Many online players use a tracker such as PokerTracker, Holdem Manager or Hand2Note. They import your hand histories and count things: how often you put money in preflop, how often you 3-bet, how often you fold to a c-bet. Those counts are useful. They answer the question how often.
Rangebet answers a different question: how much did this decision cost, against a solver, with the cards you actually held. This post is about the gap between the two, and why a stat can look perfect while you lose money in that spot.
A frequency hides which hands
Say you call 30% of the time on the button facing a 3-bet, and the solver calls 29%. A stat that counts calls says you're right on target. But that number says nothing about which hands made up your 30%.
You could be calling with a hand the solver always 4-bets, and folding one it always calls, and the two would cancel out in the count. The stat looks perfect and the spot still leaks.
This is not a made-up case. In a real database of 14,124 Ignition hands, BTN vs 3-bet was the second most costly preflop leak, with the player calling 31% against the chart's 29%. Across that database, 23 of the 42 leak spots Rangebet found had a frequency close to the chart's and the wrong hands taking each action.
A frequency has no price
Stats tell you that you differ from a target. They don't say what the difference costs. Folding a little too much in one spot might cost almost nothing; a few wrong calls in another might cost far more.
Rangebet prices every graded decision in big blinds: the EV of the best action for your exact hand minus the EV of what you did. Leaks are then ranked by bb/100, so you work on the spot that costs the most first, not the one where your number looks furthest off.
A target number is not a strategy
Most targets for HUD stats are ranges of good values, such as a 3-bet percentage for a position. A solver's answer depends on more than position: the stack depth, the size of the open, who else is in the pot, and in postflop spots the board and the bet sizes.
Rangebet picks the preflop chart for your effective stack (30 to 150bb) and the open size in the hand. Heads-up postflop pots are solved on our own GPU solver for the exact ranges, board, pot and bet sizes. Your decision is compared with the answer for that spot, not an average across many spots.
What trackers do that Rangebet doesn't
This isn't a case for dropping your tracker. They do things Rangebet doesn't try to do:
- A HUD at the table. Trackers can show stats on your opponents while you play, where the site allows it. Rangebet never runs during play. It only reads hand histories after you play.
- Long-term records. Trackers are built to hold years of results, sessions and reports.
- Every game. Rangebet grades 6-max No-Limit Hold'em cash at 30 to 150bb. Postflop it grades heads-up pots only; multiway pots are shown but not graded.
Use a tracker to see how you play. Use Rangebet to see what each decision cost, and which hands to change.
How the two fit together
- Play your sessions as normal. Your poker client saves the hand histories, and your tracker imports them as it always has.
- Drop the same files, or the whole folder, into Rangebet. Duplicates are skipped, so dropping the folder again later only adds new hands.
- Open Leaks. Pick the top spot by bb/100 and look at the hand classes that cost the most there.
- Drill that spot in the Trainer, which builds decks from your own mistakes.
Not sure where your site keeps its files? See how to export your hand histories from every poker site.
See what your stats can't: drop in your hands and get every decision priced against a solver.
Grade your hands free