Mass data analysis in poker: what it is and how to use it
Mass data analysis (MDA) in poker means studying how a whole population of players plays - not one opponent, but hundreds of thousands of them at once. This page explains what mass data analysis is, why it works, how much data it needs and what it shows, with real numbers from 324 million No-Limit Hold'em cash hands.
What is mass data analysis in poker?
Mass data analysis takes a very large database of hands played by many different players and aggregates them by spot. Instead of asking "what does this villain do?", you ask "what does the typical NL50 player do when they check-call the flop and the turn checks through?". The answer is a frequency - how often the population folds, calls, bets or raises, which bet sizes it uses and which hands it shows down.
In poker the short name is MDA, and the approach is also called population analysis or population tendencies. It is the study method behind most modern exploitative strategy in online cash games: find where the real pool deviates from a balanced (GTO) strategy, then adjust.
Why mass data analysis works
The problem with studying single opponents is sample size. With 800 hands on a villain you have maybe a handful of river decisions in any specific line - not enough to say anything. Mass data analysis solves this by pooling: the same river spot appears hundreds of thousands of times across the population, so the frequencies become stable and the differences between spots become visible.
It works because populations are consistent. Players at the same stake learn from the same content, play the same rake structure and make the same mistakes. A tendency measured on millions of hands at NL50 is a very good prediction of what an unknown NL50 player will do next.
- Preflop stats (VPIP, PFR, 3-bet, fold to 3-bet) stabilise after a few thousand hands per player.
- Postflop lines need far more: a specific turn or river line with a specific bet size can need millions of hands before the numbers are reliable.
- Showdown ranges - which hands the population really has - need the most data of all, because only a part of the hands reach showdown.
How much data does MDA need?
Poker MDA is built on 324 million NLH 6-max cash hands (1.69 billion player-hands, 979,000 unique players) from GGPoker, PokerStars, iPoker, 888poker, WPN, CoinPoker and WPT Global, from NL10 up to NL2000+. That is the scale needed to go beyond preflop averages and read river lines by bet size and board.
What mass data analysis shows: real numbers
A first look at the NLH population by stake. "Regs (hands)" is the share of hands played by regulars; "Pool bb/100" is the result of the whole population after rake.
| Stake | VPIP | PFR | 3-bet | Fold to 3-bet | Regs (hands) | Pool bb/100 |
|---|---|---|---|---|---|---|
| NL10 | 28.2% | 17.6% | 7.3% | 46.8% | 36.1% | -13.0 |
| NL25 | 28.2% | 18.1% | 7.7% | 49.2% | 45.2% | -12.7 |
| NL50 | 29.0% | 19.0% | 8.6% | 50.6% | 56.1% | -13.1 |
| NL100 | 29.3% | 19.7% | 9.1% | 51.8% | 59.9% | -12.5 |
| NL200 | 29.1% | 19.7% | 9.2% | 51.9% | 62.3% | -10.5 |
| NL500 | 29.0% | 20.5% | 9.7% | 53.0% | 67.2% | -8.8 |
| NL1000 | 29.0% | 20.9% | 10.1% | 52.3% | 71.9% | -7.6 |
| NL2000+ | 33.3% | 23.6% | 11.0% | 54.2% | 70.4% | -5.5 |
Population averages, weighted by opportunities. NLH 6-max cash, all rooms. Data as of October 2026.
- The pool gets tougher with stakes, but slowly. Regulars play 36% of the hands at NL10 and 70% at NL2000+. 3-bet frequency rises from 7.3% to 11%.
- Fold to 3-bet goes up, not down. From 47% at NL10 to 54% at NL2000+ - higher stakes fold more to 3-bets, so light 3-bets work better there.
- The pool always loses the rake. The whole population loses 13 bb/100 at micro stakes and 5.5 bb/100 at NL2000+ - roughly the rake paid at each level.
Split by player type, the same data shows where the money comes from:
| Player type | VPIP | Fold to 3-bet | River AFQ | bb/100 |
|---|---|---|---|---|
| Reg | 25.4% | 57.3% | 49.8% | -1.7 |
| Casual | 24.2% | 48.6% | 47.7% | -10.4 |
| Fish | 35.9% | 42.1% | 52.1% | -24.2 |
| Big Fish | 55.6% | 29.4% | 55.3% | -62.6 |
| All players | 29.0% | 51.2% | 50.9% | -11.4 |
River AFQ = (bets + raises) / (bets + raises + calls + folds) on the river. All stakes and rooms.
- Regs are close to break-even after rake (-1.7 bb/100); the losses of the pool come from casual players and fish.
- Big fish fold to a 3-bet only 29% of the time - against them, 3-bet for value and stop 3-bet bluffing.
- Fish are more aggressive on the river than regs (52-55% vs 50% AFQ) - their river bets are less often a strong hand than they look.
Full tables by stake, room and player type: Mass data analysis NLH.
Mass data analysis vs GTO
GTO tells you how to play against a perfect opponent. Mass data analysis tells you how the real opponents at your stakes play. The money is in the gap: every spot where the population folds too much, calls too much or never bluffs is a spot where a deviation from GTO wins more. The best workflow uses both - the solver as the baseline, MDA for the adjustments. More in Poker MDA vs GTO.
How to do mass data analysis
There are two ways to do it:
- Yourself, in a database tracker. You need a very large hand database, a tracker that can filter by line and bet size, and a lot of time building filters - one spot at a time.
- With a mass data analysis tool. Poker MDA has the data and the filters ready: choose the population (room, stakes, player type, aggression, calling tendencies - or one player and players similar to them), build the hand line and read every action, bet size and showdown range, compared with GTO. See the step-by-step guide.
Typical exploits found with MDA
- Spots where the pool folds too much to a specific bet size - and which size gives the best bluff EV.
- Lines where the pool almost never bluffs, so folding marginal hands is profitable.
- Bet sizes that are heavily polarised or heavily merged at a given stake.
- Differences between stakes and rooms - what works at NL50 may not work at NL500.
FAQ
What does MDA stand for in poker?
MDA stands for Mass Data Analysis: studying how a whole population of players plays a spot by aggregating a very large database of hands.
Is mass data analysis better than GTO?
They answer different questions. GTO is the unexploitable baseline; mass data analysis shows how real players deviate from it. Winning players use GTO as the baseline and MDA to find profitable deviations.
How many hands do you need for mass data analysis?
Preflop tendencies are visible on a few million hands, but turn and river lines by bet size need hundreds of millions. Poker MDA uses 324 million NLH cash hands.
Does mass data analysis work at low stakes?
Yes - micro and low stakes have the most players and the biggest, most stable deviations from GTO, so MDA findings are especially reliable there.