PLO variance and downswings

A 5 bb/100 winner in PLO 6-max (SD 140) should expect a worst downswing around 34 buy-ins over the next 100,000 hands, and 1 run in 20 goes deeper than 67. The same winner is still losing after 100k hands 13% of the time. Over 500,000 hands the median worst stretch reaches 59 buy-ins. A bankroll of 57 buy-ins keeps the bust risk of a million-hand run near 5% (89 buy-ins for 1%). Smaller win rates need far more: every table below spans 2-8 bb/100 and SD 120-160, so read your own cell, then run your exact numbers in the calculator.

SD columns follow the calculator's game presets: 120 = PLO full ring, 140 = PLO 6-max, 160 = PLO heads-up. The win rate barely moves the depth at this horizon; the SD owns it.

More volume means deeper worst stretches, not shallower: the longer you play, the worse your worst run gets. Volume buys income, not safety.

A 2 bb/100 winner at SD 140 spends a median of 196k hands of a 500k sample below a previous peak, and is still down money after 100k hands 33% of the time. That is what a real, positive win rate feels like from inside.

Cell format: 5% risk · 1% risk. This settles the forum argument honestly: the popular "50 buy-ins" runs a little light for a 5 bb/100 PLO 6-max winner at 5% risk (57 buy-ins), badly light for a 2 bb/100 winner (135), and comfortable for an 8 bb/100 one (36). The bigger danger is the win-rate assumption itself: bankroll requirements scale like 1 over the win rate, so being half as good as you think doubles every number here. The calculator's Bayesian tab exists for exactly that doubt.

Hold skill constant at 5 bb/100 and switch games. The NLHE 6-max model (SD 100) gives a median worst 100k-hand downswing of 22 buy-ins; PLO 6-max (SD 140) gives 34. Over 500k hands: 36 vs 59. Still losing after 100k hands: 6% vs 13%. PLO's reputation for brutal swings is not culture, it is the standard deviation, and it also means PLO bankroll advice copied from NLHE is systematically too small.

A 2 bb/100 winner at SD 140 needs 1.9M hands before the sample alone proves they win at all; even a 5 bb/100 winner needs 301k. Grade your play with review tools, not your graph.

Large steep-payout fields are a different sport: a winning Sunday-major reg is still down after a thousand of them 34% of the time, and the median worst downswing across five thousand runs past 603 buy-ins. MTT bankrolls are counted in hundreds of buy-ins for structural reasons, not cowardice.

Two practical conclusions. First, size your roll for the variance you'll actually face, because the alternative taxes your play: short money makes you avoid thin +EV spots, and that self-inflicted win-rate cut costs more than the variance itself. Second, treat your win-rate estimate as the soft spot in every table here. The depth of your downswings you cannot control; which column you're really in, you can study your way into. When a downswing hits, the question worth asking is never "when does it end" but "am I in the cell I think I'm in" - and that is a review question, not a variance question.

Built by Adam, the developer of SolvePLO. Corrections welcome: contact.

How many buy-ins do I need for PLO?

Depends on the risk you accept, and the honest range is wider than any rule of thumb. From the simulation: a 5 bb/100 winner in PLO 6-max (SD 140) needs about 57 buy-ins so that only 5% of million-hand runs ever go bust, and about 89 for 1%. A 2 bb/100 winner needs 135 and 202. The popular "50 buy-ins" answer corresponds to a real but double-digit bust risk for small winners.

How big do PLO downswings get?

For a 5 bb/100 winner at SD 140, the median worst downswing over 100,000 hands is 34 buy-ins, and 1 run in 20 exceeds 67. Over 500,000 hands the median worst stretch deepens to 59 buy-ins (95th percentile 102). Winning does not exempt you: 13% of that winner's 100k-hand runs end in the red.

How long do downswings last?

Longer than they feel. Over 500,000 hands, the same 5 bb/100 winner's longest below-peak stretch has a median of 107k hands, and 5% of runs contain one longer than 263k hands. A downswing here means the whole time between setting a bankroll peak and beating it.

Is PLO higher variance than NLHE?

Yes, materially. At the same 5 bb/100 win rate, moving from the NLHE 6-max model (SD 100) to PLO 6-max (SD 140) takes the median worst 100k-hand downswing from 22 to 34 buy-ins and the still-losing-after-100k odds from 6% to 13%. Same skill, deeper swings: that is the SD doing it.

What standard deviation should I use for PLO?

Nobody publishes measured population SDs, so treat SD as a modeled input: tracking-software populations put PLO 6-max around 140 bb/100, full ring nearer 120, heads-up 160 and up. Every table on this page spans 120-160 so you can read your own column, and the calculator accepts any value.