Granular Investment Review
Free editorial articles on transaction cost estimation and trade execution — how models measure slippage, schedule parent orders, and price the gap between a paper trade and a filled one.
Editorial coverage here treats each claim as a question about cost components, not as a verdict about whether machines beat markets.
An execution model does not delete the bid-ask spread, exchange fees, or market impact. It redistributes them across time and child orders. A useful cost estimate names each component — spread, half-spread, temporary impact, permanent impact, opportunity cost, timing risk — and reports which one the model actually tried to reduce. Editorial work here reads the breakdown, not the headline.
Linear-impact, Almgren-Chriss, and square-root rules each embed an assumption about how size moves price. Reading the model means reading the assumption: which regime it was fit on, what venue, what order size, and how stale that calibration is today.
Transaction cost backtests lean on fills that may never have been available. Editorial coverage here asks what was actually fillable, what was reconstructed, and whether the slippage number is measured or assumed.
“A transaction cost estimate is only as honest as the components it is willing to name.”
Editorial position — Granular Investment Review
A repeatable reading order for any model that claims to estimate or improve execution cost.
Reading order
Five steps, applied to any cost estimate that crosses this desk. The point is not to grade the model — it is to see which components it actually addresses.
List spread, fees, market impact, opportunity cost and timing risk before reading any single “total cost” number.
Identify whether impact is linear, square-root, or venue-fitted, and what order-size range the calibration covers.
See whether the model trades off urgency against impact, and what participation rate it assumes for the parent order.
Mark which slippage figures come from recorded fills and which are reconstructed from spread and impact proxies.
Note venue-specific fees, auctions, opening and closing auctions, and latency — the components a clean headline usually omits.
This is an editorial and informational site about AI in investment — specifically the narrow, technical question of how transaction cost is estimated and how trades are executed. It is written for readers who already meet cost models in their work and want a second reading.
No. Granular Investment Review publishes free editorial articles and answers inquiries. Nothing is sold on this site — no software, no signals, no execution, no managed accounts, no paid advice. The “Review” in the name refers to reading, not to reviewing products for sale.
No estimate of a future fill is exact. A cost model gives a conditional expectation under assumptions about impact, spread and participation. The honest output is a range with named components, not a single dollar figure presented as fact.
It can reduce drag, but only on the portion of cost the model actually addresses. Reducing measured slippage on one component can shift cost into another — timing risk, opportunity cost, or a wider market-impact regime. Editorial coverage here treats return claims as conditional, not as promised.
Yes — inquiries are welcome. We may write about the question editorially if it is of general interest, but we do not produce paid analyses, endorsements, or vendor comparisons.