Run It Twice in Poker: Same EV, Half the Variance (Computed)
Running it twice means dealing the remaining board cards two separate times after an all-in, with half the pot awarded on each runout. It does not change anyone's expected value — we show the exact numbers below — but it cuts the variance of the outcome roughly in half (51.2% reduction in a standard flush-draw all-in, computed exactly). Both players must agree, and it's only done when someone is all-in.
How running it twice works
After an all-in with cards still to come, the players agree to run the remaining streets twice:
- •All-in on the flop: two separate turn-and-river pairs are dealt. First runout decides half the pot, second decides the other half.
- •All-in on the turn: two rivers are dealt, half the pot each.
- •Cards are not reshuffled between runs — the second board comes from the same remaining deck, so the two runouts share no cards.
- •Everyone must agree (in most rooms), and the option typically exists only in cash games — tournaments almost never allow it.
Proof the EV doesn't change
Take the classic spot: you're all-in on the turn with a flush draw — 9 outs among 44 unseen cards, so one run wins 9/44 = 20.45% of the pot on average. Run it twice and enumerate every case exactly: you win both runs 3.81% of the time (9/44 x 8/43), split one-each 33.30%, and lose both 62.90%. Your expected pot share: 1.0 x 3.81% + 0.5 x 33.30% + 0 x 62.90% = 20.45% — identical to one run. This isn't a coincidence of the example: before any river is revealed, the second run's card is just as likely to be any unseen card as the first run's. Each runout has the same win probability, and halving the pot across two identical-EV runouts leaves the total unchanged. Removing an out on run one hurts run two exactly as often as removing a blank helps it.
What does change: variance, computed
Same flush-draw spot, measuring the spread of your pot share: one run has variance 0.163 (you get 100% or 0%). Two runs: variance 0.079 — 48.8% of the single-run figure, a 51.2% reduction. Intuitively, the 33.3% of outcomes that split one-each collapse toward the average instead of landing on an extreme. The mild negative correlation between runs (your outs can't repeat) makes the reduction slightly better than a true half. In bankroll terms this is free variance reduction: our downswing simulation shows a 5bb/100 winner's median worst downswing is 23.2 buy-ins per 100k hands — trimming all-in variance is one of the few levers that shrinks those swings without changing your strategy.
Should you run it twice?
If your bankroll is anywhere near the minimum for your stake: yes, always, and it costs you nothing in EV. Decline only if you have a specific reason to prefer maximum variance — you're massively bankrolled and want to end sessions faster, or you're deliberately gambling with a huge equity edge and want opponents to keep offering it (some pros always run it once precisely to advertise action). One honest psychological note: splitting pots feels like losing to some players. The math says a guaranteed-EV variance cut is a gift; take it.
Related tools
Frequently asked
Does running it twice change the odds?
No. Each runout is drawn from the same remaining deck with the same probabilities, so your expected share of the pot is mathematically identical whether you run it once, twice, or ten times. What changes is the spread: two runs cut outcome variance roughly in half (48.8% of single-run variance in our computed flush-draw example).
Who decides whether to run it twice?
House rules vary, but the standard is that all players in the all-in must agree. Some rooms let the at-risk player decide; online rooms that support it (it's a cash-game feature) usually require everyone to have the option enabled. It's agreed per hand, after the all-in.
Can you run it three or four times?
Where allowed, yes — each run gets an equal fraction of the pot, and the EV stays identical for the same reason. Variance keeps dropping with more runs, with diminishing returns: the big step is from one run to two.
Is there extra rake for running it twice?
No — the pot is the same size and raked once, then split by runouts. Since rake is a function of the pot, running it twice is EV-neutral after rake too.
What happens if each player wins one board?
The pot is split in half — in our flush-draw example this happens 33.3% of the time. That's the whole variance-reduction mechanism: a third of former coin-flip outcomes become ties, pulling results toward the long-run average.
Why do high-stakes pros run it twice?
Because at nosebleed stakes, single pots are huge relative to even large bankrolls, and variance — not EV — is what breaks players. Running it twice lets both players keep playing big pots with less risk of ruin. The famous exception is players who run it once to advertise that they gamble, hoping to get more action later.
Terms used in this article
Keep reading
New to poker? Take the free 15-minute crash course →