The Global EV tab

The Preflop tab gives you a direct comparison between what you did and what you should have done. The Postflop tab gives you clues about your deviations through the vs Reg comparison. The Global EV tab, for its part, tells you how much each hand earns you, in every situation.

That's where the name Global EV comes from: this tab is there to study your EV globally, hand by hand, situation by situation. It's the most powerful tab in the tracker, and the densest one conceptually.

A quick word on vocabulary: the EV shown in this Global EV tab is the CEV you already know from the Chips won tab of the Dashboard . Except that here it's displayed in bb per hand.

The classic mistake in Spin is staying glued to the EV (chips) curve (which, as a reminder, is what your CEV is computed from): it tells you whether you're winning or losing over time, but neither where, nor why.

The real work starts by breaking down your EV situation by situation, to see exactly where you generate it, where you let it slip away, and to isolate the leaks.

That's precisely what this Global EV tab allows, and what the global curve never will.

To carry out that breakdown, the tab puts four complementary tools at your disposal, and each one deserves a thorough explanation:

• the standard grid, what you see when you land on the tab
• the vs GTO comparison
• the Compare Actions mode
• the Best Action mode

We're going to go through all of it in detail. Hang on.

When you land on the Global EV tab for the first time, no mode is enabled. The grid directly displays your average EV in bb per hand for every combo you played.

Anyone used to solvers will recognize the format: those tools often display the theoretical EV of each hand in a given situation. The Global EV tab, for its part, displays your average real EV.

By real EV, we simply mean the EV you actually made at the tables.

Here's an example on the 3-max BTN, 20 bb+ situation:

How to read it: from the image above, we can conclude that on average at BTN, at 20+ bb, you win 0.52 bb every time you're dealt K8o.

You'll notice a gradient running from red to white to green. What does all that mean?

Each value is compared to the fold

I lied to you a little, to keep things simple. The number in each cell isn't the absolute EV of the hand. It's the EV of playing it, compared to the EV of folding it.

That makes the grid very simple to read:

• Green: you win more by playing this hand than by folding it.
• White: you'd make as much EV folding as playing it.
• Red: you'd simply be better off folding this hand.

On this grid, still at BTN 20 bb+ but with many more hands, you start to seriously wonder whether this player has a leak with his low suited connectors and his low J suited: the pattern persists despite the volume.

For now these are only hypotheses, but they deserve to be dug into. Good news: as we'll see, the tool knows how to answer that.

The KPIs at the foot of the grid

At the bottom of the grid, you'll find a few very important pieces of information:

• bb / hand: your average profitability, weighted by your volume across all combos. On the screenshot below, with 2,211 hands, the player makes +0.61 bb/hand at BTN 20 bb+.
• CEV / hand: that average EV translated into net chips.
  Watch out for the trap: this is the net chip winnings, invested blinds included (unlike bb/hand just above). In SB for example, if you fold 100% of the time, your CEV/hand will be -0.5 bb while your EV in bb/hand stays at 0, as explained earlier.

• Hands: the total number of hands aggregated in the grid.

• Uncertainty: the statistical margin of error on your data, at 90% confidence. The bigger the sample, the smaller it gets.
  The total uncertainty is displayed in parentheses (±) right next to the bb / hand and CEV / hand KPIs. The uncertainty of each cell, for its part, is displayed when you hover over the cell.

This information is very important. It directly answers the question we asked ourselves earlier:

How do you use it concretely? You compare the uncertainty with the value displayed in each cell.

• A cell displayed at -2 bb with ±0.75 of uncertainty: the true EV probably sits between -2.75 and -1.25 bb, so still negative. The signal is solid.
• A cell displayed at -0.5 bb with ±0.75 of uncertainty: the true EV could very well be positive (between -1.25 and +0.25). That's far too close to conclude anything.

Now look at the red zones we suspect of being a leak. Even if some individual cells sit close to the uncertainty, the cluster of negative-EV cells in one coherent spot changes everything: the odds that all those cells are simultaneously negative by chance are very low.

Where an isolated red cell could be a coincidence, a cluster almost never is.

A tip to gain volume

You can always add a neighboring stack size in the sidebar to pool the hands and shrink the uncertainty.

Note that the uncertainty shrinks with the square root of the sample: to halve it, you have to multiply your volume by four.

That's why the Global EV tab only really becomes usable from several thousand hands per situation onward, and why some rarer situations stay volatile for a long time.

The limits of this default grid

This grid looks nice, but it is in fact indicative only and won't be very useful to you as it stands.

Why? Because apart from this question:

... the grid won't give you many answers.

Let's take a concrete example. If you over-fold considerably in HU SB vs BB (which is often the case with beginners), the HU SB vs Fish grid will give you white cells (EV = 0) for every combo you folded.

And it's the same everywhere: nothing in this mode tells you whether you're winning "enough", "not enough", or on the contrary whether you're crushing your field in the situation you're studying. You only get access to your raw data.

It's a bit like analyzing your chips won curve on the Dashboard : over the long run you know whether you're winning or losing, but the analysis hardly goes further.

It's that limit which justifies introducing the Compare to GTO mode. But before that, let's take a moment to present the situations covered by the Global EV tab.

Global EV is organized around 8 parent positions.

Parent situations in 3-max

• Button: every time it's on you to open at BTN.
• SB facing the Button: every time the BTN has entered the pot and it's on you to act in SB.
• SB facing the BB: every time the BTN has folded and it's on you to open in SB.
• BB facing the Button: every time the BTN has entered the pot and it's on you to defend in BB, whatever the SB does.
• BB facing the SB: every time the BTN has folded, the SB has acted, and it's on you to respond in BB.
• BB multiway: every time the BTN and the SB have both entered the pot, and it's on you to act in BB.

Parent situations in Heads-Up

• SB: every hand where you're in SB.
• BB: every hand where you're in BB, whatever the SB does (including when it folds and you win the blind).

For example, by clicking on SB facing the Button, you can see the total EV of your hands for every time the BTN entered the pot, then break it down by what he did: vs limp, vs open 2x, vs all-in.

That way you can analyze your EV more precisely in the SB vs BTN open 2x situation, for example.

In this example in SB facing a 2x open from the BTN (Fish), the grid is mostly green: the player is doing rather well in this child situation. A few red cells remain on some broadway-type suited hands (QTs, Q9s, JTs, J9s).

But with such a thin sample (2,232 hands) and a huge uncertainty (the total CEV, -2.1, is displayed at ±8.8), it's impossible to see anything but noise for now.

With more volume, if those red cells persisted, you could wonder whether the action or the sizing chosen are really suited to recreational players. Good news again: we'll see a bit further on how to settle it.

Before we get into this mode, we need to lay a few foundations.

This part is a bit theoretical, but it's essential to understand what comes next, and to really get value out of the tool.

A starting point: poker is a zero-sum game

If you play a Spin against 2 opponents at your level, you'll win 0 chips per Spin in the long run.

And that holds whatever the player type: 3 fish among themselves will make 0 CEV exactly like 3 regs among themselves.

So in Spin, when you look at your CEV, you're unknowingly comparing it to a baseline worth 0. Having +50 CEV means you exploit your opponents to the tune of 50 chips per Spin.

Without ever meaning to, all this time you've been comparing your results to an equilibrium of 0 to measure your edge.

But a situation or a hand is not worth 0 EV

Let's go further. If playing against opponents at your level gives you 0 CEV overall, that does not at all mean that every one of your actions in every situation is worth 0.

It's the sum of all your actions across all your situations that will be worth 0.

That leads to two important points:

1. Each poker situation (BTN, SB vs BB, BB vs BTN, HU SB...) is not supposed to make 0 EV in the long run, even against opponents at your level. The total is, the situations taken one by one aren't.
2. Each hand within a given situation isn't supposed to make 0 EV either. Strong hands earn, weak hands lose.

Structural EV: your reference for each situation

That brings in a key concept: structural EV.

It's the EV you're supposed to make in a given situation if you play against 2 opponents at your level.

Your total structural CEV is 0 (equilibrium is 0, as we just said). But on the BTN, your structural CEV might be +20. In the SB maybe +4. In the BB maybe -24.

So it is on the BTN that you need to dig to find your room for improvement, whereas the raw reading would have suggested the opposite.

How do we know this structural EV? Thanks to solvers.

They have computed the theoretical EV in every situation for every combo, and let us know exactly the expected EV in each case.

The Compare to GTO mode

Now that these concepts are in place, we can talk about the Compare to GTO mode.

We could also call it:

or...

So, concretely, what does this mode do? It no longer shows your raw EV: it displays the gap between your EV and the EV expected in GTO.

In other words, your winnings compared to those expected in a balanced environment.

The tool calibrates the comparison by stack depth (whole stack by whole stack: 9 bb, 10 bb, 11 bb...), then takes an average weighted by the volume you played at each depth.

It's an “apples to apples” comparison: your EV at 10 bb is not compared to the GTO EV at 20 bb.

Don't worry if this isn't perfectly clear yet, we're going to take an example where we have plenty of data: HU SB All-stack vs Fish, with the Compare to GTO mode turned on.

It's often the situation with the largest cumulative volume, so the uncertainty drops very low and we can start reading the grid with confidence.

Reading the colors

The big difference with the default grid is that without the Compare to GTO mode, red = you're losing money.

Here, red = you may be winning money (or not), but less than theory would win against GTO opponents.

And green = you may be losing money (or not), but you lose less than GTO would lose vs GTO.

Reading benchmarks by stake level

• Low stakes: your grid should be mostly green. The pool makes a lot of mistakes, and every mistake you avoid turns into a positive delta vs GTO. A white grid at low stakes is a bad signal: it means you are exploiting nothing. Once again, we will come back to the special cases, so read all the way to the end.
• High stakes: you naturally get closer to white. The game is harder, opponents' mistakes are rarer. But to beat the rake and stay a winner, your overall average still has to lean slightly toward green.

Three traps to avoid

Trap 1: don't fixate on combo-by-combo values

Let's go back to our HU SB All-stack vs Fish grid. You can see that your biggest hands (AA, AKo) and a few suited broadways (ATs, KTs) show up red, while the rest of the grid is overwhelmingly green.

The uncertainty on each cell, ±0.33 bb, is still high, but a trend in that direction is taking shape.

If you have followed along so far, you might tell yourself:

It's tempting, because we don't win as much as GTO would win in an environment at equilibrium.

But careful: we are technically “exploited” on these combos relative to theory, yet that does not at all mean there is a leak.

What is really going on?

The most likely hypothesis is that recreational players in the BB don't ISO much, and more broadly stay fairly passive postflop, letting you realize your equity.

Too passive means you will less often have to pay off their ISO followed by multi-barrels, or their ISO all-in while you hold AA.

Conversely, you will more often see flops with hands like J3o, J2o, T3o.

Another warning

This does not mean you should change the way you play your strong hands in this example.

It simply means you win less than theory on these combos. Maybe you are already winning the maximum possible against a fish opponent, and that is perfectly fine.

What really matters

Without diving into combo-by-combo detail (which mostly serves to understand EV dynamics and the way the pool plays), what matters is checking whether your range as a whole makes more EV overall than theory.

If it does, you have an edge in this situation. If not, you are losing there.

And as always, look at the total uncertainty to make sure your read isn't noise.

Trap 2: don't isolate a child situation

The second trap is even sneakier. You will inevitably run into it as soon as you drill down into a child situation (HU BB vs SB all-in, vs SB open, vs SB limp, etc.).

Let's take an example by looking at our EV in HU BB vs SB all-in (Fish), All-stack:

Looking at this grid, at first glance you might think it is a disaster:

But not so fast.

What is really going on? We can assume that recreational players don't open-shove enough across the whole range of stack sizes.

The consequence of that lack of OS is that your strong hands don't get to call an opponent holding a dominated hand very often. So that kind of strong hand will make less EV than theory would have made against a competent player who would often OS with dominated hands.

And that is exactly what happens. Let's now look at the vs SB limp child situation:

We win a lot of money vs fish in this situation. Why? For the opposite reason: if the fish don't OS enough, that also means they limp too much.

And if they limp too much, they let you realize your equity more often than expected, so you win more than theory would win against a decent player who would have limped less.

So what is combo-by-combo and child-situation analysis good for?

You may legitimately wonder why you would drill down into combo-by-combo detail (trap 1) or into child situations (trap 2) if those reads can be misleading.

To find your own leaks, two complementary approaches:

• Look at the total EV of the parent situations. That is what tells you, at your scale, where you really win and where you lose.
  Note: the Summary view of this tab, which we will see at the end of the chapter, does exactly that recap of each parent situation vs Fish and vs Reg, so you can read it at a glance.
• Apply critical thinking to the red cells. Ask yourself whether they match a pattern consistent with your opponents' leaks (in which case there is probably no leak on your side, just an EV transfer), or whether on the contrary they form a suspicious cluster independent of the pool's known dynamics, which would then be worth digging into.

Trap 3: variance

This is the same issue as on every tab, but even more so here. Global EV really needs a lot of data to produce reliable signals.

Always remember to read the uncertainty at the bottom of the grid, next to the bb / hand and CEV / hand KPIs and to lean on it before drawing the slightest conclusion.

Ignore isolated red cells. A lone red cell in the middle of green ones is almost always statistical noise. On a limited sample, the EV of a single combo can swing hard because of one big hand lost. Do not overreact.

One last thing to note: on some atypical situations (Button limps, multiway spots, certain 3-bets), no GTO reference exists and this mode is disabled.

So if you can't turn this mode on, it is not a bug: it is simply that no reference exists to compare against.

Now let's move on to the Compare Actions mode, another very powerful EV tool.

The Compare Actions mode answers a question no other tool addresses so directly:

To turn it on, click the Compare Actions toggle in the sidebar. It changes the color of the cells and opens a floating panel (draggable): Action 1 on top, Action 2 below, separated by a "VS".

There you tick the actions to compare on each side. The raise sizings appear as sub-rows: so you can also compare, for example, a Raise 2x with a Raise 2.5x+.

A concrete example

Take the 3-max SB vs BB situation. You might wonder: between Raise 2x and Raise 2.5x+, which open sizing earns the most against Fish?

So we select 3-max SB vs BB vs Fish in the right-hand panel, tick the 14 bb+ stacks to get as much data as possible, then in the panel we set Raise 2x as Action 1 and Raise 2.5x+ as Action 2.

How is each cell calculated?

The difference between your average EV with Action 1 and your average EV with Action 2, on that combo. That's all there is to it.

In practice, the tool takes every time you played Action 1 on that combo, averages the EV you got, does the same for Action 2, then displays the difference.

If Action 1 earned more, the delta is positive and the cell is blue. If Action 2 is the one that earned more, the delta is negative and the cell is orange.

Two reading scales. When you hover a cell, a panel details the EV of each action on that combo: you see right away which one comes out on top.

And across the whole grid, the bb / hand KPI in the footer runs the same calculation, averaged over your volume: it's your overall EV(Action 1) − EV(Action 2).

The legend card, below the grid, sums it all up in one sentence: for example "Call earns +0.35 bb/hand more than Raise" on that sample.

Reading the screenshot

In our example, the summary below the grid shows -0.4 bb/hand: overall, Raise 2.5x+ would have been slightly more profitable than Raise 2x on that sample.

Looking at the individual cells, we find an intuitive consistency: the biggest hands (AA, KK, QQ, AKs, AQs) lean toward the big sizing (orange, Raise 2.5x+), while several hands like ATs, KQo, QJo or K9o prefer the small sizing (blue, Raise 2x).

Consistent with the idea that you extract more value with a big sizing on the top of your range against a fish.

Multi-selection

You can tick several actions per side. For example: "Fold + Call" as Action 1, "Raise + All-in" as Action 2.

That's useful to compare strategic aggregates (passive vs aggressive, for instance) rather than atomic actions.

Concretely, on each side, the tool merges the ticked actions into a single group and computes the group's average EV, weighted by the volume of each action.

If on AKs you played 80 times Call (+0.6 bb of average EV) and 40 times Fold (which is worth 0 EV in bb/hand by the grid's convention, as seen in section 1), the average EV of the "Fold + Call" group becomes:

It's this group EV that is then compared with Action 2's.

The traps of the Compare Actions mode

The Compare Actions mode is a very powerful tool, but it has one essential built-in limit.

Between the two, there are four big traps to know about.

1. Selection bias

Imagine you call your KJo at 20 bb deep mostly against very aggressive fish who seem to open-shove one hand out of two, and you fold that same combo against the tighter fish.

Your sample of "Call with KJo at 20 bb" is then a biased selection: it only contains the spots where calling already made sense (the opponent's range is so wide that KJo is profitable).

The Compare Actions mode will show you "Call worked better than Fold", but it isn't Call in absolute terms that is better, it's Call in the specific spots you picked (against those ultra-aggressive fish).

How do you know whether that bias weighs heavily in your case?

If you always play the same range vs Fish without really adapting to the profile, the question doesn't arise: your sample does reflect your overall strategy.

But if you adapt a lot (which is often the case), the "Call worked better" is only a snapshot of your choices in specific spots, not a truth you can apply everywhere.

2. Variance

We say it in almost every chapter, but here too you have to watch the number of occurrences of the actions you compare.

If one of the two actions has very few hands, its average is extremely volatile and the delta displayed can be completely misleading.

The Compare Actions mode can then display "Raise is better" when it's purely variance on the small sample.

Conversely, over 1,000 hands of Call, the average is solid.

3. The trap of silence

If you have always played a single action on a combo (100 times Call and 0 times Raise, for instance), the Compare Actions mode has nothing to compare and the cell stays empty.

But that silence is not a validation of your choice: it simply means no alternative was ever tested. The absence of a comparison says nothing about the quality of your decision, it just says it can't be measured.

And it's even trickier when three actions or more are possible. Imagine that in HU SB you can limp, open 2x or all-in, and that you never use the 2x (only limp or all-in, depending on the day).

The Compare Actions mode may tell you that all-in is better than limp, but you will never know that open 2x was perhaps by far the best option.

4. The postflop play trap

The Compare Actions mode doesn't compare theoretical strategies: it compares your real EV on each of the actions. That EV includes everything that happens after preflop: flop, turn, river.

If you play your raises very well but play your limps badly postflop, it will display "Raise is better than Limp". That is not a theoretical truth: it just means you don't play the limp as well.

In theory, with optimal postflop play, the limp could even be the most profitable option.

Worth keeping in mind: the result of the Compare Actions mode depends as much on your postflop style as on the preflop choice itself.

To finish, the Best Action mode extends the logic of Compare Actions to every available action.

For each combo, the tool picks the action that earned you the highest average EV, among those you actually played.

Cell display

The color of the cell matches the winning action:

The number in the bottom right of the cell shows the average EV of the best action (in bb/hand).

The higher that number, the more that action earned you on average on the combo.

Example: 0.35 means the winning action earned you 0.35 bb/hand on average on that combo.

When the two best actions are nearly equivalent in EV (a gap below 0.1 bb), the cell is split vertically into two colors to signal that neither really stands out.

A first example: HU SB 16 bb+ vs Fish

You can read the strategy at a glance on this grid:

• A large share of the playable combos comes out red (Raise): these are your opening hands, the most profitable with an open.
• Many combos come out green (Call, meaning limp/complete), and a few very dark cells come out All-in.
• Finally, a few blue cells (Fold at 0) appear where every action tried would have been losing.

A second example: 3-max BB vs SB limp (Fish)

Now let's turn on the Best Action mode on the 3-max BB vs SB limp (Fish) situation, over 16 bb+ to build up volume.

What can we take from it?

The top of range comes out red (Raise/ISO), sometimes with very high EVs on the big pairs and the broadways. A few very dark All-in cells show up too.

The bottom of range stays dark green (Check).

No cell switches to blue here: in BB facing a limp, folding is not a legal action (you can check for zero), so the tool never offers Fold = 0 as an alternative.

The per-cell uncertainty stays very high, even though the total uncertainty is reasonable.

At this stage, you can form hypotheses: your non all-in ISOs look very profitable compared with the check, which suggests that widening your ISO range a little could pay off. As always, it still has to be confirmed with volume before you commit.

That lets you visually compare the best action displayed with the action you normally take in game.

Important warnings

Every trap identified for the Compare Actions mode (selection bias, variance, the silence of a missing action, your own postflop play) applies in full to the Best Action mode.

So the Best Action mode gives you two useful kinds of information:

• A map of your past: on which combos you got the best results with which actions, over your whole history.
• Hypotheses to test: if Fold comes out as the winner on combos you were tempted to play, that's a lead worth digging into.

After these four tools, there's one last piece left to discover: the Summary view, the first entry in the sidebar, above the list of situations.

It condenses the 7 parent situations into a single table, building directly on the notion of structural EV seen in section 3.

A Total row at the top, then for each parent situation:

• Hands: your volume.
• Two column groups, vs Fish and vs Reg , each containing:
  Real CEV (your result, with your gap to GTO in parentheses), CEV GTO (the structural benchmark) and Rel. (the statistical reliability of the measurement).

The CEV GTO column: your benchmark for each position

The CEV GTO column shows the structural CEV of each position: what a GTO solver would theoretically get, for example a clearly positive BTN and a clearly negative BB vs BTN.

The column just to its right, with its colored badge, is the reliability indicator of the measurement: green when the gap is clear-cut, orange then red when it stays in the noise.

And every Real CEV cell shows two numbers: your CEV (average chips won or lost per tournament), then in parentheses your gap versus GTO.

Example: on the BTN row, 28.8 (+2.4) vs Reg means you won 28.8 chips per tournament against Regs, that is 2.4 more than a GTO solver on the same sample.

Reading it in two steps

Every cell reads like this: the absolute value tells you your raw result (not very useful on its own), and the delta in parentheses tells you how you play compared to GTO (exploit or leak).

Two typical readings:

• At the BTN (a structurally winning position): being above the CEV GTO on both cohorts is a clear exploit on the profitable position.
• In BB vs BTN (a structurally losing position, negative CEV GTO): losing less than GTO, or even turning the position into near break-even vs Fish, is a massive exploit, even if the absolute value stays negative.

Why the 7 rows don't add up to 0

It all comes down to what this column looks at, and above all to what it does not look at. It sees your situation, your cards, your stack. It never sees your opponent's cards.

But Fish enter at the Button far more often than GTO does. And if they can enter that often, it is necessarily with hands a perfect player would have folded: the good hands enter in every world, the "extra" entries are made with the rest.

So every extra entry is seen twice, by two judges who do not look at the same thing:

• The CEV GTO column sees one more defense spot, priced at the perfect Button's rate: the benchmark drops.
• Your table, on the other hand, sees the trash hand that was used to enter: your result benefits from it.

This CEV GTO is not the Dashboard's

The Dashboard offers a GTO benchmark of its own: the EV GTO curve in the Chips won tab. It is not the same one, and the two numbers will not look alike.

The reason is simple: they do not answer the same question.

In the first case, everyone plays GTO from start to finish. Some players' gains offset the others' losses, so the curve stays close to 0 over time, whatever your pool does. It is a stable horizon line.

In the second, it is your pool that decides which branches you play. And since it sends you into the tough situations more often (we have just seen it), the benchmark drops below 0.

The Rel. column: the first statistical filter

The Rel. badge grades each measurement: Reliable, Indicative or Insufficient, depending on whether your gap vs GTO exceeds its uncertainty or not.

It does the first sorting for you: no need to panic over a negative delta marked Insufficient.

When you hover the Rel. badge, a panel gives you the essentials straight away: your observed gap vs GTO with its uncertainty (± in CEV) and the verdict (Reliable, Indicative or Insufficient). No more comparing the loss to the uncertainty by hand, the badge does it for you.

To dig into a position, you can always click the row: it opens the parent situation in the grid, where you find the combo-by-combo detail and the total uncertainty at the bottom.

One last detail: the vs Fish and vs Reg headers are dropdown menus. You can retarget each column group on any of your opponent tags, to measure for example how well you exploit a specific archetype.

If you're reading these lines, well done: you didn't give up, and you were right not to! Global EV really is a very powerful tab for those who take the time to tame it.

So take your time. Don't feel you have to understand everything on the first pass. Have fun using the tool and, little by little, the analyses will become clearer and clearer to you.

To sum up: Global EV breaks down your EV hand by hand and situation by situation, through four complementary tools.

It is the richest tab in the tracker, but also the most demanding to interpret: you have to know what to read, and above all what not to conclude too quickly.

In the next chapter, we discover the last tab of the Leak Finder: the Overview tab, the dashboard that displays a player's complete playing profile, yours as well as that of any profile in the pool.