Blog · 13 posts · 93 min end to end

Measured,
not explained.

Build logs and measurements from the desk that writes the software. Numbers we ran ourselves, including the ones that came out against us.

Build log17 Sept 20267 min

v0.25.1: The Binance aggTrades Historical Limit, Measured

A user opened the app mid-morning and every candle since 00:00 UTC was hollow: price there, volume there, no bid × ask. Not a feed outage — a regression dated to 0.22.0, where every import call passed the same window and only ever asked for the seven candles you could see. This is what the Binance aggTrades historical limit really constrains, what one hour of BTCUSDT Spot bid × ask physically costs to rebuild over public REST, and why we fill the newest hour first, one hour per call, and flag what is still missing instead of smoothing it away.

Read it updated September 2026

Everything else

Build log09 Sept 20267 minv0.15.0: 26 of Our 27 Backtests Were Identical — and the Optimiser Ranked Them AnywayOur parameter sweep was reading keys the strategy never used. 27 runs, 26 identical, a winner picked by rounding — and a green test suite throughout.
Build log08 Sept 20266 minv0.14.0: The Order-Book Recorder Works — and Calls 100% of Its Orders CancelledOur recorder captured 5,711 order-book events and zero fills. Rithmic does not put executions on the MBO channel — and our green tests were checking a shape that never ships.
Measured27 Jul 20266 minThe Hurst Exponent Reads a Coin Toss as a Trend — We Measured By How MuchUncorrected R/S calls a memoryless series a trender at every length we tested. The correction, the dispersion it cannot fix, and the sentinel bug an adversarial review caught.
Build log23 Jul 20265 minv0.7.6: A Home Dashboard, an Agent Named Sen, and the Live Bar on Its Own ScreenThe home screen becomes a dashboard, the assistant gets a name and a hard limit, and the forming bar moves to its own monitor.
Build log21 Jul 20265 minv0.7.4: Wiring the Full Order Book (MBO), and a Strategy That Became an IndicatorFrom ~15 visible levels to the whole book: the MBO rebuild, a heatmap that finally shows two sides, and a mean-reversion signals indicator.
Microstructure21 Jul 20265 minMicro-Price vs Mid-Price: The Fair Value Your Chart HidesThe mid ignores the queues; fair value does not. The published one-tick AUC, why it is physics rather than alpha, and the 16 ms plumbing that ships it.
Microstructure21 Jul 20266 minOFI vs CVD: What the Tape Cannot See, MeasuredThe 87/33 gap between the book and the tape, the negative R² nobody quotes, and the config that ships.
Build log18 Jul 20268 minOrder Flow on macOS and Linux: Porting a Windows Footprint Platform in 158 Lines12 files, +158 lines, 0 lines of Rust. The port was never the hard part — the titlebar, Gatekeeper and a 84 MB AppImage were.
Broker plumbing15 Jul 20268 minWhy Your Footprint Shows a Different Volume Than Your Broker (and Which Number to Trust)Four platforms, four volumes, same contract, same second — none of them wrong. Which one your order flow should trust.
Pillarpillar14 Jul 202614 minOrder Flow on a Prop-Firm Account: What Actually Works on Apex, Topstep and 4PropTraderThe rules, the feeds and the drawdown maths nobody writes about because they trade their own money.
Broker plumbing14 Jul 20269 minRithmic Historical Data: What You Can Actually Backfill, and Why Your Chart Is Emptyrp_code 13, tick replay vs volume profile, and what each prop firm actually pays for.
Compliance14 Jul 20267 minFrom Read-Only Charts to Order Entry: Our Execution ControlsOur earlier read-only design and its revision.

Measured,not explained.

v0.25.1: The Binance aggTrades Historical Limit, Measured

Measured,
not explained.