Mapping Poker Strategies
I secretly tracked my friends’ online poker games, and here’s what I found.
I started playing regular online poker games with a group of friends during lockdown. Without telling them, I began logging every bet, call, raise and fold using a VBA-based Excel tracker. There were two significant findings.
A leopard never changes its spots.
The three most important metrics in poker are:
- Voluntarily Put $ in Pot (VPIP): how often you voluntarily pay money into a hand before seeing the flop
- Pre-flop Raise (PFR): how often you raise before the flop is seen
- Post-flop Aggression Frequency (Agg): how aggressively you play post-flop
Ideally, the VPIP should be 15-20%, the PFR should be 2-3% lower than the VPIP, and the Agg should be 50-60%.
In these scatter charts, each player is represented by a different symbol, and each of their games is plotted as a different point. The points belonging to each player are clustered in the same region, which suggests that each player’s strategy is very consistent between games. This can be exploited by opponents — for example, Player B should be treated warily when they finally see a flop, whereas Player F sees too many of them and can be milked more aggressively.
Best practices are really best practices.
If you search the web for “tips for amateur poker players”, you’ll find helpful articles [1, 2] containing a few ground truths: don’t drink too much, don’t see too many flops, and be decisive when you choose to see them.
In our friends’ games, the most successful players were also the ones who stuck to the pre-flop rules of low VPIP and high PFR/VPIP the most often. It’s nice when practice validates theory!
| Player | Net Winnings | VPIP | PFR | PFR/VPIP | Agg |
|---|---|---|---|---|---|
| D | $107.22 | 38% | 27% | 72% | 32% |
| B | $33.67 | 26% | 21% | 80% | 28% |
| A | $18.22 | 66% | 21% | 32% | 25% |
| E | -$6.53 | 48% | 40% | 84% | 36% |
| C | -$30.33 | 42% | 43% | 100% | 38% |
| F | -$122.23 | 83% | 54% | 65% | 31% |
The post-flop aggression was less clear-cut.
Having logged the data with a VBA-based tracker in Microsoft Excel, I used Python’s pandas and Plotly libraries for the data manipulation and visualisations respectively. Check out the Jupyter notebook here.