Impact study · v4 · October 2026

We measured what happens when poker players actually train.

Across 1M+ recorded training hands, intermediate players gained +0.061 BB per training decision on average over their first 30 days, measured on the same spots before and after training. That is the value of a better preflop answer in training, not a win rate at the table.

Read the full study ↓ Based on real user data · Updated quarterly
headline result
+0.061
BB per decision · intermediates · first 30 days
Day 71421Day 30
1M+
hands analyzed
across all cohorts
+0.061
BB per training decision
intermediate · 30 days · 95% range +0.042 to +0.079
30
day window
cumulative from day 0
2
skill cohorts
beginner · intermediate
01 · The Study

EV improvement, day by day.

Same player, same spot, before and after training, averaged from day 0. Bars show the cohort you're viewing; the line chart compares both skill levels so the gap is visible.

Bars scaled to the highest 30-day result in this cohort. Line chart uses absolute scale across cohorts.
Per-day EV gain · intermediate
Progression · both cohorts
0.000 0.020 0.040 0.060 0.080 day 7 day 14 day 21 day 30 Beginner Intermediate +0.019 +0.023 +0.025 +0.027 +0.081 +0.072 +0.065 +0.061
02 · Your Return

What that gain looks like in dollars.

Pick the stakes you actually play. Below: dollar-equivalent EV gain per 1,000 hands at the selected stake, broken out by training duration.

Your stake
BB value: $5 · skill: intermediate
Expected $ per 1,000 hands · $2/$5
Snapshots by training length
03 · Projection

An upper bound for a month.

Upper bound at $2/$5
$3,050
10,000 hands × +0.061 BB per decision × $5 per BB
10,000
hands per month
(typical recreational)
$5
per big blind
at your stake
+0.061
30-day BB per decision
intermediate cohort
For context

A recreational player typically logs 5,000–15,000 hands in a month of regular play. The projection treats every one of them as a trained decision. Many hands are routine folds that training does not change, so read it as an upper bound, not a forecast.

What this isn't

Variance still applies. This is expected value, not realized winnings, over short samples actual results will differ.

04 · How it works

Why the number moves.

The mechanism is simple: closer-to-solver decisions cost fewer BB on average. We measure the gap each week and report the delta.

STEP 01

Drill GTO ranges

Short, gamified sessions burn the correct play into muscle memory across every preflop spot.

STEP 02

Measured against solver

Every decision is scored against the GTO baseline so progress is real, not vibes.

STEP 03

Better decisions add up

Closer to GTO means more EV per decision. The projection above multiplies that by hands played, as if every hand were a trained spot: an upper bound.

05 · Where players improve

Bigger pots, bigger gains.

The gain per decision after 30 days, by type of spot, all players pooled. A mistake facing a 4-bet costs far more big blinds than a borderline open, so fixing it is worth more. Across every spot type, players fixed 68% to 83% of the spots they first got wrong.

SpotGain per decisionMissed spots fixed
Open (RFI)+0.009 BB75%
Big blind defense+0.009 BB68%
Facing an open+0.004 BB70%
Squeeze+0.007 BB70%
Facing a 3-bet+0.050 BB73%
Cold 4-bet+0.046 BB69%
Facing a 4-bet+0.070 BB70%
Facing a 4-bet all-in+0.088 BB75%
Facing a 5-bet all-in+0.360 BB83%
06 · Methodology

How we measured this.

Sample size
1M+ hands
Recorded in the app up to October 2026.
Measurement unit
BB per decision
EV of the chosen action vs the solver, on a spot the player had seen before, compared with their first try at that same spot.
Cohorts
Beginner · Intermediate
Self-selected at signup.
Difficulty
Held fixed
Every comparison is the same decision, so harder hands later in training do not count against the player. Missed spots are weighted by how often a first try is missed.
Who is counted
50+ hands
Players who answered at least 50 training hands; the median player answered 126. Test accounts are excluded.
Margin of error
+0.042 to +0.079
The 95% range for the intermediate figure (beginners: +0.014 to +0.039), computed player by player, so one heavy user cannot narrow it.
Heavy users
Not the driver
The most active 10% of players made about 80% of the measured decisions. Leaving them out raises the intermediate figure to +0.089, so the result does not rest on them.
Does it last
90 days
Counted from day 0, intermediates are at +0.055 by day 60 and +0.049 by day 90; beginners hold +0.030. The gain lasts, and most of it arrives in the first month.
What it measures
Preflop decisions
EV per decision in the app, on 6-max cash spots at 100bb. It is not a table win rate: postflop play, variance and rake are outside it.
What it does not prove
No control group
Players chose to train, and there is no untrained comparison group, so part of the gain may come from players who would have improved anyway.
The next 30 days

Start the same study
on your own game.

Same drills, same measurement. We'll show you your own EV curve as you train.

App Store Google Play
Free hands to start · cancel any time