This guide follows the documented PT4 distinction between raising and having an opportunity to raise. Its twelve-hand ledger is an original, transparent reconstruction—not an export from a tracker or a claim that an untested parser produces identical flags. The difficult cases are visible: a limp-reraise, an uncontested big blind, and an all-in that removes the option to raise. You can use the ledger to check a definition before using a HUD number to make a read.
What the PFR Number Is Actually Saying
PokerTracker's PT4 support explanation of PFR and RFI says PFR includes any preflop raise, including an initial raise, 3-bet, or 4-bet. That is broader than opening an unopened pot. If you fold the first five hands and 3-bet the sixth, that sixth hand can contribute to PFR even though someone else opened it. A description that silently replaces “raises preflop” with “opens first in” changes the statistic.
Count qualifying hands, not individual raise clicks. Suppose you open, face a 3-bet, and 4-bet in one deal. Under this article's hand-level model, the numerator increases by one, not two. The separate action log contains two raises by you; PFR's event log contains one hand with a raise. Dividing raise actions by all your preflop actions would answer another question entirely.
The working formula is PFR = qualifying raised hands / eligible raising-opportunity hands × 100. For an ordinary sample in which every hand offers a legal raise, four raised hands out of twenty eligible hands give 4 / 20 × 100 = 20%. That does not mean four of twenty actions were raises, or that you raise 20% of the times you face a 3-bet. Both of those need different event definitions and denominators.
Think of the displayed percentage as a compact report. Before interpreting it, recover the report's question: who was observed, in what games, during which dates, with which filters, and over how many eligible hands? A precise-looking number can hide a very small sample or a mixture of situations. Extra decimal places cannot supply missing context.
Choose a Definition Before Classifying a Hand
The PT4 Custom Statistics Guide provides the reproducible route: open Configure → Statistics, select the relevant cash-game or tournament statistic, then inspect Definition and Details. Definition exposes the value expression; Details provides a description and a human-readable formula. Columns can be inspected separately. The title shown in a HUD is not enough to establish the underlying calculation, particularly for a custom statistic.
Record the application version, exact stat name, game section, definition, filters, and report date together. If someone shares a screenshot labeled “PFR,” ask for the definition before arguing about a boundary case. A custom HUD can display familiar shorthand for a modified formula. Also keep cash-game and tournament reports separate unless the definition and purpose of combining them are explicit.
The limp-reraise is the useful test. A player calls the big blind, another player raises, and the original limper reraises. Under the documented “any preflop raise” scope used here, that hand qualifies: the later reraise exists even though the first voluntary action was a call. A metric restricted to entering with a raise would classify it differently. Do not silently import a first-action restriction into an any-raise definition.
That is a source-bound interpretation, not a claim that we inspected the database expression in a particular installed PT4 build. For a software bug report, verify the actual stat and imported hand. If a tracker disagrees, first determine whether the difference is intentional semantics, a filter, an incomplete import, or a parser issue. A historical forum answer for PT3 is not proof of PT4 behavior, and neither proves how another product calculates PFR.
| Question | Model used in this guide | What would change the result? |
|---|---|---|
| Does a 3-bet qualify? | Yes; it is a preflop raise | A statistic limited to opening raises |
| Does opening and 4-betting count twice? | No; one qualifying hand | An action-count metric rather than hand frequency |
| Does a limp-reraise qualify? | Yes, under the documented any-raise scope | A first-voluntary-action restriction |
| Does a forced blind count as a raise? | No | Confusing a forced post with voluntary action |
| Does an impossible raise create an opportunity? | No, in the opportunity model | A raw dealt-hand denominator |
| Is the app's displayed Hands figure always the denominator? | Not assumed | Opportunity exclusions and report filters |
Audit This Twelve-Hand Ledger
The following is a synthetic sequence of independent No-Limit Hold'em hands for one player. Ordinary rows assume enough chips and a legal raise option. “Opportunity” is a binary hand flag, not a count of every occasion the action comes around. A walk means everyone folds to your big blind without requiring a voluntary decision; this reconstruction excludes it from both voluntary-entry and raising opportunities. That exclusion is stated explicitly rather than asserted as a verified parser result.

For RFI, this ledger counts an opportunity when the action first reaches you in an unopened pot; a player has not voluntarily entered before you. Completing after a limper is not an RFI opportunity here. A big-blind walk is not one either. On a hand where you could open but limp instead, an RFI opportunity exists even if you later reraise. The opportunity and the successful event are separate flags.
| Hand | Your position and preflop sequence | PFR event / opportunity | VPIP event / opportunity | RFI event / opportunity |
|---|---|---|---|---|
| 1 | UTG: fold in an unopened pot | 0 / 1 | 0 / 1 | 0 / 1 |
| 2 | HJ: open, then fold to a 3-bet | 1 / 1 | 1 / 1 | 1 / 1 |
| 3 | CO: face an open and 3-bet | 1 / 1 | 1 / 1 | 0 / 0 |
| 4 | BTN: cold-call an open | 0 / 1 | 1 / 1 | 0 / 0 |
| 5 | SB: complete after a button limp | 0 / 1 | 1 / 1 | 0 / 0 |
| 6 | BB: check after a small-blind limp | 0 / 1 | 0 / 1 | 0 / 0 |
| 7 | UTG: limp, then reraise another player's raise | 1 / 1 | 1 / 1 | 0 / 1 |
| 8 | CO: open, then 4-bet | 1 / 1 | 1 / 1 | 1 / 1 |
| 9 | BB: receive a walk; make no voluntary decision | 0 / 0 | 0 / 0 | 0 / 0 |
| 10 | BTN: post a forced ante, then fold in an unopened pot | 0 / 1 | 0 / 1 | 0 / 1 |
| 11 | BB: face a shove larger than your stack; fold, with no legal raise | 0 / 0 | 0 / 1 | 0 / 0 |
| 12 | BB: face a normal open and call, with raising available | 0 / 1 | 1 / 1 | 0 / 0 |
| Total | Twelve dealt hands; different eligible sets | 4 / 10 | 7 / 11 | 2 / 5 |
The reconstructed PFR is 40%, VPIP is 63.64%, and RFI is 40%. PFR and RFI happen to match numerically, but their meanings do not. PFR counts four raised hands among ten raising-opportunity hands; RFI counts two opens among five unopened-pot opportunities. Matching percentages are not evidence of matching event sets.
Using all twelve dealt hands as the PFR denominator would produce 4 / 12 = 33.33%. Counting both raises in hand 8 would instead corrupt the numerator. Neither mismatch should be corrected by changing your impression of the player's style. First repair the classification. The table gives you the intermediate counts needed to identify which assumption caused the difference.
PFR vs. VPIP: Why the Gap Needs Context
VPIP asks whether a player voluntarily puts money in before the flop; PFR asks about raising. PokerTracker's Basic HUD Guide identifies them as separate frequencies. A voluntary call can increase VPIP without being a PFR event. A forced ante or blind is not itself voluntary entry, and checking the big blind does not mean the player voluntarily called a bet.
Jonathan Little identifies VPIP and preflop-raise figures in an English-language tracker example with English captions. This excerpt is for off-table interpretation, not installation or permission to use a HUD during play; the sample figures are not targets, and the gap alone does not count every call.
With a shared eligible set, raising hands are a subset of voluntary-entry hands, so PFR cannot exceed VPIP under that shared-denominator model. Equality is possible: all voluntary entries in the sample could be raises. A 20/20 sample is not mathematically impossible. Conversely, a wide gap tells you that many voluntary entries were not raised hands under the selected rules; it does not reveal every action taken within those hands.
Do not turn that subset observation into “VPIP must always be higher on every tracker.” PT4 documents the exception directly. Its example has a player facing an all-in larger than their stack: they can call or fold but cannot raise. The two stats then have different opportunities. After another hand containing a raise, its example shows VPIP at 50% and PFR at 100%. The percentages differ because their denominators differ, not because a raise somehow stopped being voluntary.
The ledger's hand 11 is the same kind of diagnostic situation, with original inputs. The player folds to an oversized shove. That adds a voluntary-entry opportunity without a successful entry, while it adds no raising opportunity. In a tiny sample, one such row can move the two displayed percentages sharply. Inspect the underlying hands before treating an unusual relationship as evidence of bad data.
For player analysis, read a gap as a prompt to investigate. Filter for cold calls, small-blind completes, and limps, then inspect positions and prices. Several calls can come from reasonable big-blind defenses; several others can come from expensive cold calls against tight early-position ranges. PFR alone does not tell you which category dominates. The separate VPIP guide covers voluntary entry in more detail.
PFR vs. RFI: Same Street, Different Opportunity
RFI means raise first in. It isolates opening an unopened pot, while PFR can include raises after someone else enters. The PT4 support discussion distinguishes the two; PokerCoaching's chart instructions likewise define RFI charts as the situation where the action folds to you. Those charts are not a universal PFR target and come with their own stack and sizing assumptions.
Take hand 3: someone opens before you, and you 3-bet. That is a PFR event, but it cannot be an RFI event because the pot was already opened. Hand 7 is subtler: you had an unopened-pot opportunity but chose to limp. Your later reraise qualifies for the any-raise PFR model, yet it does not retroactively make the first voluntary entry a raise. The RFI numerator stays zero.
Suppose a button receives forty unopened-pot opportunities in one sample and raises twenty of them. That RFI is 50%. If the same sample contains one hundred total raising-opportunity hands and thirty raised hands, PFR is 30%. There is no contradiction: 20 / 40 and 30 / 100 summarize different sets. The other ten raises can occur after an opponent has entered, and the other sixty eligible hands need not offer an unopened pot.
| Metric | Event being counted | Denominator to establish | What it does not answer |
|---|---|---|---|
| PFR | A qualifying preflop raised hand | Eligible raising-opportunity hands under the chosen definition | How often the player opens when folded to |
| VPIP | A hand with voluntary preflop investment | Eligible voluntary-entry hands | How often the entry was specifically a raise |
| RFI | A raise as the first voluntary entrant | Eligible unopened-pot opportunities | How often the player reraises an opener |
| 3-bet frequency | A qualifying 3-bet | Eligible 3-bet opportunities under that stat | The player's overall PFR |
Position Splits Are More Useful Than a Universal Target
A combined PFR mixes early-position opens, late-position steals, blind reraises, and folds against prior action. Two players can reach the same overall percentage through very different decisions. Before comparing them, separate table size, format, position, and effective stack. A short-handed cash sample and a late-stage tournament sample are not interchangeable just because both use the label PFR.
Consider two original position summaries. Player A raises 10 of 50 eligible early-position hands and 20 of 50 late-position hands: overall 30 / 100 = 30%. Player B raises 15 of each 50-hand group: also 30%. The combined number hides A's 20%/40% split and B's 30%/30% split. Neither synthetic pattern establishes good play; it simply demonstrates information lost through aggregation.
| Synthetic player | Early-position raised / eligible | Late-position raised / eligible | Combined PFR |
|---|---|---|---|
| A | 10 / 50 = 20% | 20 / 50 = 40% | 30 / 100 = 30% |
| B | 15 / 50 = 30% | 15 / 50 = 30% | 30 / 100 = 30% |
Use this distinction when reviewing your own database. If your overall number changes, check whether your decisions changed or the mix of seats and games changed. Playing more heads-up hands can alter a combined report without proving that an existing six-max opening strategy became looser. The position guide is the starting point for naming the seats consistently.
Then study an actual decision class. “I want a higher PFR” is too broad to act on. “I repeatedly limp first in from the cutoff in these reviewed hands” gives you something concrete to investigate. Equally, a higher PFR is not automatically an improvement if it comes from 3-betting too many hands against a range that rarely folds. A frequency is a description of choices, not an independent objective.
Small Samples: Keep the Counts Beside the Percentage
At ten eligible hands, one event changes the result by ten percentage points. At one hundred, one event changes it by one point; at one thousand, by 0.1 point, holding the denominator fixed. This is arithmetic, not a reliability threshold. An observed 20% from 2 / 10 and 20% from 200 / 1,000 have the same displayed rate but very different amounts of evidence behind them.
More hands do not automatically solve every problem. A long sample can combine different stakes, changing strategies, short stacks, and old sessions. Filtering it into a relevant position or situation reduces the count again. Keep the selected period and opportunity total next to the rate; do not claim that a fixed hand count turns an opponent into a known quantity.
A practical note might read: “PFR 20%: 12 qualifying raised hands / 60 eligible hands; six-max cash; observed period stated; not a position-specific range.” That note preserves more useful information than “tight player.” It still does not reveal which hands were raised, how sizing changed, or whether the player adapts to a particular opponent. Use observed showdowns and action context to refine the read without pretending they supply a complete range.
Reproduce a Tracker Check Without Guessing
Start with a small set of complete hand histories you are permitted to use. Preserve the originals, note the import time, and choose one player. Read the actual statistic definition before calculating anything. Mark the opportunity and event separately on every hand; include a reason for excluded rows. Use the same game type and report filters when comparing your count with the application's number.
Next, reconcile the numerator. Did the report count an open, a 3-bet, a limp-reraise, or a repeated raise within one deal? If the raised-hand count matches, reconcile the denominator. A walk, a no-raise-option all-in, an incomplete hand, or a different date filter can explain a mismatch. Do not change two assumptions at once: you will lose the ability to identify the cause.
When reporting an unresolved mismatch, provide the application version, stat definition, anonymized complete hand, expected flags, observed flags, and exact report filters. Keep credentials and other players' unnecessary personal information out of the report. For PT4, the documented Statistics window is the place to establish what the stat is intended to calculate; a forum slogan cannot substitute for that inspection.
Finally, return to the poker question. Use PFR to find hands worth reviewing, then examine the opponent's position, your position, the price, and the effective stack. The 3-bet guide covers that distinct decision family, while hand analysis gives you a broader review process. A useful PFR read starts with a definition and ends with a specific hand—not with chasing a supposedly perfect percentage.
Sources
- PokerTracker 4 — Basic HUD Guide: definitions of common HUD frequencies; not an export of an installed stat's SQL.
- PokerTracker 4 — HUD Troubleshooting, “PFR > VPIP”: actual-opportunity distinction and a no-raise-option example.
- PokerTracker 4 — Custom Statistics Guide: how to inspect Definition, Details, expressions, and columns.
- PokerTracker PT4 Support — Difference Between PFR and Raise First In: moderator explanation dated September 15, 2013; any-raise scope versus first-in entry, not a current installed-build test.
- PokerCoaching — Downloadable Preflop Charts, Instructions: RFI situation and chart-specific assumptions; its ranges are not adopted as universal targets here.
The hand ledger, position comparison, counts, and arithmetic are original educational examples. Sources checked October 28, 2026. No tracker import test or player-performance claim is implied.

