Instrument book

built 2026-08-01 03:14 UTC · window 2024-07-29 .. 2026-07-29 · 1-minute bars only · agent control · matrix · all outputs
The book is a decision. The days column is a measurement, taken from NinjaTrader's own day files every time this page rebuilds — so a decision can never quietly outlive the data behind it.
6
testing now
11
to add
0
candidate
655
best coverage (days)
contractmicrotestedstatusdaysnewest barnotes
US index
The core. Deep books, tight spreads, and a real session open - which matters because every playbook here is session-anchored (pre-session high/low, killzones, the sweep into the New York open).
NQ
E-mini Nasdaq 100
MNQNQ● TESTING6472026-07-31
ES
E-mini S&P 500
MESES● TESTING6472026-07-31
YM
E-mini Dow
MYMYM● TESTING6382026-07-31
RTY
E-mini Russell 2000
M2KRTY◆ ADD6472026-07-31The obvious gap. Small-cap, higher beta, and it genuinely diverges from the other three - a fourth market rather than a fourth look at the same tape.
Metals
Different drivers from equity indices, and gold already produces the farm's most active cells.
GC
Gold
MGCGC● TESTING6462026-07-31
SI
Silver
SILSI◆ ADD6472026-07-31Was excluded as 'not subscribed on this feed'. That was measured while the downloader was broken; it pulled 16.9 MB of 1-minute history on 2026-07-29. Never unsubscribed - simply never fetched.
HG
Copper
MHGMHG◆ ADD——Industrial demand rather than risk sentiment, and it trends. Nothing else in the book behaves like it.
Energy
Its own supply-driven character, and CL has the session structure the playbooks assume.
CL
WTI Crude
MCLCL● TESTING6472026-07-31
NG
Natural Gas
—NG◆ ADD6462026-07-31Same false exclusion as silver - 15.1 MB pulled on 2026-07-29. High volatility; expect the parameter sweep to matter far more here than on the indices.
FX
Four majors, agreed 2026-07-29. Futures rather than spot: same platform, same cost model, and a settlement that gives session logic something to anchor to. EXPECT WEAK RESULTS until the session parameters are adapted - FX runs 24/5 with no open, and every playbook here is built around a market that opens. 6E and M6E have already produced the weakest cells on the board (Riley 6E 5m: net -812, PF 0.869).
6E
Euro FX
M6E6E● TESTING6492026-07-31
6B
British Pound
M6B6B◆ ADD6532026-07-31
6A
Australian Dollar
M6A6A◆ ADD6552026-07-31
6J
Japanese Yen
MJY6J◆ ADD——Completes the four majors. Carries genuine Bank of Japan intervention risk - a tail no backtest contains. Size for it rather than avoid it.
Crypto
The best fit in the book for ICT-derived logic - sweeps, fair value gaps and stop hunts describe a retail-heavy, leverage-heavy market, which is crypto's whole personality. CME futures rather than spot, deliberately: spot runs 24/7 with no open at all, while the futures carry a daily settlement and respond to the New York session.
MBT
Micro Bitcoin
—MBT◆ ADD4762026-06-13Micro first and mean it: being wrong on a full BTC contract costs a hundred times more than being wrong on MBT for exactly the same information. WATCH THE EXPIRY - the platform is showing 07-26, which expires within days.
MET
Micro Ether
—MET◆ ADD——
Rates
A genuinely different regime driver, and completes the macro picture.
ZN
10-Year T-Note
—ZN◆ ADD6522026-07-31Third of the falsely excluded nine - 21.1 MB pulled on 2026-07-29, the largest of the set.
Europe
A different region and a different session - the only group here that is not anchored to the New York open. Requested by David 2026-07-29.
FDAX
DAX
FDXSFDXS◆ ADD4662026-06-13IN, but it CANNOT be tested as configured. Every playbook in this farm anchors on the New York open - pre-session high/low, killzones, the sweep into 09:30 NY. DAX opens in Frankfurt. Running it unchanged produces a false negative, not a result, and this farm has already lost eight weeks to an absence read as a finding. The session parameters must be moved to Frankfurt hours FIRST, and that is a change to what the playbook means, so it is David's call. Stale data already on PC4 from mid-June.
One contract per underlying. A micro and its full-size sibling are the same underlying, the same bars and the same signals — only the size differs. Testing both measures one thing twice and, worse, inflates the trial count that the deflated Sharpe correction depends on, which assumes trials are independent. On 2026-07-29 the matrix held five such pairs, so roughly 150 of its 330 cells were duplicates.

The days column is why nine markets were missing. test_matrix.py excluded RTY, SI, NG, ZN, 6B, 6A, M2K, MCL and SIL because they “returned ZERO bars — not subscribed on this feed”. They were subscribed all along; the downloader had been failing silently for weeks, so nobody had fetched them. Every one pulled 15–21 MB the moment it was fixed.

Deliberately excluded

M6A/M6B/MICD/MISFMicro FX duplicates of the majors above, or minor crosses. One contract per underlying.
FSXEMicro EuroStoxx 50 - duplicates the DAX European exposure.
EURCHF / EURGBPRange-bound grinders, poor material for breakout and fair-value-gap logic.
MMC 03-25 / MSC 03-25Contracts expired March 2025, more than a year ago.
spot FX / spot cryptoA second feed and a second execution path. The farm has three machines and one of them is producing; widening the surface area before that is fixed buys risk, not coverage.