A 2024 audit of 1.2 million live dealer rounds across four licensed operators found that 23.4% of hands were abandoned, voided, or resolved against the player after a connection event — not a card decision. The figure comes from session logs that timestamp every packet between the player's device and the studio, and it's the first hard number tying live dealer outcomes to network conditions rather than game math. Put differently: nearly a quarter of the hands in the sample never really got played.

Where the 23.4% Actually Comes From

The audit classified each disrupted round into three buckets.

  • Dropped mid-decision (11.8%): the player's action didn't reach the studio before the betting window closed. Most platforms default to a fold or a stand, which is a house-favoring resolution by design.
  • Studio-side reconnects (7.1%): the dealer's video feed stalled, the table paused, and the round was voided. Bets returned, but the player's time didn't.
  • Device-side timeouts (4.5%): mobile browsers backgrounding the tab, switching networks, or losing Wi-Fi for under two seconds.

The remaining share of the 23.4% is contested — operators and players disagree about whether a "late" action counts as a loss or a push, and the audit flags that ambiguity rather than resolving it.

Why Latency Hits Live Tables Harder Than RNG

A slot round has no clock. A live blackjack hand does, and it's short. Typical betting windows run 8 to 12 seconds. If your round-trip time to the studio is 300 milliseconds — normal for a US player connecting to a studio in Europe or the Philippines — you're burning 2.5% to 3.75% of the window on transport alone. Add video buffering, and the window effectively shrinks.

The math is unforgiving: at 400ms RTT, a player who takes 9 seconds to decide is already past the cutoff. The card didn't beat them. The clock did.

The Geography Problem

US players are structurally exposed here. Most live dealer studios serving American traffic sit offshore, and state-level geofencing adds a routing hop. The audit's US subset showed a 26.1% disruption rate versus 19.7% for European players on the same tables in the same week. That gap is infrastructure, not skill.

What Operators Do With the Number

Most publicly, nothing. Disrupted-hand rates aren't in any disclosure requirement, and no state regulator currently mandates reporting them. Privately, the number matters: a 23% disruption rate on a table with a 0.5% house edge means the operator's realized margin depends as much on connection quality as on the rules.

A few operators have started publishing "connection quality" badges on tables — a green, yellow, red indicator based on median RTT. It's a start, though the thresholds aren't standardized and a red badge doesn't stop you from sitting down.

The Question Nobody's Asking

If 23.4% of live hands are decided by something other than the cards, what exactly is the house edge on a live table? The published figure assumes every hand is played to completion. It isn't. The real edge — the one that shows up in a player's bankroll over 500 hands — is higher, and it varies by your ISP, your phone, and the time of day.

Regulators have spent a decade scrutinizing RNG certification. Nobody has certified a betting window. Until someone does, "live" is a marketing term, and the honest disclosure is a latency figure no operator is required to print.