Expected Value of Common Plays
Every Scrabble decision has a measurable expected value — the total point impact across current and future turns. Most players make decisions by feel. Champions make decisions by EV. This guide quantifies the value of common play types so you can make data-driven choices instead of gut-based ones.
30 pts
Average turn score (experts)
72-85
Average bingo score
+22
EV gain from smart exchange
8-10
Value of holding an S tile
Understanding Expected Value in Scrabble
EV combines two components: immediate score and leave value. Leave value is the expected average score from the tiles remaining on your rack after playing. A play's true EV is: immediate points + leave value.
💡 The EV Formula
True EV = Score This Turn + Expected Score Next Turn (from leave). A play scoring 28 with leave ERS (expected next-turn score: 35) has True EV of 63. A play scoring 35 with leave QVW (expected next-turn score: 12) has True EV of only 47. The "worse" immediate score wins.
This is why experts sometimes play lower-scoring words. They're optimising over two turns, not one. The numbers always favour good leave over marginal extra points.
EV of Play Types — The Data
Based on simulation data from thousands of games, here's what each play type is actually worth:
Bingo (7 tiles, 50pt bonus)
Average immediate: 72-85 pts. Leave: 0 tiles (full redraw). Two-turn EV: 102-115. Best single-turn play type. Worth sacrificing 10-15pts setup.
Power tile on premium (2-3 tiles)
Average immediate: 28-45 pts. Leave: 4-5 tiles (good control). Two-turn EV: 60-75. Efficient and safe. Use when no bingo available.
Parallel play (3-5 tiles)
Average immediate: 24-38 pts. Leave: 2-4 tiles. Two-turn EV: 55-68. Great on tight boards. Multiple crosswords compound value.
Hook play (1-2 tiles + extension)
Average immediate: 18-30 pts. Leave: 5-6 tiles (excellent control). Two-turn EV: 50-65. Best for rack-building. S hooks worth ~10pts in leave value.
Standard play (4-5 tiles)
Average immediate: 22-32 pts. Leave: 2-3 tiles. Two-turn EV: 50-60. The bread-and-butter play. Aim for good leave composition.
Short dump (2 tiles, weak)
Average immediate: 8-15 pts. Leave: 5 tiles (but were already bad). Two-turn EV: 35-45. Only justified when building toward a specific stem.
Exchange (0 immediate points)
Average immediate: 0 pts. Leave: 0-2 tiles (fresh start). Two-turn EV: 25-35. Only correct when best play scores under 15 AND leave is terrible.
Exchange EV — When Zero Points Now Is Correct
Exchanging scores nothing this turn. So it needs to gain 30+ points in improved next-turn scoring to be worthwhile. Here's when the math works:
✓ Exchange IS Correct
Best play scores <12. Leave from best play is QVWU-level. Bag has 30+ tiles (good redraw odds). You hold a blank but can't bingo (exchange everything except blank).
✗ Exchange Is WRONG
Any play scores 20+. Bag has <10 tiles (can't improve much). You're behind by 80+ (need every point). Rack is mediocre but playable (AEEINT).
🧩 Exchange Decision Calculation
Find your best available play and score it: say 10 pts with leave QVWU
Estimate next turn from that leave: QVWU → maybe 8 pts. Two-turn total: 18
Estimate exchange outcome: fresh rack → average 30 pts next turn. Two-turn total: 30
Exchange gains +12 over two turns. Correct decision: exchange.
📚 Dig Deeper
Leave Values — What Tiles Are Worth Keeping
Every tile has an estimated leave value — how many points it contributes to your next turn on average. These values are computed from simulation:
These values mean: keeping an S tile on your rack is worth approximately 8 extra points on your next turn compared to playing it. This is why experts hold S unless the immediate gain from playing it exceeds 8 points.
Applying EV to Real Decisions
When choosing between two plays, calculate rough EV for each. The winner isn't always the higher-scoring play:
Play A: QUARTZ (32pts)
Leave: ING (+6 leave value). True EV: 32 + 6 = 38. Dumps Q efficiently. Good leave fragment.
Play B: RITZ (34pts)
Leave: QUANG (-1 leave value). True EV: 34 + (-1) = 33. Scores more but leaves Q problem. Worse EV.
Play A wins despite scoring 2 fewer immediate points. Over two turns, dumping Q and keeping -ING produces more total value. This kind of calculation — quick, approximate, done in 30 seconds — is what separates competitive players from casual ones.
EV Exceptions — When Immediate Score Wins
EV analysis assumes you'll have another turn. In certain situations, immediate score matters more than leave:
⚠️ Endgame (bag empty)
Leave value is irrelevant — no tiles to draw. Play for maximum immediate score every single turn. Dump high-point tiles to avoid unplayed penalties.
⚠️ Game-deciding spread
In tournaments, winning by 1 point = winning by 100. If a play clinches the game this turn, take it regardless of leave. Don't risk a lost turn.
⚠️ Clock pressure
Under time pressure, take the highest-scoring obvious play. EV calculation requires time. A 28-point play made in 10 seconds beats a 32-point play you never find.
🔤 Calculate the best scoring plays for your current rack
Open Word Finder →
Loading comments...