HRT takes retail execution-quality crown in slower August market

Hudson River Trading (HRT) and Susquehanna (SIG) won the August price improvement race. This may be the symptom of the securities, order type, and order size they choose to favourably market for retail brokers, Global Trading analysis reveals.

US equity retail volumes cooled by about 11% month-on-month in August according to our data from Bmll Lab, and aggregate price improvement (PI) delivered by market makers fell to US$382m from US$447m in July (-14.5%), a touch below this year’s monthly average of US$388m (-1.6%).
On execution quality (E/Q), Hudson River Trading (HRT) posted the lowest share-weighted median E/Q at 0.315, with Susquehanna (SIG) next at 0.335, both clear of their competitive field despite broad, visible shifts in the shape of every market maker’s E/Q distribution.

Substantial price improvement versus the national best bid-offer (NBBO) are disclosed in 605 forms: Citadel Securities delivered US$130.9m on 251.5bn shares (-10% MoM) , Virtu US$86.2m on 162 bn shares (-8% MoM), SIG US$65.4m (-15% MoM), Hudson River Trading US$57.5m (-10%), with Jane Street US$26.3m (-32%) and Two Sigma US$11.3m (-13% MoM).
In August, volumes for our lit proxy printed 1.23 trillion shares versus 1.39 trillion shares (-11% MoM). Trades’ E/Q aggregate for our market proxy delivered US$632m versus US$707m improvement in July equally, down 10% MoM.

Looking at our 0 to 2 E/Q (the realised spread versus prevailing spread ratio) distributions of price improvements or volume traded, which we build for each market maker, and comparing the shape of E/Q density distributions, it is apparent why SIG and HRT achieve best execution for retail traders on aggregate: Most of their volume sits in very tight bins with E/Q below 0.4 and the their median E/Q is also the lowest.

HRT improved at the margin in August and became the best market maker to trade against with on average for retail; Its median E/Q edged down from 0.344 to 0.315  and the share of ticker / order size / order type E/Q prints with execution quality better than 0.4, where retail trades at better than 40% of the distance between the NBBO and the mid price, rose to 66.6% (from 64.0% in August).

The mode, which is the value that appears the most in the distribution stayed low with a peak around 0.275. This signals consistency rather than just statistical mix artefact, where one type of trade or securities filles would skew the whole E/Q distribution, for market orders, which are most orders that they fill. HRT’s PI of US$57.5m fell in line with volumes (-16.6% MoM, not because execution quality slipped.

SIG became a close second on median quality but saw its distribution shift right: its median E/Q went from 0.295 to 0.335 and the sub-0.40 E/Q share fell to 52.4% from 61.9%. The peak migrated from 0.285 to 0.365 pointing to slightly wider realized spreads for retail trading against it. PI landed at US$65.4m (-14.9% MoM), roughly in line with the market-wide volume pullback.

Citadel Securities provided retail traders with the month’s largest volume of PI, as the biggest market maker, at US$130.9m (-13.9% MoM) but showed the most pronounced execution quality deterioration, as shown by the right shift in its E/Q distribution; Its median E/Q went from 0.405 to 0.515. The better than 0.4 E/Q-share of execution also fell from 47.6% to 27.5%. The distribution’s peak moved from 0.285 to 0.485, indicating, maybe, a heavier mix in names/contexts where the realised spread is inherently larger.

Virtu median E/Q also degraded in August to 0.515 from 0.444 in July. PI delivered to retail were US$86.2m (-14.0% MoM).

In very liquid, directional exchange traded funds (ETFs), market makers’ E/Q tends to converge. For example, July’s triple levered Nasdaq ETFs (TQQQ) and its reverse triple short (SQQQ) were traded by most market makers with low E/Q dispersion (standard deviation of E/Q at 0.11), and large-cap stocks like Apple or Tesla also showed tight clustering. Most market makers have the same colours in these securities on our heat map reflecting similar execution quality.

In more idiosyncratic or leveraged exposure securities, E/Q diverges. July’s silver trust (PSLV) and single-name outliers like Lucid group (LCID) showed much wider E/Q dispersion across makers (standard deviation of E/Q up to 0.25–0.39). That’s probably where skill, inventory, and microstructure tactics bite and where mix changes can move a firm’s aggregate E/Q even when its process quality is steady.

In August’s, the top 10 retail traded ETFs were dominated by levered ETFs such as the triple semiconductor ETF and inverse, the triple Nasdaq and inverse ETFs, the double and inverse Tesla ETFs, or more esoteric ones like the ultra income ETF ULTY that sells weekly covered call. For these, order-type filled by market makers diverged in ways that line up with the E/Q differences per market maker. SIG received and executed the highest share as market orders (65% on ETFs), with Citadel Securities and HRT in the mid-50s and Two Sigma a touch lower; Jane Street stood out for the largest marketable-limit share filled (≈50%) and the highest inside-quote filled among peers.

Where the ETFs are most liquid and directional, like the triple Nasdaqs, or semiconductor ETFS, E/Qs are fairly clustered, but Jane Street’s more limit/inside mix seem to correlate with its median E/Q to keep its execution quality looking worse while market-heavy routing shows up as greener, that is a better E/Q. HRT’s best in class median-level E/Q is consistent with this profile—even though its top10 ETF set is smaller in our per-ticker files, its book keeps a larger share of executions in sub-0.40 E/Q.

The top 10 retail traded stocks (Opendoor, Nvidia, Tesla, Intel, Palantir, as well as penny stocks like Tilray, Denison Mines et al.) tell a similar August story. In the deepest names (INTC, AAL, NVDA), makers’ E/Q is tightly grouped, and differences are small; in idiosyncratic names (OPEN, PLTR, TLRY) dispersion widens and order-type mix seem to qualify execution quality.

Susquehanna again shows the highest of at market routed executed stocks (≈80% on stocks), Hudson River is next, while Jane Street carries a higher marketable-limit and inside-quote proportion which seems to lead to worse E/Q in these names. *

In short, execution quality rankings seem at least driven, in part, by which names each market makers leaned into and even more how much of that flow they executed as market vs. marketable-limit orders or outside the quote orders. Jane Street the largest ETF market maker worldwide is unlikely to be offering the worst execution quality for them, rather retail broker route to them specific orders (marketable limit, outside of quote) that skew the execution quality average picture.