Beyond Downswings: Why Extended Breakeven Runs Are Inevitable
COAUTHORED BY HANSEN HE
Poker is an industry of feast or famine. It’s common for me to receive coaching inquiries from players who have been breaking even for their last 100,000 hands or more, desperately trying to understand what has gone wrong with their game.
The truth is, these players are not necessarily playing poorly. Massive samples without profit are common and should actually be expected. In this article, I will share new variance calculations that quantify how common and severe unprofitable samples truly are.
My hope is that you’ll come away from this article with a new appreciation for variance. If you’re considering going pro, you'll gain a realistic expectation that you could go a solid year without making any money, allowing you to plan accordingly.
It’s arguably more important than ever to understand variance and prepare for lengthy unprofitable stretches. As GGPoker continues to dominate large parts of the market, players are increasingly funneled into that ecosystem, and are forced to grind razor-thin edges over enormous samples.
Five years ago, I cowrote the article Selection Bias in Variance Calculations with Robert Wells. In that work, we published variance calculations demonstrating the severity of downswings that poker players can expect over large samples.
In this article, my coauthor Hansen He and I continue this research. Instead of measuring downswings, we’ll measure unprofitable samples. An unprofitable sample is defined as any stretch of hands where, at the last hand, your profit is no greater than one buy-in more than your first hand. Additionally, unprofitable samples must not overlap.
Variance Calculations
Here’s how the simulations worked: We simulated 1,000 theoretical poker players, each with the same win rate and standard deviation. Each played the same number of hands, and we measured the length of each unprofitable sample they endured.
The simulation was repeated with four different win rates. Below are the results for a 1 million-hand sample size, i.e., 1,000 players playing 1 million hands each.
1 Million Hands (Standard Deviation = 115 bb/100)
Average Maximum Unprofitable Sample. On average, this is the longest unprofitable streak you should expect to experience in a sample of this size.
Average Number of Unprofitable 50K-Hand Samples. On average, this is how many independent unprofitable 50,000-hand streaks you should expect to see in a sample of this size.
Probability of Unprofitable 100K-Hand Sample. This is the chance that you will experience at least one unprofitable 100,000-hand streak in a sample of this size.
These simulations reveal that unprofitable samples are not only common but can be quite severe, especially at lower win rates. For instance:
At a 2.5 bb/100 win rate, there's a 97% chance of hitting an unprofitable 100K-hand stretch.
Even at a 7.5 bb/100 win rate, your expected largest unprofitable sample is around 86K hands.
This data underscores the importance of understanding and preparing for variance in poker. Even elite players will go through extended periods without profit, and recognizing this can help in managing expectations and bankroll effectively.
Remember, these figures are averages: if you run especially bad, those unprofitable stretches can last much longer. The variance becomes even more pronounced when you mix stakes, since a downswing at higher limits could force you to move down for an extended period to recover your losses.
Here are the results of the simulations for two other sample sizes: 500K, and 250K hands, respectively:
500K Hands (Standard Deviation = 115 bb/100)
250K Hands (Standard Deviation = 115 bb/100)
What does all this data tell us? Even for consistent winners, extensive breakeven or negative stretches are an unavoidable reality of poker. The simulations we’ve covered—and their sobering statistics—demonstrate that multiple 50K- or 100K-hand unprofitable stretches are not only possible, but likely. This isn’t limited to modest win rates of 2.5 bb/100; it also extends to higher rates of 5 bb/100 and beyond.
Poker players often refer to “the long run” as a safe haven from variance, which is partially true—but only at the very end of a large sample. Along the path to that distant endpoint, volatility reigns, and you can experience bad luck beyond anything you imagined possible.
During these rough stretches, it’s tempting to abandon your strategy or walk away from the game entirely. But in today’s online poker landscape—where win rates are thinner and volume requirements are higher than ever—persistence in the face of variance is necessary for survival.
The good news is that understanding variance can be empowering. By recognizing the severity and inevitability of these spells, you’ll have more realistic expectations and practice better bankroll management. This makes the grind more sustainable, both financially and psychologically. When those inevitable breakeven streaks appear, you’ll see them as part of the process rather than a sign that you’ve suddenly “forgotten” how to play. Instead, you can stay focused on the one thing within your control: executing your strategy as best as possible, and continually improving your game.
ADDENDUM
The table below shows the 500K-hand samples simulated with a standard deviation of 90 bb/100, which is likely to be closer to your observed standard deviation in online cash games.
It’s important to note that these simulations often underestimate variance because real-world players tend to perform worse during downswings and extended breakeven periods. Consequently, I prefer to use a slightly higher standard deviation in my simulations than what is observed in my database.
500K Hands (Standard Deviation = 90 bb/100)
