What Is EV Loss in Poker? The Stat That Ranks Your Leaks

Your trainer told you the fold cost 0.42 big blinds. You nodded, clicked next, and kept playing, because 0.42 of anything doesn't sound like a number that deserves a reaction. Four hundred hands later you folded there again. That's the trap with EV loss: the per-hand figure always looks small, and the yearly figure never is.

EV loss is the expected value you gave up by taking one action instead of the best action available in that exact spot. It's measured in big blinds per decision, or as a share of the pot, and it's the only common poker metric that puts a price on a mistake rather than just naming it. Sort your hands by EV loss and you get a to-do list. Sort them by how bad they felt and you get a grudge.

Short answer: EV loss in poker is the difference in expected value between the action you took and the highest-EV action available, measured in big blinds. Multiply that figure by how often the spot occurs and you get the leak's real cost in big blinds per 100 hands.

Key takeaways

• EV loss prices one decision. Frequency turns it into a win rate. An error worth 0.42 bb in a spot that appears 15 times per 100 hands costs 6.3 bb/100.
• GTO Wizard's published worked examples put common preflop errors between 0.08 and 0.49 big blinds, and a single turn sizing error at 0.87 big blinds.
• Solver EV loss and the EV difference in your tracker are different metrics measuring different things. One is your decisions, the other is your run-outs.
• Your single worst hand is almost never your biggest leak. Frequency beats magnitude over a year.
• There is no published standard for a "good" EV loss per hand, so the only benchmark worth using is your own previous number.

What does EV loss mean in poker?

EV loss is the expected value you give up by choosing one action over the highest-value action available in the same spot. Expected value (EV) is the average result of a decision if you could replay that exact situation thousands of times, and in cash games it's expressed in big blinds. So if calling is worth 1.20 bb and folding is worth 0.00 bb, folding carries an EV loss of 1.20 bb. You didn't lose 1.20 bb in that hand. You lose it every time you play that hand, forever, averaged out.

GTO Wizard's help documentation defines EV loss as "how much you'd lose compared to if you had played GTO," and splits it into readings that answer different questions:

That last row matters more than it looks. Give up 2 bb in a 40 bb pot and you played badly. Give up 2 bb in a 4 bb pot and you played criminally. Raw big blinds flatten that difference, which is why percentage of pot exists.

One thing to be precise about: EV loss is always measured against a baseline, and that baseline is usually a solver's equilibrium strategy. That's what you're being marked against, not some abstract notion of perfect play. For the plain-English version of what the baseline is and what it isn't, our guide to what GTO poker actually means covers it without the maths.

How do you calculate EV loss on a single decision?

EV loss equals the EV of the best available action minus the EV of the action you chose. A solver gives you both numbers. You do one subtraction.

GTO Wizard's article on which mistakes cost you the most money runs that arithmetic on real spots. Calling with AJo in one particular preflop node makes 0.07 bb. Calling with A9o in the same node loses 0.42 bb. The gap between them is 0.49 bb, which the article describes as a 49 bb/100 long-term error if you always make it. A single pip of difference, opening A9o instead of ATo, costs 0.08 bb. On the turn, picking the wrong bet size in one of their examples costs 0.87 bb, an 87 bb/100 error.

Those numbers look tiny, and then they don't. Here's the conversion that does the damage:

1. Get the EV loss for one occurrence, in big blinds, from a solver output or a trainer's feedback screen.
2. Estimate how often that spot occurs per 100 hands. Not per session. Per 100 hands.
3. Multiply. The product is the leak's cost in bb/100, which is the same unit as your win rate, so you can subtract it directly.

Put that next to a real win rate. BlackRain79's win-rate benchmarks for small stakes, updated for 2026, put an elite NL10 result at 8 bb/100. A 0.42 bb error in a spot that shows up 15 times per 100 hands costs 6.3 bb/100, which is most of a world-class NL10 win rate, handed back by one habit. At NL10, where the big blind is 10 cents, playing 10,000 hands a month, that's 630 big blinds, or $63, every month, to a mistake you cannot feel.

You'll never notice it at the table. It doesn't produce a bad session. It produces a flat year.

Is EV loss the same as the EV difference in my tracker?

No, and this is where most of the confusion lives, because two completely different metrics ended up with almost the same name.

The EV line in a tracker graph compares what you actually won to what your equity said you should have won at the moment the chips went in. PokerTracker got tired of the confusion and renamed the stat in version 4, arguing that "EV by definition is forward-looking" while the all-in stat "is results oriented." It's now called All-In Equity Adjusted Winnings. Nobody says that out loud, which is a shame, because the name was the whole fix.

Here's an actual opinion: the all-in EV graph is the most overrated screen in poker software. It's the bad beat story in chart form, told by you, to you, on a loop, and it has never once produced an action item. PokerTracker themselves note the stat breaks down in multiway pots and whenever opponent cards were never revealed. If you're running 4 bb/100 under your all-in EV line over 30,000 hands, the correct response is to close the tab.

If you're unsure whether your problem is that line or a genuine leak, the diagnostic in why am I losing at poker separates variance, leaks, and an unbeatable rake structure, in that order.

What is a good EV loss per hand against a trainer?

There is no published benchmark, and anyone quoting you a hard threshold is guessing. The figure depends entirely on which spots the trainer serves. Big blind defence against a button open is a well-trodden node where good players hold EV loss very low. A 4-bet pot on a paired turn out of position is a node where almost everyone bleeds, regs included.

Regulars in poker theory forums throw around roughly 0.02 big blinds of EV loss per hand as a very strong trainer result. Treat that as folklore. It's a plausible order of magnitude, not measured data, and it means nothing without the node mix behind it.

The benchmark that works is you, last month, on the same kind of spot. Two rules keep that comparison honest:

• Compare like for like. Your EV loss in blind-versus-blind pots against your EV loss in blind-versus-blind pots. An aggregate score across a different node mix is noise wearing a grade.
• Read average EV loss per mistake separately from average EV loss per hand. A player making rare, huge errors and a player making constant, tiny ones can post identical per-hand numbers and need completely different fixes.

Why does EV loss rank your leaks better than any other stat?

Because a leak is size of error multiplied by frequency of error, and EV loss is the only common metric that carries both. Every other way of ranking mistakes ranks them by how much they hurt, which tracks pot size and memory, and not at all cost.

GTO Wizard's guide to fixing a poker leak says it plainly: "The 'most costly' mistakes are a combination of the degree of the mistake and the frequency of the mistake occurring." Their example is worth sitting with. Your single biggest blunder might live in a UTG versus UTG+1 4-bet pot, a spot you'll see a handful of times all year. Meanwhile you're making a small, consistent error in button versus big blind single-raised pots, which the same article notes account for 23% of all single-raised pots. A single-raised pot (SRP) is a pot where exactly one player raised preflop and everyone else called or folded, which makes it the most common pot type in no-limit hold'em by a distance.

The rare disaster feels like the leak. The boring one is the leak.

That ordering is why our leak report sorts by EV loss rather than by how much you lost in the hand. Ranking by dollars lost puts your coolers at the top, and a cooler isn't a mistake. Our ranked index of 21 common poker leaks runs the same logic across the whole game, with a rough cost and a fix attached to each one.

How to actually drill this

You can now calculate EV loss, which on its own is worth about nothing. The number tells you which habit to change. Changing a habit takes reps, and reps are boring in a way that reading an article about expected value is not.

1. Pick the single highest bb/100 line from your own multiplication. Not the biggest per-hand EV loss. The biggest product of EV loss and frequency. Write the spot down with every parameter named: position, stack depth, table size, board texture.
2. Run 20 reps a day in that one spot, five days a week, for two weeks. Twenty reps is about five minutes. That's 200 reps total, roughly what it takes for a spot to stop being a decision and start being a reflex.
3. Re-measure the same node and compare the two EV loss figures. Same spot, same filter, same unit. If the number hasn't moved after 200 reps, you drilled a neighbouring spot instead of the one costing you money, so narrow the parameters and go again.
4. Recalculate the ranking and take the next line down. The order changes once you fix the top one. It's supposed to.

Fitting five minutes into a week that already contains a job is its own problem, and our 15-minute daily study routine has the calendar for it. The wider measure, compare, drill, re-measure loop lives in the five-step framework for fixing poker leaks.

This is what LeakSeek is built around: five-minute drills on real spots, feedback on the decision the moment you make it, and a leak report ordered by EV loss rather than by how much the hand stung. It's phone-only, and after the 7-day trial you get one free session a day unless you subscribe, so it won't replace a desktop solver if you want to build your own sims. This worked breakdown of a single hand shows what an EV-ranked read on one decision actually says.

Nobody has ever tilted over 0.42 big blinds. That's exactly why the mistake survives in your game for three years while the cooler that cost you a buy-in gets replayed all week. The mistakes that quietly end poker careers are boring and repetitive. Go find a boring one.