--- title: "Hikaru Nakamura blitz analysis: Stockfish on 2,000 games" description: "You out-play your opponents almost everywhere. Your edge is thinnest in the first 10 moves. A free Chess Goat report on Hikaru Nakamura's public chess.com games: improving or lucky, leaks ranked by rating points, and their best moves." canonical: "https://trychessgoat.com/r/hikaru-85291b45/" last_updated: 2026-10-09 --- chess.com / Hikaru · 46,865 games · Stockfish 19 on the latest 2,000 (Mar 11 – Sep 24 2026) # You out-play your opponents almost everywhere. Your edge is thinnest in the first 10 moves. 87% of your games are 3+0 blitz, and your rating went from 3423 to 3440 over the last 90 days. Over those games you made 0.46 big mistakes per 40 moves; your opponents made 0.83. So there is no glaring weakness here. This report ranks six leaks by how much more they cost you than they cost your opponents in the same games. Openings and endgames are where that margin is thinnest: about 1 rating point each. 92.5%your accuracy, last 30d. Opponents: 89.2%. You outplay them in 86% of games. 42%of your big mistakes are made from winning positions: 428 of 1,027 1:11median clock when you blunder a winning position 10games lost after being +5 or better, 7% of all your losses ## The short answer **Analysed:** Hikaru’s chess.com blitz games: 46,865 games · Stockfish 19 on the latest 2,000 (Mar 11 – Sep 24 2026). Report date: September 24, 2026. **Verdict:** Holding at the top, not climbing. Last 90d vs Mar–Jun: accuracy -0.04 (p = 0.82); big mistakes per game +0.04 (p = 0.41); score -0.01 (p = 0.57). None of it is beyond noise. Wins from lost positions: 4% vs 5%, so the results aren’t luck either. **Biggest leaks:** 1. [Openings that start you in a hole](https://trychessgoat.com/#leak-opening): White: +0.4 by move 10 · Black: −0.3 · ≈19 rating points, 1 fewer than your opponents 2. [Endgames: the last point](https://trychessgoat.com/#leak-endgame): 367 endgame mistakes · 57% in rook+minor endings · ≈8 rating points, 1 fewer than your opponents 3. [Not taking what’s offered](https://trychessgoat.com/#leak-missed): 302 missed punishments · 23 free pieces · 5 mates · ≈22 rating points, 7 fewer than your opponents [Free Goat Card for Hikaru](https://trychessgoat.com/u/hikaru) · [How to review your own games for free](https://trychessgoat.com/improve/free-game-review/) · [Time pressure in chess, measured](https://trychessgoat.com/improve/time-pressure-data/) At a glance · Last 90d vs. Mar–Jun ## Your game at a glance - Openings2,970▼ 20 - Tactics3,430▲ 10 - Strategy3,150▼ 30 - Endgames3,200▲ 20 - !!Brilliant moves232in 2,000 games Each number is the rating of a player who plays that part of the game the way you do: your opponents’ average rating (2,980), adjusted by how much win-% you lose per move there compared with them, in the same games. Arrows: the change against your opponents compared with Mar–Jun; grey means it is within the noise. A **brilliant move** is the engine’s top choice that gives up at least two points of material. Last 30d & Last 90d vs. Mar–Jun ## Improving, or lucky? **Holding at the top, not climbing.** Last 90d vs Mar–Jun: accuracy -0.04 (p = 0.82); big mistakes per game +0.04 (p = 0.41); score -0.01 (p = 0.57). None of it is beyond noise. Wins from lost positions: 4% vs 5%, so the results aren’t luck either. Metric Last 30d Last 90d Mar–Jun Read Results Games 335 990 1,010 Rating, start → end 3396 → 3440 3423 → 3440 3292 → 3420 flat Score 89.3% 89.0% 89.8% flat (p = 0.57) Skill: what the engine sees Your accuracy 92.5 92.2 92.2 flat (p = 0.82) Opponents’ accuracy 89.2 88.9 88.7 flat Big mistakes per 40 moves 0.43 0.46 0.45 flat (p = 0.41) Opponent blunders you punish 83% 81% 80% flat More accurate than opponent 86% 86% 87% flat Eval at move 10 as White +0.4 +0.4 +0.4 flat Eval at move 10 as Black −0.3 −0.4 −0.3 flat Conversion & luck Score after reaching +5 98.8% 98.6% 99.0% flat Score after being −5 or worse 36.9% 35.3% 40.4% saving fewer Losses that were +5 at some point 4% 8% 5% flat Wins that came from a lost position 4% 4% 5% flat Wins on the clock while losing on the board 1% 2% 1% flat Losses on time 8% 13% 8% worse ### Rating, rolling 50-game average Rolling 50-game average: 3316 → 3439. Most of the climb came in the spring. Since then it has held. ### Accuracy, rolling 50-game average The blue line never dips below the orange one: you 93.0, opponents 89.9. The gap is steady, not widening. YouOpp ### Your clock: wins vs losses Time left on your clock after your move, in the games you won and the games you lost. 1. Move 20 Wins2:20 Losses2:10 In your wins you reach move 20 with 2:20; in your losses with 2:10. Only 10 seconds apart, so your losses are rarely decided by the clock this early. Medians over your 3+0 games (1,502 wins, 116 losses). 2. Move 30 Wins1:42 Losses1:30 In your wins you reach move 30 with 1:42; in your losses with 1:30. The gap grows to 12 seconds. Medians over your 3+0 games (1,198 wins, 99 losses). At this level the rating won’t jump from fixing one big weakness, because there isn’t one. What is left is small: your Black openings (−0.3 by move 10) and the few won positions that slip when the clock runs low. Strengths ## What you do well Not everything is a leak. These parts of your game already work, measured the same way as the leaks below: against your opponents, in the same games. - **Tactics: 3,430**450 above your opponents’ average of 2,980: you see and play tactics better than players at your level. - **Tactical defence**Short combinations cost you less than they cost your opponents: ≈23 points, against ≈84. - **Closing out wins**You close out won games better than your opponents do: ≈16 points lost from winning positions, against ≈39. - **Few pieces hung**You hang pieces less often than your opponents: one-move oversights cost you ≈9 rating points, them ≈25. - **Punishing mistakes**You cash in on their mistakes more often than they cash in on yours: ≈22 points lost there, against ≈29. ### Your best moves Black to move · Move 30 · 1:51 on your clock · level positionYou found a strong sacrifice here: it leaves your queen en prise. Can you find it again? Black to move · Move 28 · 1:49 on your clock · you were +3.3You found a strong sacrifice here: it leaves your queen en prise. Can you find it again? White to move · Move 33 · 0:46 on your clock · level positionYou found a strong sacrifice here: it leaves your queen en prise. Can you find it again? What the engine found · 1,027 big mistakes classified ## 6 leaks, ranked by rating points lost beyond your opponents You beat your opponents in all six of these leaks, by 1 to 61 rating points. Read #1 as **where your edge is thinnest**, not where you are weak. Every position below is from one of your own games in the last 120 days. Decide on your move first. Then open the answer and step through what you played and what was best with the ← → buttons (or your arrow keys). ### 1 · Openings that start you in a hole White: +0.4 by move 10 · Black: −0.3 · ≈19 rating points, 1 fewer than your opponents > You should keep in mind no names, nor numbers, nor isolated incidents, not even results, but only methods. — Emanuel Lasker[Wikiquote](https://en.wikiquote.org/wiki/Emanuel_Lasker) By move 10 you are +0.4 as White and −0.3 as Black. In 6% of your Black games you are already 1.5 pawns down. Similar-rated opponents lose about as much in the opening as you do: your edge here is 1 rating point, the thinnest of the six. Black is where it shows. The seven lowest move-10 evals among your openings are all Black lines, led by the Caro-Kann Gurgenidze (14 games) and the Dragon Yugoslav Attack (18 games). You still score 93–94% in both, so this is a small edge to win back, not a hole you keep falling into. Coach’s notes 1Why it happens At 3+0 your median opening move takes 0.8 seconds. That speed is right when the position is one you know. As Black it sometimes isn’t: −0.3 by move 10 on average, and 1.5 pawns down in 6% of games. One exact position cost you six times: you played **…Qxc5** where **5…Bxc3+ 6.bxc3** came first. 2The habit When a Black game leaves your usual lines before move 10, **spend a few seconds on the first new move**. You can afford it: you reach move 10 with the full 180 seconds. 3The drill After each session, open the Black games that were worse than −0.5 at move 10 and find the move where it turned. Check it with an engine and a database of recent elite games, then add the fix to your file. Start with the Week 3 lines. 4What success looks like Average eval at move 10 as Black above **−0.3** (now −0.3). White to move · Move 8 · 2:37 on your clock · level positionMove 8 as White. Black’s pawn on c3 hits b2, and the queen stares down the e-file at your king. Which move deals with both? You played Be2?? White to move · Move 7 · 2:03 on your clock · you were +5.7Move 7 as White, and you are +5.7. Black’s queen on a5 pins your c3-knight. How do you break the pin? You played Bd3?? White to move · Move 12 · 2:05 on your clock · level positionMove 12 as White. The c5-bishop and the e7-queen both look at e3. Where does your e-pawn go? You played e3?? Black to move · Move 11 · 2:55 on your clock · you were +4.8Move 11 as Black. White’s d7-bishop and c3-knight are both attacked. Which one do you take? You played …Nxd7?? Study with - [Lichess Analysis board + Opening Explorer](https://lichess.org/analysis)practiceCheck your own opening lines against master games. The explorer needs a free Lichess login. Practice this: [Opening rules that win games →](https://trychessgoat.com/learn/#/lesson/opening-rules) ### 2 · Endgames: the last point 367 endgame mistakes · 57% in rook+minor endings · ≈8 rating points, 1 fewer than your opponents > In order to improve your game you must study the endgame before anything else; for, whereas the endings can be studied and mastered by themselves, the middlegame and the opening must be studied in relation to the endgame. — José Raúl Capablanca[Wikiquote](https://en.wikiquote.org/wiki/Jos%C3%A9_Ra%C3%BAl_Capablanca) 52% of your games reach an endgame. From a winning endgame you score 99%; from an equal one, 86%. 57% of your endgame mistakes come in rook+minor endings. Your opponents lose nearly as much in endgames as you do: your edge is 1 rating point, tied for the thinnest. The puzzles below share one thing, the clock. Three of the four came with 24 seconds or less. Coach’s notes 1Why it happens These are not knowledge gaps. Below are a mate in one, a pawn that takes a queen, and a rook that lets a king off the edge. They come at 0:24, 0:07 and 0:08. Your big-mistake rate goes from 0.6 per 100 moves with more than 90 seconds left to 3.9 with under 10. 2The habit In a won ending with under 30 seconds, play **checks and captures of big pieces first**. The pawn grab can wait. 3The drill Set up the four endings below and play them out against the engine with 30 seconds on your clock, until you win each one cleanly twice. Then read the matching rook+minor sections of **Fundamental Chess Endings**. 4What success looks like Keep converting at least **99%** of your winning endgames (now 99%). White to move · Move 81 · 0:24 on your clock · you had mate in 1Endgame, 24 seconds left, and you have mate in one. Checks first. You played Rxa4?? White to move · Move 97 · 0:07 on your clock · you had mate in 9Rook and knight against rook, 7 seconds left. Black’s king is stuck on the edge. How do you keep it there? You played Rf5?? White to move · Move 81 · 0:08 on your clock · you had mate in 138 seconds left. Count before you move: what can your pawns take? You played Kc4?? Black to move · Move 39 · 1:03 on your clock · you were +7.0You are +7.0 as Black. White’s king is on b1 behind a2. Grab f3, or go for the king? You played …Rxf3?? Practice this: [Converting a won game: trade pieces, not pawns →](https://trychessgoat.com/learn/#/lesson/converting) ### 3 · Not taking what’s offered 302 missed punishments · 23 free pieces · 5 mates · ≈22 rating points, 7 fewer than your opponents > I have a weakness for any piece in excess of my opponent's numbers - from pawn to queen. — Tigran Petrosian[Wikiquote](https://en.wikiquote.org/wiki/Tigran_Petrosian) Your opponents hand you material, and 81% of the time you take it. This leak costs you 7 rating points fewer than it costs your opponents. Still, you passed on 23 free pieces and 5 mates. In 54% of the misses the winning move was quiet: an attack on a piece with nowhere to go. Those are the hardest to see at blitz speed. Coach’s notes 1Why it happens Your misses come at speed. The four below were played with 2:05, 1:01, 0:20 and 1:58 left. At that pace you play your own plan, and the opponent’s mistake goes past. 2The habit After every opponent move, ask **“What did that move leave loose?”** One second, before your own plan. 3The drill Collect the quiet misses from your own games, the trapped-piece wins first: 54% of your misses were quiet. Solve each from the diagram in 10 seconds, then spar the engine from it at the clock you had. 4What success looks like Punish at least **88%** of your opponents’ blunders (now 81%). White to move · Move 16 · 2:05 on your clock · you were +4.7Black’s h6-queen is its only real attacker. What do you do with it? You played Rh1?? Black to move · Move 26 · 1:01 on your clock · you were +1.9White’s queen is on h6 next to your king. The e5-pawn is free. Is it? You played …Nxe5?? White to move · Move 58 · 0:20 on your clock · you had mate in 3Black just made a mistake. You have mate in three. Start with checks. You played Bd5?? White to move · Move 26 · 1:58 on your clock · you were +4.2The g6-knight is attacked by your rook. Why can’t Black take back with the f-pawn? You played Qd4?? Practice this: [Is it really safe? Counting attackers and defenders →](https://trychessgoat.com/learn/#/lesson/count-defenders) ### 4 · One-move oversights ≈87 mistakes · 8% of all · ≈9 rating points, 16 fewer than your opponents > Playing Real Chess (when you DO consistently analyze to see if each candidate move is safe) will not make you a good player, but playing Hope Chess will prevent you from becoming one. — Dan Heisman[chess.com](https://www.chess.com/article/view/passive-vs-basic-hope-chess) One-move oversights are rare for you: about 87 mistakes, 8% of all. 54 times a piece went to a square where it could be taken, 31 times an attacked piece stayed where it was. They cost your opponents 16 rating points more than they cost you. When it does happen, it’s the clock. Your blunder rate goes from 0.6 per 100 moves with time to 3.9 under 10 seconds. Coach’s notes 1Why it happens You don’t miss these for lack of skill. You miss them when the clock is low: 0.6 big mistakes per 100 moves with more than 90 seconds, 3.9 with under 10. The check is there; short of time, it gets skipped. 2The habit Under 20 seconds, keep one question: **can the piece I’m moving be taken?** 3The drill 13% of your losses in the last 90 days were on time. After each one, find the move where you spent the seconds you needed later. 4What success looks like Big mistakes at or below **0.4** per 40 moves (now 0.46), measured on your next 50 games. Black to move · Move 54 · 0:52 on your clock · level positionYour knight wants to give check. There are two check squares. Which one is safe? You played …Nd3+?? Practice this: [The blunder check →](https://trychessgoat.com/learn/#/lesson/blunder-check) ### 5 · Giving back won games Reached +5 in 1,586 games · lost 10 · drew 18 · ≈16 rating points, 23 fewer than your opponents > First restrain, next blockade, lastly destroy. — Aron Nimzowitsch[ChessBase](https://en.chessbase.com/post/edward-winter-s-che-explorations-11-) You reached +5 in 1,586 games and lost 10 of them. It costs your opponents 23 rating points more than it costs you. 42% of your big mistakes come from winning positions, and the median clock when you drop a win is 1:11. The puzzles below show the pattern: a pawn grab or a slow move when a check or a forcing move finished the game. Coach’s notes 1Why it happens When you are winning, the shortest path is usually a check. In two of the four positions below you took a pawn when a check was there (27…Ne2+, 28…Qh7+ with mate). The other two were slow moves while the opponent had one idea left. 2The habit When you are +5, **checks first, then captures, then pawn grabs**. Ask what their one idea is before you play for more. 3The drill Replay your 10 games lost from +5 and your 18 draws from +5. Mark the first move where a check or a forcing move was better than what you played. 4What success looks like Stay at or above **99%** after reaching +5 (now 99%). Black to move · Move 27 · 0:31 on your clock · you were +5.3You are +5.3. Your queen guards the e3-rook. What happens if it takes on d5? You played …Qxd5?? Black to move · Move 48 · 0:47 on your clock · you were +6.8You are +6.8. White has a passed d-pawn. Which is faster, your attack or that pawn? You played …g4?? Black to move · Move 28 · 1:59 on your clock · you had mate in 2You have mate in two. Look at checks before captures. You played …Rxf6?? White to move · Move 103 · 0:07 on your clock · you were +6.2You are +6.2, with 7 seconds left. Which of your pawns needs a guard? You played Bd3?? Practice this: [Converting a won game: trade pieces, not pawns →](https://trychessgoat.com/learn/#/lesson/converting) ### 6 · Two-move tactics against you ≈370 mistakes punished by a short combination · ≈23 rating points, 61 fewer than your opponents > After certain particular dispositions of pieces have proved his undoing, the beginner will develop the perception of threats. — Edward Lasker[Project Gutenberg](https://www.gutenberg.org/ebooks/5614) This is your biggest edge: 61 rating points better than your opponents. When short combinations do land, they are capture sequences on under-defended pieces (88), attacks on pieces with no retreat (99), checks (76) and forks (32). All four below came with 1:37 or less on your clock. Coach’s notes 1Why it happens When they do land, the setup is the same: a piece on a square their next move hits. The a4-queen after …Rc5, the king on g7, where one bishop check also hit c7, the queen on d4 in front of a knight. You looked at your move, not at their most forcing reply. 2The habit With under 90 seconds, before you release the move, ask **what is their most forcing reply?** 3The drill Once a week, go through your losses and mark the first move where their forcing reply was already on the board. Spar the engine from that position at the clock you had, until you hold it. 4What success looks like At the next audit, no more than **24%** of your big mistakes come from a short combination (now 36%). Black to move · Move 24 · 1:33 on your clock · you were +2.9You are +2.9 and White’s king sits on b1. Is there a capture that opens it up? You played …Rc5?? Black to move · Move 33 · 1:37 on your clock · you were +4.4You are +4.4. If e5 falls, what does the g3-bishop hit? Where should your king not go? You played …Kg7?? White to move · Move 54 · 0:11 on your clock · level position11 seconds left, and Black’s d2-pawn is one step from queening. Push b7, or deal with the pawn? You played b7?? White to move · Move 43 · 0:25 on your clock · you were +3.7Your a8-rook pins the f8-bishop to Black’s king. How do you win it? You played Qd4?? Practice this: [Forks: one piece, two targets →](https://trychessgoat.com/learn/#/lesson/forks) Lesson plan · 8 weeks · ~30 minutes a day ## The plan ### Four rules, starting today 1. **Black prep on three lines.** As Black you are −0.3 by move 10. The Gurgenidze, the Yugoslav Attack and the Pterodactyl Eastern are where it starts. Week 3 is for them. 2. **Won endings, low clock: forcing moves first.** Winning endgames converted: 99%. Keep it there. 3. **Their move: which piece just lost its squares?** You cash in 81% of their blunders. 54% of the misses were quiet wins against a piece with nowhere to go. 4. **Spend the seconds before move 30.** In losses you reach move 30 with 1:30, in wins with 1:42. Your big-mistake rate goes from 0.6 per 100 moves above 90 seconds to 3.9 under 10. Your in-game checklist ### Before every move 1. Out of prep as Black: a few seconds on the first new move. 2. Their move: which piece just lost its squares? 3. My move: what is their most forcing reply? 4. Winning, low clock: checks first, pawns last. Chess Goat · trychessgoat.com ### Week by week Weeks 1–2 · Leaks 3 + 4 ### Your own positions, calculated Start · Black to move Your game, move 30. You were +4.3. What was the threat? - Daily 15 min: one or two exercises from Aagaard’s **Grandmaster Preparation: Calculation**. Write the full line down before you touch a piece, then check it with the engine. - Daily 10 min: spar the engine from the three positions above with 60 seconds on your clock. Play each one until you win it twice. - Games: 3+0, rated. After each loss, find the first move where the eval dropped and note which it was: their threat, or your clock. In the game after a loss you score 83%, against 89% otherwise, so do this review before the next game. Target: big mistakes ≤ 0.4 per 40 moves (now 0.46) Week 3 · Leak 1 ### Black prep: Gurgenidze, Yugoslav, Pterodactyl Start · Black to move Your game, move 11. Move 11 as Black. White’s d7-bishop and c3-knight are both attacked. Which one do you take? - With an engine and a database of recent elite games, go through your Black games in the **Caro-Kann Gurgenidze** (−0.7 by move 10, 14 games), the **Dragon Yugoslav Attack** (−0.5, 18 games) and the **Pterodactyl Eastern** (−0.3, 33 games, 85% score). - Fix one move-10 position a day. Start with the one you repeated six times: **…Qxc5**, where **5…Bxc3+ 6.bxc3** comes first. - Spar the engine from each fixed position at 3+0, as Black, until the new move comes at your normal opening speed. - Play 20 rated games as Black this week (your Black score: 89%). Log every game worse than −0.5 at move 10. Target: eval at move 10 as Black ≥ −0.3 (now −0.3) Weeks 4–5 · Leaks 3 + 6 ### Quiet wins and forcing replies Start · White to move Your game, move 26. The g6-knight is attacked by your rook. Why can’t Black take back with the f-pawn? - Daily 15 min: one composed endgame study. Solve it fully from the diagram, without moving pieces, before you check. Studies train the quiet winning move, and 54% of your misses were quiet. - Daily 10 min: Aagaard’s **Grandmaster Preparation: Positional Play**, the prophylaxis chapters. Ask what the opponent wants before you calculate your own line. - Spar the engine from the two positions above and the eight in leaks 3 and 6, starting from the clock you had in the game. - In your games: after their move, look for the piece that has run out of squares. Before yours, find their most forcing reply. Target: punish ≥ 88% of opponent blunders (now 81%) Week 6 · Leak 5 ### Closing out +5 games Start · Black to move Your game, move 48. You are +6.8. White has a passed d-pawn. Which is faster, your attack or that pawn? - Replay your 10 losses and 18 draws from +5 with the engine. Mark the first move where a check or a forcing move was better than what you played. - From each marked position, play the engine with 1:11 on your clock, your median when a win slips. Finish each one twice. - When you are +5: checks, then captures, then pawn grabs. Ask what their one idea is before you play for more. Target: score from +5 ≥ 99% (now 99%) Week 7 · Leak 2 ### Rook+minor endings at blitz speed Start · White to move Your game, move 97. Rook and knight against rook, 7 seconds left. Black’s king is stuck on the edge. How do you keep it there? - Rook+minor endings are 57% of your endgame mistakes (211 of 367). Work through the rook+minor chapters of Müller & Lamprecht’s **Fundamental Chess Endings** and **Dvoretsky’s Endgame Manual**. Check each key position with the engine. - Spar the engine from the two positions above and from your other rook+minor mistakes with 30 seconds on your clock, until you win each cleanly twice. - From an equal endgame you score 86%. Shereshevsky’s **Endgame Strategy** is the reference for pressing small edges in long endings. - Under 10 seconds: checks and captures of big pieces first. The pawn grab can wait. Target: winning endgames converted ≥ 99% (now 99%) Week 8 · Re-measure ### Audit again - Play 60 rated 3+0 games, half of them as Black. - Re-run this analysis on them. You pass if big mistakes are ≤ 0.4 per 40 moves (now 0.46) and you score 99% from +5. - If you don’t pass, repeat the week whose target you missed. Method Games: the latest 2,000 blitz games from the chess.com public API (Mar 11 – Sep 24 2026). Engine: Stockfish 19, 40,000 nodes per position; puzzle lines at 3,000,000. Win% model and accuracy formula: lichess. A **big mistake** loses ≥ 20 win-% in one move. Rating-point costs: 695 Elo per point of expected score, counting only mistakes the opponent cashed in. Trends: Welch z-test against Mar–Jun. Want this for your own games? Send your chess.com username; the report arrives as a link like this one. [Get your Goat Report · $10$5](https://trychessgoat.com/) Analysis by [Stockfish](https://stockfishchess.org/), the free open-source chess engine. Chess Goat is not affiliated with chess.com, or with any player, coach, author or channel quoted or linked in this report.