How Long Do Poker Downswings Last?

Downswings are measured in hands, not weeks — and the honest unit changes the answer. For a solid 2.5bb/100 winner, a typical 100,000-hand year contains a deepest downswing of about 27 buy-ins, with the longest stretch below a previous peak running around 43,000 hands. Over a longer career, a 30-buy-in downswing is a when, not an if. The full distributions, by win rate, are below.

That's the answer almost nobody searching this question gets. Ask the top results and you'll be told a downswing can last "a few hands to thousands of hands," or "weeks to months," which is the kind of sentence that is technically true of everything and useful for nothing. It's a shrug with a byline. You didn't type this query because you wanted a range wide enough to hold every player who ever lived. You typed it because you're inside one, and what you're actually asking is narrower and colder: when does this end, and can I outlast it?

So let's do what the shrug never does. Let's compute it.

The Model, Stated Plainly

Numbers first, assumptions on the table — because a downswing number without its assumptions is just a vibe wearing a decimal point.

We modeled a standard online cash game the way every serious variance calculator does: results per 100 hands drawn from a normal distribution, with a mean equal to your true win rate and a standard deviation of 100bb/100 — the defensible figure for 6-max no-limit, where real-world values run roughly 90–100 (higher for heads-up or PLO; if your tracker says 90, your numbers are modestly kinder than these). We ran three true win rates — 1, 2.5, and 5bb/100 — through thousands of simulated careers: 4,000 separate 100,000-hand stretches per win rate, plus 600 full million-hand careers. A downswing here means the real thing: the fall from a high-water mark until that mark is passed again. Depth is how far below the peak you sink; length is how many hands you spend under water before you surface.

Know the model's limits, because they all point the same direction. It assumes your win rate holds constant — that you play your A-game through the entire storm, never tilt, never face tougher lineups, never lose sleep. Real per-hand results also have fatter tails than a normal curve fully captures. Every one of those simplifications flatters you. Real downswings run worse than these tables, not better. Read them as the floor of what variance alone does to a player who never once cracks.

The Table: What a Year of Downswings Actually Looks Like

Here is what 100,000 hands — roughly a serious online year — actually contains, purely from variance, for a genuine winner playing flawlessly the whole way:

True win rate	Median deepest downswing	Nightmare (1-in-20) depth	Median longest stretch below a peak	Chance of a 20-buy-in downswing	Chance of a 30-buy-in downswing
1bb/100	32 buy-ins	63 buy-ins	~54,000 hands	88%	55%
2.5bb/100	27 buy-ins	53 buy-ins	~43,000 hands	79%	40%
5bb/100	22 buy-ins	42 buy-ins	~31,000 hands	61%	20%

Read the middle row, because the middle row is most players. A real 2.5bb/100 winner — a professional edge, rent-paying, no doubt about it — should expect, as the median outcome of an ordinary year, to lose 27 buy-ins from a peak at some point, and to spend a stretch of about 43,000 consecutive hands below a previous high. Four times in ten, the year contains a 30-buy-in hole. Not a bad year. The middle year. The one the coin flips you half the time.

Stretch the horizon to a million hands — a career, or most of one — and the language of probability stops being the right language. The chance of experiencing at least one 30-buy-in downswing somewhere in a million hands: 94% for the 5bb/100 crusher, 99.8% at 2.5bb/100, and effectively certain at 1bb/100. At 2.5bb/100, a 30-buy-in downswing is not a tail risk you insure against. It is a scheduled stop. You don't know the date; you know the visit.

And the duration of that visit, peak to full recovery? In the million-hand simulations, a downswing that reaches 30 buy-ins deep takes a median of about 100,000 hands to fully resolve at 2.5bb/100 — and about one such downswing in nine still hadn't found a new peak within the simulation's window at all, so the true median is slightly worse than that. At 5bb/100 the median recovery is around 73,000 hands. At 1bb/100 it's roughly 136,000 hands, with a fifth still unresolved. That is what "it'll turn around" costs in the only unit that counts.

Now convert the unit, because this is where the number becomes a life. An online grinder putting in 40,000 hands a month walks a 43,000-hand trough in about five weeks — brutal, survivable, over by the next season. A live player at 25–30 hands an hour, playing a serious 1,000-hour year, collects maybe 25,000–30,000 hands annually. The identical, statistically ordinary downswing that costs the online player a hard month costs the live player more than a full year of his life. Same math. Same cards. The difference is only how fast you move through the sample — which is why "how many months" was never an answerable question, and why everyone who answers it in months is guessing.

One more honest note: these are variance-only numbers. If your actual stretch is far outside them, the arithmetic is allowed to say so — that's not a comfort machine, and a distribution that could only ever acquit you would be worthless. Whether the red in your graph is the weather or partly your own play is a different question with its own essay: are you running bad, or playing bad?

Why "How Long" Is Really an Identity Question

Look at the shape of the question you asked. Not "how deep," which is a bankroll question with a bankroll answer. How long. Duration. And duration-terror has a specific signature: it's the fear of a man serving a sentence with no stated end.

That's the thing the weeks-to-months articles never touch, and it's the thing that actually has you at this page. A defined sentence, even a harsh one, can be endured — you mark the wall, you count down. What breaks people is the indefinite one. Three months in, five months in, the number you've stopped saying out loud, and the question underneath the search query isn't statistical at all. It's the one that only speaks at two in the morning, when the session ended an hour ago and you're still in the chair, looking at a graph that goes down: when does the punishment end? And beneath that, quieter: is it a punishment? Am I still good? Was I ever?

Here's what the table above does to that question, if you let it. It removes the sentence entirely — because a sentence implies a judge, and there isn't one. A 43,000-hand trough is not a verdict on you. It is the median behavior of the noise, the thing that happens to the flawless player in the ordinary year. The standard deviation of this game is 100bb/100; a strong edge is 5. The signal is a whisper inside a scream, and over any stretch your nervous system can actually feel — a night, a month, a quarter — the graph is nearly all scream. It cannot convict you on that timescale. It also, and this is the part the upswing never taught you, cannot acquit you. It is weather. You are asking the weather when it will stop raining on you personally, and the honest answer is that it isn't raining on you personally. It's just raining.

Which means the duration question, pressed hard enough, stops being about the downswing and becomes about the relationship you've built with a number — the years of letting the graph tell you who you are, which felt fine on the way up and is eating you alive on the way down. The full anatomy of that — the 3am self-trial, the crown that was also noise, the way out — is the entire subject of The Downswing, and it's more than one essay can carry.

What Actually Shortens It — and What Only Feels Like It Does

Nothing shortens the downswing. Start there, because everything sold to you in this state pretends otherwise. No study binge, no site change, no seat change, no deserving. The cards do not know you're in one. What you can change is not the swing but the experience of the swing — and, critically, whether you're still solvent and still sane when it ends. Three moves do real work.

Run the two clocks apart. A downswing is a race between two clocks: variance grinding toward your true win rate, measured in hands, and your bankroll running toward zero, measured in money. Whichever finishes first decides everything — a genuine winner who goes broke at the bottom of an ordinary downswing never collects on an edge that was real the whole time. The table tells you what the first clock demands: tens of thousands of hands, sometimes six figures of them. Your only job is making sure the second clock loses that race, and that's arithmetic, not character. The standard risk-of-ruin formula (RoR = e^(−2·WR·B/SD²)) says a 5bb/100 winner with 20 buy-ins goes broke about 14% of the time on pure variance; with 50 buy-ins, 0.7%. A 2bb/100 winner with 30 buy-ins goes broke nearly 30% of the time. Moving down when the roll thins isn't retreat. It's multiplying your buy-ins deep — the one dial that takes ruin from one-in-seven toward one-in-a-hundred-and-forty — with your own hand.

Move the scoreboard. The results graph updates on a timescale where it means nothing; your decisions are gradeable tonight. "Did I choose well with what I could know?" has an answer before you stand up from the table. "Am I winning?" does not — not this month, not this year. Players who function through long downswings, rather than merely surviving them, are the ones who moved their self-evaluation off the outcome and onto the choice, in writing, on a schedule. That conversion — from an unanswerable question to an answerable one — is what actually changes how long a downswing feels, because it ends the nightly wait for a verdict that the math says cannot arrive.

Count hands, not months. Your suffering runs on calendar time; the sample runs on volume. Five months of low-volume pain can be 20,000 hands — a stretch the table above regards as almost nothing, noise inside the noise. When the weight of how long this has gone on lands on you, translate it out of months and into hands, against the distributions above. The weight often turns out to be a rumor.

And if the darkness under the question has stopped being about poker — if "how long" has quietly become "whether" — that deserves a straight answer too, not a pep talk: should I quit poker?

Where Your Stretch Falls

Everything above is the distribution for three benchmark win rates at one standard deviation. Your stretch has its own coordinates: your win rate, your tracker's actual standard deviation, your hands played, your buy-ins lost. The Verdict takes those four numbers and places your downswing on the curve — tells you whether the thing that's been eating you for three months is a 1-in-3 stretch that variance hands out routinely, or a genuine outlier that's earned a harder look at the play itself. Five minutes, and you'll know which conversation you should actually be having. Most players are shocked in the direction of ordinary.

So: how long do poker downswings last? Tens of thousands of hands as the ordinary case, six figures of hands in the tail, longer than your nervous system believes possible and shorter than your fear insists at 2am — and the length was never the real question. The real question is what you've asked the graph to tell you about yourself while you wait, and that question has a longer answer than a table can hold. It's a book called The Downswing. The whole thing is on the shelf.