Scrabble Word Finder

Complete Rack Leave Guide

20 min read

Ask any tournament Scrabble player what separates experts from intermediates and they'll give the same answer: rack leave. Beginners obsess over the word they play. Experts obsess equally over the tiles they keep. Your leave — the tiles remaining after your play — determines what you can score next turn, whether you'll draw into a bingo, and how many options you'll have regardless of what tiles come from the bag. This guide teaches you to think in leaves.

+10-15

Phantom pts from great leave

3:4

Ideal vowel:consonant ratio

ERST

Best 4-tile leave

8 pts

Max sacrifice for better leave

What Rack Leave Really Means — Beyond the Obvious

Every play in Scrabble is two decisions packaged as one: which word to play AND which tiles to keep. Most players only see the first decision. When you play TRAIN for 20 points from a rack of TRAINEE, you're also choosing to keep EE. That two-tile leave (duplicate E's, no consonant balance) is poor — it constrains your next turn before you even draw.

The alternative might be NEATER for 16 points, keeping I. A single balanced tile leave. Yes, you scored 4 fewer points, but your next rack will have 6 fresh tiles plus a flexible vowel rather than 5 fresh tiles plus two duplicate vowels that fight each other for word space. Over a full game, these 4-8 point sacrifices compound into 30-50 additional points from consistently stronger racks.

💡 The Leave Equation

True play value = Immediate score + Leave equity. A 25-point play with +5 leave equity beats a 30-point play with -3 leave equity. The first play is truly worth 30; the second is truly worth 27. Stop looking only at the scoreboard number.

Evaluating Leaves Numerically — The Equity System

Tournament players assign equity values to every possible leave combination. These values come from computer simulations that play millions of games from each leave position and measure average future scores. You don't need to memorise thousands of values — understanding the principles lets you evaluate any leave quickly.

Leave Equity Why
? (blank)+25 to +30Enables bingos; can be any letter
S+8 to +10Hooks onto almost any word; plurals + bingos
ERST+7Best 4-tile leave: balanced, bingo-friendly
ERS+6Strong 3-tile: S hook + good consonant-vowel mix
ER+4Productive pair: -ER suffix, many extensions
E+2Single flexible vowel, neutral to positive
empty (all tiles played)0Baseline — drawing 7 fresh tiles is average
VV-8Duplicate + hard letter = very constrained next turn
QU-6Committed to Q-word; limits options severely
UUVW-12Worst common leave: duplicates + awkward consonants

The practical takeaway: a leave with positive equity effectively adds those phantom points to your play's value. Choosing a 22-point play that leaves ERST (+7) over a 28-point play that leaves VV (-8) is a 15-point equity difference that favours the lower-scoring play. The 22+7=29 beats 28-8=20 in true value.

The Best and Worst Leaves — Reference Tables

Knowing the extreme ends of the leave spectrum helps calibrate your instincts. Here are the strongest and weakest common leaves you'll encounter:

✓ Top 1-Tile Leaves

Blank (+28), S (+9), E (+2), R (+1), A (+1). Single tiles that add immediate flexibility to any draw.

✗ Worst 1-Tile Leaves

Q (-5), V (-3), U (-1). These tiles constrain draws and limit word formation with most combinations.

✓ Top 2-Tile Leaves

?S (+35), ?E (+30), ST (+5), ER (+4), ES (+4). Pairs with either blank or high-synergy combinations.

✗ Worst 2-Tile Leaves

QV (-10), VV (-8), UU (-6), WW (-6). Duplicates and dead-weight combinations that cripple next turn.

Vowel-Consonant Balance — The 3:4 Rule

The ideal post-draw rack (7 tiles) should contain 3 vowels and 4 consonants (or sometimes 2:5 for consonant-heavy bingo stems). Your leave should aim to produce this balance after you draw replacement tiles. Since you're drawing randomly, you can't guarantee the ratio — but you can set up conditions that make balance likely.

🧩 Balance Reasoning

1

The tile bag is roughly 44% vowels, 56% consonants. Leaving 1 consonant + drawing 6 random tiles gives expected 2.6 vowels + 4.4 consonants = fine balance.

2

Leaving 2 vowels + drawing 5 gives expected 4.2 vowels total — too vowel-heavy. Avoid leaving 2+ vowels unless one is a blank.

3

Leaving 3 consonants + drawing 4 gives expected 4.8 consonants — slightly heavy but workable. The -ING suffix loves consonant draws.

Exception: when you hold a blank, balance matters less because the blank can serve as whatever letter you need. A leave of blank + 2 consonants is fine because the blank compensates for any vowel shortage in your draw. Similarly, S tiles have enough hooking power that S + consonant leaves are strong regardless of vowel count.

Synergy Tiles — Combinations That Multiply Value

Individual tile values are useful, but the real power of leave evaluation comes from understanding synergy. Certain letter combinations produce far more valid words together than their individual values would suggest. Keeping these combinations intact — even at a scoring cost — dramatically improves your next turn's ceiling.

High Synergy Pairs

E+R (thousands of -ER words), I+N (-ING stem), S+T (-ST endings, SATIRE stem), A+N (-ANT, -AN words), E+D (-ED past tenses)

Bingo Stem Fragments

SATI (+any 3 = 100+ bingos), ERAT (+any 3 = 80+), INES (+any 3 = 70+), TION (productive suffix start)

Anti-Synergy Pairs

Q+V (no words use both without U), W+W (minimal WW words), U+U (duplicate vowels), V+V (1 word: REVVED)

False Friends

Q+U looks good but commits you to a Q-word. I+I seems balanced but duplicate vowels limit. J+consonant is dead weight.

When to Sacrifice Points for Better Leave

The million-point question: how much should you pay for a better leave? The answer depends on game state, but general thresholds apply:

0-5 points difference: Almost always take the better leave. The equity gain exceeds the score sacrifice every time at this threshold.

6-10 points difference: Take the better leave if the equity difference is large (e.g., keeping blank vs not, or ERST vs UUVW). Otherwise, score.

11-20 points difference: Usually score. Only sacrifice this much for a blank-saving play or when the bad leave is catastrophic (-10 or worse equity).

20+ points difference: Always score. No leave improvement is worth giving up 20+ points of immediate scoring. The bird in hand wins.

Context modifiers: when you're behind by 50+, score aggressively (you need points now, not future potential). When you're ahead by 50+, take the better leave (protect your advantage with consistency). When it's early game, leave quality matters most (many turns to benefit). When it's late game, immediate scoring dominates (fewer turns to cash in leave equity).

The S-Tile Dilemma — When to Play Your Most Valuable Letter

The S tile is worth only 1 point in face value but approximately 8-10 points in strategic value. It hooks onto nearly any word (forming the plural), enabling parallel plays that score in two directions. It also appears in most bingo combinations. So when is it correct to play an S?

👑 The S Rule

Play S Only for 8+ Extra Points

The S must earn its keep — it's worth approximately 8-10 phantom points held

Only spend an S if doing so gains at least 8 more points than your best S-retaining play. If CATS scores 18 and CAT scores 12, the S earned 6 extra — not enough. Keep it. If CATS on DWS scores 30 and CAT scores 12, the S earned 18 — well worth it.

Exceptions to the S rule: when you hold 2+ S tiles (spend one, keep one), when you're in late endgame (no future value left), or when the S enables a bingo (50-point bonus easily covers S's strategic cost). But in general, playing S for small gains is one of the most common mistakes intermediate players make.

Practice Method — Training Your Leave Instincts

Leave evaluation becomes instinctive with practice, but you need structured training — not just playing more games. Here's a method used by tournament players to develop leave judgment rapidly:

🎯 Post-Game Leave Audit

After every game, review each turn. Write down what you played and what you kept. For each leave, ask: was there a better leave available within 8 points? Note patterns in your mistakes — most players consistently err toward scoring over leave quality.

🎯 Random Rack Drill

Draw 7 random tiles. Find your 3 best plays (imagining a partially open board). For each play, evaluate the leave: vowel/consonant balance? Synergy? Duplicates? Equity estimate? This builds rapid leave assessment under no time pressure.

🎯 Computer Comparison

Play a game and record your choices. Then run the same positions through a Scrabble engine and compare leaves. When the engine chooses a different play, study why — it's almost always a leave difference. Track how many points per game you lose to leave errors.

Most intermediate players lose 20-40 points per game to leave errors — choosing slightly higher-scoring plays with significantly worse leaves. Reducing this to 5-10 points (expert level) is achievable within a few months of focused practice. The key is awareness: once you start seeing every play as a score + leave combination rather than just a score, improvement happens naturally.

Leave Equity Tables — The Full Reference

Serious players study leave equity tables — massive lookup references computed from millions of simulated games. While you can't memorise every combination, knowing the relative values of common leaves transforms your decision-making. Here are the key groupings you should internalise:

💡 Understanding Equity Numbers

Equity is measured relative to an empty rack drawing 7 tiles (equity = 0). A leave of +7 means your next turn averages 7 points better than drawing 7 random tiles. A leave of -8 means your next turn averages 8 points worse. These numbers come from simulating millions of draws from each position.

Leave Size Best Leaves Worst Leaves Range
1 tile? (+28), S (+9), E (+2)Q (-5), V (-3), U (-1)33 pts
2 tiles?S (+35), ?R (+30), ST (+5)QV (-10), VV (-8), UU (-6)45 pts
3 tiles?ST (+36), ERS (+6), EST (+5)QVW (-14), UUV (-11)50 pts
4 tiles?ERS (+37), ERST (+7)QUUV (-16), UVWW (-13)53 pts
5 tiles?ERST (+38), AEINR (+6)Extremely rare (exchange)55+ pts

The pattern is clear: blanks dominate the equity table at every leave size. A blank in your leave is worth roughly 25-30 points of equity regardless of what other tiles accompany it. This is why experts will sacrifice enormous immediate scoring to retain a blank — a 40-point play that keeps the blank often beats a 55-point play that uses it, because the blank's future value exceeds the 15-point difference.

The RETAINS hierarchy: Among 6-letter leaves (when you play just 1 tile), the letters R-E-T-A-I-N-S form the strongest combination. Any 4-5 letter subset of RETAINS is a premium leave. Memorise this word — if your leave contains mostly RETAINS letters, you're in strong shape regardless of the specific combination.

Diminishing returns on leave size: A 1-tile leave draws 6 tiles (lots of randomness). A 4-tile leave draws only 3 tiles (your leave dominates your next rack). Larger leaves mean less randomness and more control — but also mean you played fewer tiles this turn, likely for a lower score. The sweet spot is 2-3 tile leaves for balancing control with scoring.

Common Leave Mistakes

Understanding common errors is often more instructive than studying correct plays. These mistakes account for the majority of leave-related point losses among intermediate and advancing players:

✗ Mistake 1: Keeping Q "for later"

Players hold Q hoping to draw U for a big score. But Q has -5 equity per turn held. Holding Q for 3 turns costs ~15 phantom points. Unless you're within 1-2 draws of the endgame and Q-words are playable on the board, dump it or exchange immediately. QI scores only 11 — rarely worth multiple turns of suffering.

✗ Mistake 2: Playing S for marginal gains

Adding S to a word for 3-5 extra points is one of the most common leaks. S is worth ~9 phantom points in equity. Playing DOGS (10) instead of DOG (8) gains 2 points but costs 9 in leave equity — a net loss of 7. Only play S when it gains 8+ extra points or enables a bingo.

✗ Mistake 3: Ignoring duplicates

Keeping EE, II, or OO because "vowels are good" misses the point. Duplicate tiles fight for the same positions in words. Leave EE makes fewer words than leave ER with the same number of tiles. Always break up duplicates unless the only alternative is a catastrophic leave (like QV).

✗ Mistake 4: Tunnel vision on longest word

Players find a 7-letter word on their rack and play it without checking alternatives. A 5-letter play for 5 fewer points but keeping ERST is often superior to a 7-letter play that leaves nothing. Always check: "Is there a shorter play that scores close but keeps better tiles?"

✗ Mistake 5: Not considering the bag

Leave quality is relative to what's left to draw. Keeping E when 10 of 12 E's are already played is worse than keeping E early game (fewer E's in the bag means your E may produce EE duplicates). Late game, consider what's actually available in the unseen pool when evaluating leaves.

A useful exercise: after each game, identify your single worst leave decision. Calculate how many points it truly cost (score difference + equity difference). Most players find they make 1-2 "big" leave errors per game worth 10-15 points each. Eliminating just these major mistakes — even if smaller errors persist — often adds 20-30 points to your per-game average.

Advanced Leave Theory

Beyond basic equity evaluation, advanced players consider second-order effects that the simple "leave value + score" equation doesn't capture. These factors separate expert-level from merely competent leave play:

💡 Dynamic Equity — Leave Values Change

Static equity tables assume a full, random bag. But mid-game, the bag's composition changes based on what's been played. Leave E is worth less when 10 E's are already played (fewer E-compatible tiles remain to draw). Leave S is worth more when the board has many hookable words open. Dynamic equity adjusts leave values based on current game state.

Board-dependent leave value: A leave containing S is worth more on an open board with many hookable endpoints than on a closed board where few S-hook spots exist. Similarly, leave containing a blank is worth more when triple-word squares are accessible (the blank enables longer words reaching premium squares). Always assess your leave relative to the board, not in isolation.

Opponent-aware leave strategy: If your opponent just exchanged 7 tiles (likely had a terrible rack and now has fresh tiles), you might prioritise scoring over leave quality for this turn — they're about to score well regardless of your leave. Conversely, if your opponent is clearly bingo-hunting (played a short word keeping 5-6 tiles), a strong leave of your own increases your chances of counter-bingoing.

The tempo concept: Keeping a great leave is an investment — you expect to cash it in next turn. But if your opponent blocks the best spots between now and then, your leave's value evaporates. When the board is closing rapidly, take immediate points because future opportunity is shrinking. When the board is wide open, invest in leave because opportunities will persist.

Variance management: Tournament strategy sometimes favours volatile leaves when behind and stable leaves when ahead. Keeping a blank (high variance — could enable a bingo or produce nothing special) makes sense when you need a big swing. Keeping ERST (consistent, always produces playable racks) makes sense when protecting a lead — you don't need a miracle, just steady scoring.

The most advanced concept is pre-endgame leave planning. When roughly 15 tiles remain in the bag, you can begin tracking (see our tile tracking guide) and know approximately what's left. At this point, your leave decisions should account for the specific tiles you're likely to draw — not the average of all possible draws. If you know the bag is S-heavy, keeping consonants becomes more valuable. If the bag is vowel-depleted, keeping a vowel prevents a consonant disaster.

20-40

Pts lost/game by intermediates

5-10

Pts lost/game by experts

15 tiles

Bag size for pre-endgame planning

3 months

To reach expert leave play

The journey from intermediate to expert leave play follows a predictable progression: first, learn to spot obvious mistakes (playing S for marginal gains, keeping Q too long). Then, develop equity intuition for common leaves (ERST is good, duplicates are bad). Next, incorporate board awareness (is the leave useful given current board state). Finally, integrate all factors simultaneously — score, equity, board, opponent, tempo, and variance — into a single fluid assessment that takes seconds rather than minutes. This is the art of competitive Scrabble, and rack leave is its beating heart.

🔗 Try our free Scrabble Word Finder — instant results, no sign-up

Open Word Finder →

💬 Comments

Loading comments...

Leave a comment

0/2000