Managed Trading vs In-House Trading: What It Does to Hold
US sportsbooks held about 10% of handle in 2025 on a product whose base vig is 4.5%. Here's how a managed trading service and an in-house desk each change that number, and which one fits your GGR.
US sportsbooks kept roughly 10% of every dollar wagered in 2025, per Legal Sports Report. The standard two-way market they sell, priced at -110 each side, carries a theoretical hold of about 4.5%. The gap between those two numbers is where a sportsbook makes its money, and who controls it, your own desk or a supplier's managed trading service, is the biggest commercial decision in the sportsbook contract.
The question arrives in one form: what's the difference between a platform's managed trading service and running my own traders, and what does that do to margin? Short answer: managed trading gives you a shared price at a cost that scales with revenue, and a hold near the supplier's client average. In-house trading gives you a price nobody else has, at a fixed cost that only beats revenue share once you're big, and a hold that ends up above or below the shared one depending on how good your traders are.
Hold is two businesses stacked on top of each other
Hold isn't one number. It's two products moving in opposite directions.
The straight-bet book (moneylines, spreads, totals) is the liquid, competitively priced part. The overround on a -110/-110 market implies 104.76% total probability, or a theoretical hold of around 4.5% of handle. Realized hold runs below that: sharp money finds the soft side, promotional credits give margin back, competitors undercut visible lines. New Jersey reported a 4.4% hold on non-parlay bets in September 2024; Illinois posted 4.9% on straights across 2023.
The parlay book is where the money is. Every leg carries its own margin, the legs compound, and correlation on same-game parlays is priced with its own cushion. Illinois held 18.2% on parlays in 2023; New Jersey hit 24.2% in that same September 2024 month. Multi-leg bet builders reportedly run 20-35% theoretical hold depending on leg count. That's why New Jersey's annual hold climbed to 9.65% in 2025 from 8.54% a year earlier, per iGamingBusiness, while the vig on a spread didn't move.
So a "record hold" headline usually means parlay mix grew faster than straight-bet margin fell.
Why straight-bet hold keeps compressing
Four forces squeeze the core product at once, and none of them reverse.
Sharper markets. Feeds are faster, models are better, and syndicates move on stale prices in seconds. The window where a book carries a mispriced line has collapsed from minutes to fractions of a second, and less mispricing means less margin you can carry without bleeding to sharps.
Promotional leakage. Odds boosts, profit boosts and no-sweat bets are hold handed back, and promo intensity rises with acquisition costs.
Price competition. In mature markets the marginal bettor shops lines, so books shave the vig on NFL sides and totals to keep the volume that matters. NFL single-game hold reportedly fell about 73 basis points year over year in the most recent full season, an approximate trade-press figure. The NFL is the canary: the most liquid, most modeled, most competitively priced US market, so compression there leads everything behind it.
In-play share. Roughly 47% of global online wagers were placed in-play in 2024, and in mature markets the share has reportedly passed 50%. Genius Sports says 59% of bets by its BetVision users in 2024 were in-play (vendor-reported). Live markets carry wider overround than pre-match main lines, but the most liquid ones are sharply priced and expensive to run: 24/7 coverage, feed redundancy, suspension logic. Video adds to that bill. Watch-and-bet drives in-play turnover, but streaming rights for a serious multi-market book reportedly run into six or seven figures a year, and a stream that beats your odds feed is a courtsiding machine you built yourself. In-play is a growth engine, not a margin cushion.
Add it up: the part of the book a desk prices most carefully is the part where margin is thinnest.
What managed trading does to your hold
A managed trading service means a supplier's desk compiles the odds, moves the lines, sets liability limits and runs the risk for your book, and you steer through configuration. In the iGamingHub catalog, Sportradar is the clearest example of the positioning: its card describes operators outsourcing full trading risk on its Oren platform, on a hybrid revenue model, with a 16-32 week launch window. Kambi sells a full turnkey, API-first sportsbook on revenue share, headless-capable, licensed across MGA, SGA, Alderney, Romania, Colombia and Argentina, with 12-28 weeks to launch. The two are set side by side in Kambi vs Sportradar.
The hold consequences are structural, not a matter of supplier quality.
Your straight-bet price is shared. Every client of the desk shows roughly the same line, so you can't win volume by being sharper than the operator next door on the same supplier. You can set target overround by sport and tier within the supplier's policy, but the probability estimate underneath isn't yours to change.
Your parlay book is stronger than you could build. Correlation pricing across thousands of events is where a shared desk earns its keep: a startup on managed trading gets same-game parlay depth on day one, while one building it in-house gets it in year two, if at all. Since parlays are where blended hold above 8% comes from, this is usually the bigger effect.
Your cost scales with GGR. Revenue share means a bad month costs less and a great month costs more. Altenar (fixed fee, 6-12 week launch, Ontario among its licences) and Digitain (fixed fee, 10-24 weeks, 150+ operator clients on its card) show the other pricing shape, where trading is a line item rather than a slice. BetConstruct runs a hybrid model around an in-house sportsbook with AI-driven real-time risk management, per its card (compared in BetConstruct vs Digitain). GR8 Tech, hybrid, 8-16 weeks, positions its stack as tested at $1B+ GGR scale; Amelco, fixed fee, headless, is built for New Jersey, Pennsylvania, Colorado and Indiana. None of these are endorsements, just the pricing shapes on offer.
Net effect, typically: blended hold lands close to the supplier's client average, the 8-10% region for a parlay-heavy US book and lower for a European straight-heavy one. The floor is high, the ceiling is capped.
What in-house trading does to your hold
An in-house desk means you employ compilers, in-play traders and risk analysts across time zones, license the raw data and pricing models yourself, and own margin strategy end to end. Typical scale: 15-40+ people for a multi-sport, multi-market book, and 12-24 months to competence (editorial estimates, not catalog data).
The margin upside comes from three places. Differentiated prices: if your model beats the shared desk on a sport your customers care about, you win sharp-adjacent volume at a vig the shared desk can't carry. Product control: your own boosts, novelty markets and cash-out margin, a second bite at the overround on every bet settled early. Factoring and limits set to your own book rather than a supplier's policy. Together those can lift straight-bet hold tens of basis points above the shared line, real money on NFL-sized handle.
The downside is symmetrical. A mediocre in-house desk holds less than the shared one, because sharps find it faster than it finds them, and it costs the same either way. Several tier-1 US operators migrated off managed trading because at their scale fixed cost beat revenue share; almost nobody below tier 1 has followed, and those who tried mostly spent two years rebuilding a parlay product while blended hold went backwards.
The break-even heuristic: below roughly EUR50M annual GGR, the fixed cost of a credible desk rarely beats a revenue share, and the parlay gap alone usually outweighs any straight-bet gain. That threshold is approximate and depends on sport mix.
| Model | Cost structure | Typical hold impact | Who it suits (typical) |
|---|---|---|---|
| Managed trading service | Revenue share or hybrid on GGR, bundled into the platform fee; low upfront | Blended hold near the supplier's client average; strong parlay hold from day one; straight-bet price shared with other clients | Startups and mid-size multi-market operators; anyone whose edge isn't pricing |
| Hybrid (supplier desk plus own margin configuration and a 3-6 person risk team) | Revenue share plus a small fixed payroll | Supplier parlay depth kept; own overround and limits per sport; tens of bps of straight-bet hold recovered on core sports | Mid-size books with one or two core sports they know better than the supplier does |
| In-house desk | Fixed: 15-40+ staff, data rights, tooling; 12-24 months to competence | Unique prices and products; straight-bet hold beats or trails the shared line depending on desk quality; parlay hold takes years to match | Tier-1 operators and trading-led brands above roughly EUR50M GGR |
What this means for operators
Report straight-bet hold and parlay hold separately, monthly, by sport. That's the report that tells you whether your trading model is paying off, not the blended headline.
Under the GGR threshold, take managed trading and spend the money on acquisition and product. The parlay book you get on day one is worth more than any straight-bet edge you'll build in year one; how to launch a sportsbook shows where trading sits in the build order relative to platform, payments and licensing.
Mid-size with a real thesis about one or two sports? The hybrid row is usually right: keep the supplier's desk for the long tail and parlays, and staff a small risk team to set overround and factoring where you know your customers better than the supplier does.
Only bring the whole desk in-house when fixed costs beat revenue share at your scale and you can say, in one sentence, why your prices will differ from everyone else's. If the sentence is "we'll be sharper", you don't have a thesis. Operators running casino alongside sportsbook should also weigh how much sportsbook margin matters to group GGR at all, covered in casino and sportsbook convergence.
Methodology and sources
US hold figures come from state regulator filings as aggregated by Legal Sports Report and iGamingBusiness; the September 2024 New Jersey split is from the state's monthly revenue report, and the American Gaming Association's tracker is the reference for headline revenue. The 73 basis point NFL figure is an approximate trade-press number, not independently reconciled here. In-play share is an industry estimate; the 59% BetVision figure is Genius Sports' own report. Vendor facts (revenue model, launch windows, licences, positioning) come from each supplier's card in the iGamingHub catalog as of September 2026. Team sizes, the EUR50M threshold and the hold ranges in the table are editorial estimates, marked typical. Nothing here is a rating.