bb/100 vs Dollars per Hour: Why the Signs Can Differ

bb/100 measures your recorded result in big blinds per 100 hands. An hourly result divides by time instead. At mixed stakes, a positive bb/100 can coexist with a dollar loss because the blinds you lose can be worth more money than the blinds you win. Changing the time denominator alone cannot reverse the sign of the same result.

To reconcile the figures, convert each constant-stake segment into big blinds before adding it, count the actual hands, and use one shared clock for simultaneous tables. The invented examples below produce +5bb/100 and −$5 per hour without an arithmetic contradiction.

Give each rate a numerator and a unit

Here, bb means one big blind. At $0.25/$0.50, one bb is $0.50. PokerTracker's current unit guide uses big blinds for no-limit and pot-limit BB/100, but big bets for fixed-limit BB/100. Check the definition in your report; capitalization alone is not a universal unit key.

• bb/100: total net big blinds ÷ total hands × 100.
• bb/hour: total net big blinds ÷ elapsed playing hours.
• $/hour: total net dollars ÷ those same hours.

Our scope is completed no-limit cash play in US dollars. The net dollar inputs already include pot rake. They exclude rewards, tips, travel, taxes, staking and special cash-out products. These are results from play under a stated convention, not a complete personal-income calculation. A deposit or withdrawal is not a hand result.

Use the source's actual net-result definition. Do not subtract rake again when it is already included, or substitute an all-in-adjusted line for realized results. Our rake calculation guide addresses the separate task of understanding a game's charges.

Positive big blinds, negative dollars

Imagine two completed, consecutive sessions with no break between them. Session A lasts five hours; B lasts one. The hand counts and results are invented inputs, not estimates of typical poker volume or performance.

On a narrow screen, scroll the table to see the hand and rate columns.

Convert within each row: +$50 ÷ $0.50 = +100bb, while −$80 ÷ $2 = −40bb. Then add the big-blind results: +100 − 40 = +60bb. The combined rate is 60 ÷ 1,200 × 100 = +5bb/100.

The dollar total is +$50 − $80 = −$30. Over six hours, that is −$5/hour. The normalized result is +60 ÷ 6 = +10bb/hour. Neither tells the whole story alone: the larger-stake loss costs more dollars despite containing fewer blinds.

This unit distinction is established. PokerTracker's official support explanation explicitly notes that mixed-stakes winnings and BB/100 can have opposite signs. Our session figures provide a separate, reproducible example.

Combine hands before averaging rates

Averaging the two displayed rates gives (+10 − 20) ÷ 2 = −5bb/100. That treats the 200-hand session as equally influential as the 1,000-hand session. It is an equal-session average, not the rate across the combined hands.

The hand-weighted calculation is (1,000 × 10 + 200 × −20) ÷ 1,200 = +5bb/100. Adding net big blinds and dividing by total hands gives the same answer with less opportunity to lose track of the weights.

Keep the separate stake rows when interpreting the report. A mixed-stakes summary does not establish your result at either individual stake. If the big blind changes within a session, split the record at that change before normalizing.

The bridge to hourly results needs the same sample

bb/hour = bb/100 × hands/hour ÷ 100. In the two-session example, 1,200 ÷ 6 = 200 hands/hour, so 5 × 200 ÷ 100 = 10bb/hour. Both rates must describe the same selected play and the same clock convention.

At one constant stake, multiply bb/hour by the dollar value of that big blind to obtain $/hour. For a separate control, keep the same +100bb and −40bb results but give both sessions a $1 big blind. Their cash results are now +$100 and −$40: +$60 total, or +$10/hour. The combined rate stays +5bb/100.

For the original mixed-stakes case, there is no single actual big blind for the combined play. Multiplying +10bb/hour by a positive average blind cannot produce the correct −$5/hour. Calculate dollars from the underlying rows instead. The control changes the stakes and therefore the dollar results; it is not another description of the original sessions.

Two simultaneous tables share one hour

Now take a separate example: two $1/$2 tables, both played from 10:00 to 11:00. Each records 100 hands and +5bb, or +$10. The combined record contains 200 hands and +10bb during one elapsed hour.

Two tables. One shared clock.

10:0011:00

Table 1 · 100 hands · +5bb

Table 2 · 100 hands · +5bb

Total: 200 hands · +10bb · 1 elapsed hour

The results are +5bb/100, +10bb per elapsed hour and +$20 per elapsed hour. Dividing by two table-hours produces +5bb per table-hour, a different statistic. Label it that way if it is the quantity you intend to report.

For our elapsed-time convention, merge overlapping playing intervals and count their shared minutes once. For example, 10:00–11:00 and 10:30–11:30 cover 90 elapsed minutes, not 120. Exclude recorded breaks. A wider measure including breaks, waiting or travel can also be useful, but it needs its own label and matching time records. This is a declared worksheet convention, not a claim about every tracker's session settings.

These completed inputs do not show that adding a table preserves your per-hand results. They make no prediction about attention, opponents, future volume or winnings.

Audit the record before interpreting the rate

1. Keep one row per constant-stake segment. Record the game, currency, big blind, net-result definition, actual hand count and playing interval.
2. Keep the same selection. Include the losses and folds in the selected play; do not divide its net result by only the hands you voluntarily entered. Count each unique hand once across tables, even when its clock time overlaps another hand.
3. Keep missing inputs missing. If hands are unknown, bb/100 is unresolved; an assumed hands-per-hour figure cannot recover the count. A verified result and elapsed time can still support an hourly figure. If time is unknown, calculate only the rates whose denominators you have.
4. Keep money separate from normalized units. Check the dollars against the matching play records. Account-balance changes can also include transfers or rewards outside those records.

Open the three-example rate inspector to compare the mixed-stakes case, the constant-stake control and the simultaneous tables. It works offline after downloading and accepts no personal records. The six-row CSV contains the exact inputs and row rates. Tournament entries need a different measure; see the re-entry ROI ledger.

Use a results report to choose a study question

A correctly calculated rate describes the selected record. It does not prove a stable edge or identify which decision caused a loss. Separate decision review from the outcome, retain the hand's context, and investigate a specific uncertainty.

For off-table decision practice, GTO Gecko lets you work through simulated scenarios and compare choices with precomputed solver output. A paid plan is required, and the available libraries depend on its game type and tier. Choose a relevant available scenario and keep that feedback separate from your cash results. This article's rate inspector is an independent teaching resource.

Disclosure: GTO Solutions AS publishes both GTO Gecko and this blog.

Method and limits

All three cases are authored arithmetic fixtures, not player data, a simulation or a performance study. The inputs, exact results, Python generator, independent JavaScript verifier and reproduction instructions preserve the calculation. There is no random seed because no sampling occurs.

The examples assume complete, correctly labeled records. They do not estimate statistical uncertainty, establish a suitable sample size, model changing stakes during a hand or recommend stakes, table counts or future income. Header artwork is AI-generated conceptual imagery; the body tables and clock diagram carry the evidence.