Understanding Downswings and Extended Breakeven Streaks

14 Feb 2025

Intermediate

This material is for medium-skilled players

In this episode, we're going to talk about something that very few poker players fully understand — the true power of variance. I'll be drawing on two posts from my blog, one I co-authored with Robert Wells and another with Hanson He, to shed light on some of the biggest misconceptions about downswings, breakeven stretches, and why they might not be as rare or surprising as you'd think.

Let's start with a little bit of a confession. One of my favorite pastimes used to be plugging massive downswings into the Prime Dope variance calculator, just to see how unlucky I'd been. I'd lose 20, 30, maybe 40 buy-ins, throw the data into Prime Dope, and let it spit out some percentage telling me how rare or improbable my situation was. 

But after a while, something didn't quite add up. You see, these calculators would say something like: «You had a 1 in 14,000 chance of going on this 50 buy-in downswing over 50,000 hands». Yet I kept seeing these supposedly 1 in 14,000 events cropping up way more than I had expected. Not just for me, but for other strong players in my network too. 

Your win rate is assumed to be constant. There's no allowance for tilt or adjustments in your play. In reality, if you're stuck with 50 buy-ins, you might not be playing your A-game. But still, these huge negative runs seem to be happening too frequently. 

The key insight here is selection bias. It's like flipping a coin 10 times, getting 10 tails, and saying: «Wow, that's a 1 in 1,000 event!». Well, sure, if you only flipped the coin 10 times total. If you actually flipped it 500 times, and then found a streak of 10 tails, it's not nearly as shocking. The same goes for poker. You might run a variance simulation for just one 50,000 hand sample, but in your career, you're playing hundreds of thousands or even millions of hands. 

To illustrate all this, my friend Rob Wells and I ran some simulations. We took 10,000 theoretical poker players, each with the same win rate and standard deviation. Then we had them all play the same large sample of hands, up to 3 million in some simulations, and recorded the worst downswing each player experienced. 

And here's what those simulations showed:

• Players with a 5 BB per hundred win rate averaged a worst downswing of 70 buy-ins over a 3 million hand career;
• Players with a 2.5 BB per hundred win rate averaged about 105 buy-ins as their worst downswing. 

Now that's on average! Some players fared better, and some saw even bigger downswings. And we also looked at confidence intervals. For instance, at a 90% confidence interval, the 2.5 BB per hundred winner's max downswing over this period, was 163 buy-ins. That means it's very unlikely, but it certainly isn't off the table. 

We looked at smaller samples, too. At 250,000 hands, which could be played over the course of 6 months or 1 year, a 2.5 BB per hundred player could expect their worst downswing to be 49 buy-ins. That means about once or twice a year, if you're putting in around 20,000 or 40,000 hands a month, you'd be due for a monster downswing. 

It doesn't mean that you're guaranteed to get smacked by variance over that period, but it's in the realm of normal expectation. So that covers downswings. 

But what about breaking even for seemingly forever? I've heard from plenty of players who have gone 100,000 hands or more without a profit. They'll say: «I must be doing something horribly wrong». But a long break-even stretch isn't necessarily the sign of bad play. It might just be variance.

In my recent collaboration with Hanson He, we pushed the variance research further and looked at unprofitable samples. For these calculations, an unprofitable sample is any stretch of hands where, by the time you reach the final hand in that stretch, you haven't made more than one buy-in of profit since the first hand. It's a dry run, basically. 

Our simulation showed that these unprofitable samples are common, especially when your win rate is on the lower side. At 2.5 BB per hundred, which again is actually above average on a lot of sites, there's a whopping 97% chance you'll hit an unprofitable 100,000 hand stretch in a 1 million hand sample. 

Your longest unprofitable streak in a 1 million hand sample is expected to be 291,000 hands. That means almost a full year of break-even, even if you're putting in 25,000 hands per month. Even for a smaller sample of 500,000 hands, the probability of a 2.5 BB per hundred winner going on a 100,000 hand break-even stretch is still 82%, with their average maximum unprofitable sample being 182,000 hands. 

Now, does this mean that you should just shrug and accept a year-long downswing or break-even stretch with no questions asked? Of course not. If you never do a study session or evaluate your game, you might be using variance as a convenient excuse for actual leaks. 

If you know these stretches are possible, and indeed likely at some point in your career, you're far less likely to panic. Whether it's a 70x in downswing or a 100,000 hand break-even slog, you're prepared. If you're properly bankrolled, you've got the mindset to stay the course. If your fundamentals are strong, tilt is minimal, and you're continuously learning, these stretches will eventually end. 

We often hear one-size-fits-all guidelines like, keep 100 buy-ins for cash games. But the truth is, your personal risk tolerance, win rate, and game type matter more than a blanket rule. A 3 BB per hundred winner in a tough environment has different needs than a 10 BB per hundred winner in a softer pool. What the data tells us is that you could easily see 40-70 buy-in downswings multiple times per year if you are putting in significant volume. And for break-even stretches, going 100,000 hands without profit, even as a decent winner, isn't a freak occurrence — it’s actually expected. 

So if your bankroll can't handle that type of variance, you might find yourself stressed, anxious, or worse, forced to move down limits when you'd prefer to stand your ground. 

We've covered a lot, so let's recap the essentials. First: always consider selection bias when you're running variance calculations. Big downswings look super rare when you only consider a single stretch of hands, but over a long career, these rare events become much more probable. 

Second: depending on your win rate, you should anticipate sizable downswings, in the 40-80 buy-in range, even if you're winning at a healthy clip. Third: going 50-100,000 hands without profit can be normal, even for very good players. And fourth: knowing these runs can and will eventually happen helps you stay calm and avoid panic-induced strategic changes that will worsen the problem. 

The silver lining is that understanding variance can be empowering. When a big downswing or break-even stretch hits, you'll realize it's part of the process. You're less likely to freak out, second-guess every hand, or book an impromptu trip to Costa Rica and get one-shotted by Ayahuasca. Instead, you can lean on the work you've put into your game, keep a level head, and stay focused on good decision-making. You can click on Getcoach blog poker articles in the show notes to find even more statistics on the probabilities of downswings and break-even stretches. So thanks for reading, take care, and may the poker gods smile upon you. Most of the time, at least!