i keep this paper close. always. 2007. numbers shift. mental model: permanent. Drepper: CPU isn't the bottleneck - memory is. L1→4 · L2→12 · L3→40 · DRAM→200 cycles the programmer who internalized this writes different data structures. Agree? #hft #lowlatency #tradingtech

97% of limit orders cancel before they ever trade. The dominant event in any order book is not a fill. It's a cancellation. If your signal stack consumes the trade tape and discards the cancel stream, you model the residual. A breakdown: #hft #marketmicrostructure

It wasn’t AI, it was the COVID hiring binge interest rate spike hangover.

checkout our last blog post about Cancel-to-Trade Ratio Watching Cancel-to-Trade Ratio Drift in Real Time: What Batch Reports Miss Before the Flag Arrives. - by @sisSoftware #visualHFT #electronictrading #tca #analytics

90% of buy-side equity desks run TCA. Every last bit of it is post-trade. Coalition Greenwich surveyed 103 traders in 2024. Nine out of ten are measuring execution quality after the damage is done. The market has moved. The fills are locked. The alpha is gone. The report arrives the next morning. Knight Capital lost $460 million in 45 minutes on August 1, 2012. Post-trade analysis documented the loss beautifully. What it did not do — could not do — was interrupt the bleed while it was happening. There was no live order flow signal. There was no volume-informed toxicity measure updating in real time. The desk was flying blind at the worst possible moment. That is the structural gap that post-trade TCA was never designed to close. VPIN — Volume-Synchronized Probability of Informed Trading — updates on the volume clock, not the calendar clock. When informed order flow accelerates, VPIN signals the shift while execution is still in progress. Easley, Lopez de Prado, and O'Hara established the theoretical foundation in 2012 in the Review of Financial Studies. The methodology has been available for over a decade. #VisualHFT operationalizes it with five simultaneous live signals against real exchange feeds: VPIN, multi-level LOB imbalance across 10 depth levels, Market Resilience, and OTT/TTO ratios — all updating live. The CME Group's own 2025 research concluded that order book depth alone is insufficient for assessing real liquidity conditions. That is the exchange operator, not a vendor, stating the case. #FINRA Rule 6152, approved June 2025, makes execution quality data publicly auditable. Execution performance is now market-visible. Desks that cannot explain their intra-session routing decisions face a transparency deficit that post-trade reports cannot resolve retroactively. VisualHFT is open-source, Apache-2.0, with 1,100 GitHub stars, 219 forks, and 511 commits. Every computation is auditable in the source code. No black box. Full breakdown in the article — link in first comment. #VisualHFT #MarketMicrostructure #VPIN #HFT

We just hit 1k stars on the open-source VisualHFT repo. It’s not a big number, sure, but for us, it’s a bit more than that. If you're managing a desk and you feel like you're guessing why your slippage is high, have a look at the repo. Point it at your order book and see the FIX or binary flow for what it actually is. No pitch, just a tool for anyone who wants actually to see what’s happening on the wire. #hft #marketmicrostructure #electronictrading #fintech

We are getting ready to launch... > New website > Sign up for early access #visualHFT

Liquidity could be hidden, but that, does not mean you cannot discovered deep into the microstructure. We are building advanced metrics. #visualhft #markets #microstructure #liquidity

Every finance person needs to watch this episode from season 4 of Industry: