Case study · Market making · November 2026 · Vandergrid

Tighter spreads and deeper books on three exchanges

Since we took over quoting for a mid-cap infrastructure token, its touch spread has tightened by 59% on one exchange and 49% on another, and resting depth within 2% of mid has risen between 101% and 377% across all three venues. Spreads were already tight on the third exchange, so depth was the gap there.

−59%

touch spread, Exchange A

−49%

touch spread, Exchange B

up to 4.8x

resting depth

3 venues

client-funded

Starting point

Wide spreads and thin books on all three venues.

Before we began quoting, the median touch spread was 81.7 bps on Exchange A and 109.8 bps on Exchange B. Depth within 1% of mid was a few hundred dollars or less per side everywhere.

Exhibit 1The improvement came in stages, not in one step

Daily median touch spread per venue, from every captured book snapshot. Phases shaded by quoting model; the dashed line is a model change. Source: our own order book capture.

What we changed

Client-funded quoting for a retainer, with no token loan and no options.

The client funds the trading accounts and keeps custody and P&L; we run the quoting. Spread roughly halved on the two venues that had room, in two steps rather than one.

Exhibit 2Spread roughly halved on two venues

Median touch spread, before window against the current quoting model. Within-day medians, then the median across days. Source: our own order book capture.

Depth, not spread

Exchange C was already tight at 39.8 bps, so depth was the gap there.

Its bid depth rose 255% and its ask depth 377%. Every side of every book is at least double what it was.

Depth is shown as an index because the token’s price fell over the period, so dollar depth understates what was added. The index is built on token sums with the before window set to 100.

Exhibit 3Resting depth at least doubled on every side of every book

Depth within 2% of mid, token sums, indexed to the before window = 100. Source: our own order book capture.

The independent check

A second collector, different cadence, agrees within 5% on all three.

Every captured order book snapshot in each window, from our FlowTrack surveillance capture. Each statistic is formed within the day first and then the median is taken across days, so uneven capture rates cannot bias the result. We cross-checked the before window against a second, independent capture that used a different collector and cadence: the two agree on touch spread to within 5% on all three exchanges.

Exhibit 4An independent capture agrees within 5%

Before-window median touch spread from our capture against a second collector with a different cadence, whose window ends before our first fill. Source: both captures.

No volume figure

Our own fills are a large share of the reported tape, so volume would describe us.

We report book quality instead, meaning spread and depth, because that is what any other trader experiences.

Reference data

Before windows: May to June 2026 on Exchanges A and B, May to early September 2026 on Exchange C. "Now" is our current quoting model, measured to late October 2026.

	Exchange AA	Exchange BB	Exchange CC
Median touch spread, before → now	81.7 → 33.1 bps (−59%)	109.8 → 56.0 bps (−49%)	39.8 → 39.4 bps (unchanged)
Spread per unit of daily volatility	−52%	−15%	−17%
Bid depth within 2% (in tokens)	+233%	+101%	+255%
Ask depth within 2% (in tokens)	+213%	+347%	+377%

Depth is measured in tokens rather than dollars because the token’s price fell over the period, so dollar depth understates what was added.