a P&L line yet
20 brief public-chain policy cases. Zero executed matched spreads. Entries failed the volume gate. No 48h historical chain, paired venue fills or realized options profit has been validated. Candles alone cannot fill that gap.
Historical stress archive
All 3,000 recorded runs. Different policy from today’s submission; overlapping proxy periods, not independent samples.
All recorded close fills. Older campaigns do not record direction or entry timing; those fields are shown as unavailable, never inferred.
Side
Exit UTC
Entry price
Exit price
Contracts
Reason
Net P&L
Signal confusion matrix
Whole selected campaign, not the filtered subset. Rows = forward market class; columns = policy signal. Nominal 6h, ±0.5%; 4h grids resolve to 4h. Forward outcomes are diagnostic labels, not inputs or trade returns.
Inspect the market inputs
Full sanitized OHLCV for all historical stress and fresh evaluation datasets. The fresh files include 15h of warm-up before the 38h10m evaluated segment. Data records can overlap across sources and timeframes.
Charts aggregate to at most 300 time buckets: price is each bucket’s last close; volume is the bucket sum. Full OHLCV arrays are in the download. No historical option chain, Greeks, GEX or order-book depth is fabricated.
An inventory of research artifacts, not a single performance leaderboard. Repeated measurements, old profiles and synthetic fixtures stay visible and separate.
Sanitized analytical reports; matching measurements are flagged, not counted as new evidence
Excluded from edge claims. Known look-ahead bias: a feature uses shift(-1). Different older simulator, no verified fill lifecycle. These recorded results must not be treated as achievable returns.