How Does Board Position Decide Your Cribbage Strategy?

Board position in cribbage is the score plus who deals next, read against par, a target hole for each deal. A player ahead of par protects the lead by playing off, and a player behind it takes risks by playing on. The first dealer starts ahead, so the first non-dealer attacks from the opening lead.

What is the 26 theory in cribbage?

The Twenty-Six Theory is DeLynn Colvert's board strategy, set out in Play Winning Cribbage and republished as a series of tips on the American Cribbage Congress website. Its premise is that dealer and non-dealer together score about 26 points a deal. An average game can then be projected deal by deal, and every decision judged against that projection. The two per-role averages behind the 26 are on the dealer page.

Colvert's measured averages sum to 26.4, but he rounds down on purpose. Slowing a game takes one player, while pairs, runs and fifteens need both. His projection on the 26 basis follows, with first count alternating as the roles swap.

After nine average deals the first dealer sits one hole short of game, with first count due on deal 10, and the first non-dealer seven short. Colvert turns that gap into two jobs. The first dealer can give away nine points and still count out first, and so defends from the first card. The first non-dealer needs seven points above average for first count within ten of the game hole on deal 9, and so attacks from the first card. That ninth deal decides the game more often than any other: for the first dealer to count out a deal earlier would take seventeen points above average.

His first-hand postures follow from those jobs, one for the non-dealer and one for the dealer.

What are par holes?

Par holes are Colvert's target scores after each completed deal, one list for the first dealer and one for the first non-dealer. He asks readers to memorise both and note which deals give them first count. The lists below are his figures, set out in our own table with the first-count deals in bold. Par is a target rather than an average: the first non-dealer's par is the average-game score plus seven at every deal, and the first dealer's is the average-game score minus nine.

That construction makes the two lists the same holes one deal apart: the dealer's track counts out with first count on deal 10, the non-dealer's on deal 9. Colvert's chart note says par assumes an average game and shifts in an unusually fast or slow one. Reaching par gives about an even chance of winning, so the aim is to pass par without giving up board position. Each point past par improves the odds.

His worked reading of a 36-36 tie after deal 2 shows the lists in use. The first non-dealer is three past a par of 33. The first dealer is far past the 17 of the slow track but seven short of the 43 that the faster, deal-8 track demands. So the dealer must either speed the game up by a full deal or slow it enough that first count falls on deal 10.

What are positional holes?

Positional holes are John Chambers's term for the lowest scores from which your next three counts can be expected to carry you home. The Cribbage Forum's convention sets them at 18, 44, 70 and 96, each to be reached before your opponent does, and Chambers suggests marking them on the board. His derivation works back from home: the players average slightly over 26 a deal, so the last hole sits 25 short of 121, at 96, and the rest 26 apart.

Colvert's first-count pars and Chambers's positional holes are one hole apart: one idea with different bookkeeping, not rival theories. The skunk positional holes, 14, 40 and 66, are the same ladder measured back from the skunk line between holes 90 and 91 instead of from 121. The aim is to reach each before your opponent reaches the matching positional hole. In ACC qualifying play a skunk scores three game points to a win's two, so a player out of position may still play for the skunk, or against it.

Chambers's two-step rule says how the holes are used. Play your position first: knowing the averages and what you need to reach the next hole tells you whether to peg. Take a calculated risk only once position alone will not get you home. A dealer at 100 facing a pone at 104 declines to pair the lead: the dealer has three counts coming, the pone one.

What does the simulation say about par?

Every engine figure on this page is computed from the site's odds data set, src/_data/odds.json, by the Nobs engine, with the method on the odds page. The corpus is 25,000 fixed-seed games of ONYX, strongest of the browser game's three bots, playing itself. Bot-versus-bot play is a proxy for skilled play, not a claim about human games. Each game's final scoring item is clamped at 121, dragging the late-deal means down.

Through deal 6 the corpus reproduces the spacing behind Colvert's par. The first dealer's mean sits about ten holes past the dealer's par at every deal, the first pone's about six short of the pone's. The bots also run slightly faster than the 26 line: 43.14 and 36.98 after three deals against 42 and 36, and 53.52 and 53.25 after four against 52.

From deal 8 the columns fall away from the 26 line through selection, not pace. The deal-9 means cover only the 17,834 games still undecided at its start, the deal-10 means just 6,802. Games decided early have left the sample; each winner's last count is cut at 121. The first dealer's lead over par, about ten holes through deal 7, shrinks to under two at deal 9 and turns negative at deal 10. The first pone's shortfall widens from about five holes to ten. The late rows describe long games, not average ones.

How do you know if you are ahead or behind?

Standing is judged at the start of a deal against the next positional hole. The Cribbage Forum's vocabulary for it, summarised below, ends in one signed number per player, which is how the Ask-the-Experts columns describe a position.

The Forum's start-of-game example shows the arithmetic. The first pone, expecting about ten points on deal 1, will count eight short of 18 and so starts at minus 8. The first dealer, expecting about 26 over the first two deals, will be eight past 18 when counting as pone on deal 2 and so starts at plus 8. Later, surplus and deficit are measured against the next hole from an actual count, or as an expectation assuming average scoring.

That phrase, after counting as pone, is the half step that makes standing depend on the role. The pone counts first, so on the deal that decides the game the pone's hand is scored before the dealer's hand and crib, which are never counted if the pone goes out. Colvert alternates first count deal by deal for the same reason. The Forum's two units of board strategy follow from that order. Three counts are the dealer's hand and crib on one deal followed by that player's hand as pone on the next, with only one round of pegging between them. First count is the pone's hand count. The rule itself is on the counting order and pone pages.

What do playing on and playing off mean?

The Cribbage Forum's glossary sorts play into five modes plus one special case. Desperation offense and desperation defense are the extremes, and playing on and playing off are the everyday leanings on either side of balanced play.

Prevent defense, the Forum's special case of playing off, accepts small scores to cut the risk of a big one, typically with a big lead late on. Chambers's summary test settles most choices in one question. Play on when the points on offer would carry you to your target within your next count or three, provided what you concede leaves the opponent no better placed; otherwise play off.

The bots in our cribbage app Nobs play a mechanical version of the same scale. Position-aware bots project when each side will reach 121 at the literature's pace, in half-deal steps because the pone counts before the dealer. A bot on course to finish at least a full deal ahead switches to defense, one that far behind to offense, and otherwise plays neutral: playing off, playing on and cautious offense.

In defense the Nobs pegging brain weights the opponent's likely reply one and a half times as heavily and, as pone, balks harder; in offense that weight is halved. Prevent and desperation defense are not separate states, because near the end an exact endgame search maximising win probability takes over. Maximising points per hand is not always the goal.

What are the four streets?

The streets are the board's four 30-hole quarters, covered as pace markers on the streets page. Chambers's street tips add a posture for each, read from whose deal it is, both players' standing against the next positional hole, and the hand. He bands hands as poor (0 to 5), average (6 to 9), good (10 to 13) or excellent (14 or more).

On First Street, 1 to 30, the first non-dealer can afford risks, with three streets left to recover any loss, while the first dealer usually plays off. Chambers's deal-1 test is whether the non-dealer reaches about 17 with the dealer not far behind, which removes the dealer's edge.

On Second Street, 31 to 60, the goal is the mid-forties on your own deal with the opponent trailing; from there your next three counts should carry you to 70. Chambers adds a double skunk reminder: pass 60 before the opponent goes out.

On Third Street, 61 to 90, the objective is at least hole 70 with your three counts, setting up Fourth Street position. With that position on the opponent's deal, giving up hand points to starve the crib can be the right trade.

On Fourth Street, 91 to 120, the best position is at least 96 on your own deal with the opponent behind; hole 121 is the game hole, outside the street. Chambers's conditional rules follow, deal by deal.

How big is the dealer's advantage?

In the same corpus the first dealer won 14,100 of 25,000 games, 56.4%, with a 95% confidence interval of 55.8% to 57.0%. Nine deals was the commonest length, 44.1% of games, and 90.5% ended in eight, nine or ten deals. Colvert's figures, from his tip on the dealer's edge and charts from samples of a thousand games or more, sit beside the engine's below. The corpus is a companion to his data, not a test.

Colvert has the dealer winning nine-deal games about 250 to 170, usually by counting the ninth-deal crib while the first non-dealer comes up about five short. The corpus does not split its endings by game length or role, so it cannot test that figure. It does record which count reached 121, whatever the game's length or winner: a hand count in 61.1% of games, pegging in 19.9%, the crib in 18.4% and his heels in 0.5%.

Is it better to go first or second? There is no choice, because the cut for the first deal decides it, and "first" is ambiguous anyway: the non-dealer leads and counts first, the dealer holds the crib. The measurable edge belongs to the first dealer, and Colvert and Chambers agree on the consequence: the first non-dealer attacks from deal 1. Reaching about 17 there with the dealer not far behind takes the edge away.

How much is a point of position worth?

Schell's rule of thumb, from four Ask-the-Experts answers, covers a dealer starting a deal near the Fourth Street positional hole, 96. There each extra point the pone gives up costs about 7%, sometimes quoted as 6%, of the pone's back-end chance of winning, the games the pone wins because the dealer fails to get out in time. He applies it loosely, as a rule of thumb, and calls the positional hole the sweet spot where a single point swings the most.

The table gives the dealer's win rate at the start of a deal, by the dealer's street and margin in ten-point bands, deals in parentheses. Level counts as ahead, the outer bands are pooled, and cells with a few dozen deals are noise.

A point is worth far more late than early. Stepping from the band just behind to the one just ahead lifts the dealer's rate from 45.9% to 58.4% on First Street but from 40.1% to 75.9% on Fourth. The Fourth Street table below uses five-point bands.

Around level, each five-point band moves the dealer's rate seventeen to nineteen percentage points: 49.2% just behind, 67.8% level to four ahead, 84.6% five to nine ahead. That is below Schell's 7% a point, but his figure is for a dealer near 96 and the table pools every Fourth Street start. For the midgame it is the only measurement available.

What changes on Fourth Street?

On Fourth Street the objective stops being points and becomes the probability of counting out on your next one or three counts before the opponent does. Chambers treats it as the street where judgement matters most. His standing warning in the same tip is against pairing when your position is already good, because a six-point reply can cost the game.

Chambers's two closing cases show the counting-out test at work. Level at 118 on your own deal, the opponent counts first, so he keeps four different cards and plays on. At 116 on the opponent's deal he keeps just enough to go out plus cards that can play off. The three-count and first-count arithmetic, the specific count, Barlow's rule and the bitter end are the endgame page's subject.

By the Nobs Cribbage team Last reviewed 2026-09-03