Blog · Blitz study · October 2026
What stronger players do differently in blitz
We analyzed over 61,000 blitz games from 624 Chess.com players. The analysis includes people who asked for a report in our Reddit thread, randomly chosen rated and titled players, and the top 10 of the blitz leaderboard. As expected, stronger players make fewer blunders and capitalize on their opponents' mistakes more effectively, but some things remain surprisingly consistent across all levels.
Enter it and every chart below marks the players at your level.
With time on the clock, strong players almost never blunder
A blunder is a move that loses more than two pawns of evaluation. With more than half the clock left, the median player under 600 blunders on 10.8% of moves, about one in 9. Grandmasters blunder on 1.0% and the top 10 on 0.5%, about one in 200.
Blunders with more than half the clock left
Share of moves, median player in each group
elohell.app/blog/blitz-study · 61,000 blitz games
Under 600: the median player blunders on 10.8% of moves with more than half the clock left, about one move in 9.
600–900: the median player blunders on 8.2% of moves with more than half the clock left, about one move in 12.
900–1200: the median player blunders on 6.5% of moves with more than half the clock left, about one move in 15.
1200–1500: the median player blunders on 6.1% of moves with more than half the clock left, about one move in 16.
1500–1800: the median player blunders on 5.2% of moves with more than half the clock left, about one move in 19.
1800–2100: the median player blunders on 4.2% of moves with more than half the clock left, about one move in 20.
2100–2400: the median player blunders on 3.5% of moves with more than half the clock left, about one move in 30.
FMs and NMs: the median player blunders on 1.8% of moves with more than half the clock left, about one move in 60.
IMs: the median player blunders on 1.5% of moves with more than half the clock left, about one move in 70.
GMs: the median player blunders on 1.0% of moves with more than half the clock left, about one move in 100.
The top 10: the median player blunders on 0.5% of moves with more than half the clock left, about one move in 200.
Time trouble is where strong players blunder
With under 10% of the clock left, every group blunders more, and the jump grows with rating: 1.7 times as often under 600, 8.6 times for grandmasters and 16.6 times for the top 10.
Blunders by clock
Share of moves, median player. Under each group: how many times as often they blunder with under 10% left.
elohell.app/blog/blitz-study · 61,000 blitz games
Under 600: 10.8% of moves with more than half the clock left, 18.2% with under 10% left, 1.7 times as often.
600–900: 8.2% of moves with more than half the clock left, 16.0% with under 10% left, 2.0 times as often.
900–1200: 6.5% of moves with more than half the clock left, 14.8% with under 10% left, 2.2 times as often.
1200–1500: 6.1% of moves with more than half the clock left, 13.2% with under 10% left, 2.3 times as often.
1500–1800: 5.2% of moves with more than half the clock left, 13.4% with under 10% left, 2.6 times as often.
1800–2100: 4.2% of moves with more than half the clock left, 11.4% with under 10% left, 2.7 times as often.
2100–2400: 3.5% of moves with more than half the clock left, 11.8% with under 10% left, 3.3 times as often.
FMs and NMs: 1.8% of moves with more than half the clock left, 10.0% with under 10% left, 6.2 times as often.
IMs: 1.5% of moves with more than half the clock left, 8.7% with under 10% left, 5.9 times as often.
GMs: 1.0% of moves with more than half the clock left, 8.7% with under 10% left, 8.6 times as often.
The top 10: 0.5% of moves with more than half the clock left, 7.2% with under 10% left, 16.6 times as often.
Lower-rated games are often decided by move 15
After 10 moves, 25% of games under 600 are still level, with neither side a pawn up; for grandmasters it is 82%. By move 15, 69% of games under 600 are already decided, against 8% for grandmasters and 4% for the top 10.
Games already decided
Share of games where one side is three pawns or more ahead, or the game is over, median player
elohell.app/blog/blitz-study · 61,000 blitz games
Under 600: after 10 moves 25% of games are still level and 44% are decided. By move 15, 69% are decided.
600–900: after 10 moves 38% of games are still level and 28% are decided. By move 15, 53% are decided.
900–1200: after 10 moves 47% of games are still level and 20% are decided. By move 15, 42% are decided.
1200–1500: after 10 moves 52% of games are still level and 16% are decided. By move 15, 35% are decided.
1500–1800: after 10 moves 57% of games are still level and 12% are decided. By move 15, 29% are decided.
1800–2100: after 10 moves 61% of games are still level and 10% are decided. By move 15, 25% are decided.
2100–2400: after 10 moves 68% of games are still level and 6% are decided. By move 15, 19% are decided.
FMs and NMs: after 10 moves 75% of games are still level and 4% are decided. By move 15, 13% are decided.
IMs: after 10 moves 77% of games are still level and 2% are decided. By move 15, 12% are decided.
GMs: after 10 moves 82% of games are still level and 2% are decided. By move 15, 8% are decided.
The top 10: after 10 moves 88% of games are still level and 2% are decided. By move 15, 4% are decided.
Stronger players find more tactics
When a move won material by force, stronger players found it more often: forks 54% under 600 to 86% for the top 10, pins 54% to 92%, skewers 55% to 90%. Free pieces are the easiest: 80% are taken under 600.
Tactics found
Share of chances taken, median player in each group
Free pieces
Forks
Pins
Skewers
| Free pieces | Forks | Pins | Skewers | |
|---|---|---|---|---|
| <600 | 80% | 54% | 54% | 55% |
| 600–900 | 84% | 63% | 58% | 62% |
| 900–1200 | 87% | 65% | 67% | 70% |
| 1200–1500 | 88% | 67% | 67% | 65% |
| 1500–1800 | 88% | 69% | 67% | 70% |
| 1800–2100 | 91% | 73% | 70% | 72% |
| 2100–2400 | 92% | 75% | 75% | 75% |
| Titled players | ||||
| FM/NM | 93% | 79% | 89% | 80% |
| IM | 94% | 81% | 81% | 83% |
| GM | 95% | 83% | 90% | 85% |
| Top 10 | 97% | 86% | 92% | 90% |
elohell.app/blog/blitz-study · 61,000 blitz games
Under 600: free pieces 80% found, forks 54%, pins 54%, skewers 55%.
600–900: free pieces 84% found, forks 63%, pins 58%, skewers 62%.
900–1200: free pieces 87% found, forks 65%, pins 67%, skewers 70%.
1200–1500: free pieces 88% found, forks 67%, pins 67%, skewers 65%.
1500–1800: free pieces 88% found, forks 69%, pins 67%, skewers 70%.
1800–2100: free pieces 91% found, forks 73%, pins 70%, skewers 72%.
2100–2400: free pieces 92% found, forks 75%, pins 75%, skewers 75%.
FMs and NMs: free pieces 93% found, forks 79%, pins 89%, skewers 80%.
IMs: free pieces 94% found, forks 81%, pins 81%, skewers 83%.
GMs: free pieces 95% found, forks 83%, pins 90%, skewers 85%.
The top 10: free pieces 97% found, forks 86%, pins 92%, skewers 90%.
About 70% of winning positions get converted until master level
Once two pawns up, the median player wins 64% of the time under 600 and between 67% and 71% from 600 to 2100. Titled players convert 74% to 79% and the top 10 91%, partly because their opponents are weaker: by a median 59 points for grandmasters and 190 for the top 10.
Winning positions converted
Share of games won after reaching a position two pawns ahead, median player. Under each group: median rating edge over the opponent.
elohell.app/blog/blitz-study · 61,000 blitz games
Under 600: the median player converts 64% of winning positions. Their median opponent was rated about the same as them.
600–900: the median player converts 67% of winning positions. Their median opponent was rated about the same as them.
900–1200: the median player converts 69% of winning positions. Their median opponent was rated about the same as them.
1200–1500: the median player converts 71% of winning positions. Their median opponent was rated about the same as them.
1500–1800: the median player converts 70% of winning positions. Their median opponent was rated about the same as them.
1800–2100: the median player converts 71% of winning positions. Their median opponent was rated about the same as them.
2100–2400: the median player converts 74% of winning positions. Their median opponent was rated about the same as them.
FMs and NMs: the median player converts 74% of winning positions. Their median opponent was rated 18 points below them.
IMs: the median player converts 78% of winning positions. Their median opponent was rated 22 points below them.
GMs: the median player converts 79% of winning positions. Their median opponent was rated 59 points below them.
The top 10: the median player converts 91% of winning positions. Their median opponent was rated 190 points below them.
Lower-rated players get checkmated. Stronger players resign.
Under 600, 37% of lost games end in checkmate; for grandmasters it is 12%. Stronger players resign instead: 30% of losses under 600 and 62% for the top 10.
How lost games end
Share of all lost games in each group
elohell.app/blog/blitz-study · 61,000 blitz games
Under 600: 37% of losses by checkmate, 26% on time, 30% by resignation.
600–900: 37% of losses by checkmate, 16% on time, 38% by resignation.
900–1200: 28% of losses by checkmate, 20% on time, 46% by resignation.
1200–1500: 27% of losses by checkmate, 18% on time, 49% by resignation.
1500–1800: 23% of losses by checkmate, 22% on time, 49% by resignation.
1800–2100: 20% of losses by checkmate, 25% on time, 51% by resignation.
2100–2400: 22% of losses by checkmate, 24% on time, 51% by resignation.
FMs and NMs: 16% of losses by checkmate, 27% on time, 56% by resignation.
IMs: 14% of losses by checkmate, 25% on time, 58% by resignation.
GMs: 12% of losses by checkmate, 35% on time, 52% by resignation.
The top 10: 13% of losses by checkmate, 25% on time, 62% by resignation.
Stronger players hold drawn endgames far more often
Under 600, only 20% of rook endgames that start drawn are held, against 44% for grandmasters and 55% for the top 10. Winning endgames are easier to keep at every level. Held means the player didn't make it worse; the rules are in How we did this.
Drawn endgames held
Endgames that started drawn, with neither side two pawns ahead: share where the player never dropped to losing. All of the group's endgames of that type.
Queen endings
Rook endings
Minor-piece endings
| Queen | Rook | Minor | |
|---|---|---|---|
| <600 | 24% | 20% | 18% |
| 600–900 | 24% | 23% | 25% |
| 900–1200 | 28% | 25% | 29% |
| 1200–1500 | 31% | 26% | 26% |
| 1500–1800 | 31% | 29% | 31% |
| 1800–2100 | 37% | 29% | 31% |
| 2100–2400 | 37% | 36% | 35% |
| Titled players | |||
| FM/NM | 43% | 41% | 43% |
| IM | 49% | 42% | 45% |
| GM | 48% | 44% | 59% |
| Top 10 | 57% | 55% | 58% |
elohell.app/blog/blitz-study · 61,000 blitz games
Winning endgames held
Share of endgames that started two pawns or more ahead where the player stayed winning. All of the group's endgames of that type. There were fewer than 40 minor-piece endgames for the top 10 players, so those have been excluded because of the small sample size.
Queen endings
Rook endings
Minor-piece endings
| Queen | Rook | Minor | |
|---|---|---|---|
| <600 | 70% | 57% | 69% |
| 600–900 | 66% | 59% | 66% |
| 900–1200 | 66% | 59% | 61% |
| 1200–1500 | 66% | 60% | 70% |
| 1500–1800 | 63% | 59% | 67% |
| 1800–2100 | 67% | 63% | 72% |
| 2100–2400 | 66% | 61% | 72% |
| Titled players | |||
| FM/NM | 76% | 67% | 76% |
| IM | 72% | 68% | 80% |
| GM | 73% | 74% | 77% |
| Top 10 | 76% | 79% | – |
elohell.app/blog/blitz-study · 61,000 blitz games
Endgame is the least accurate phase
Share of players whose endgame accuracy is lower than their opening and middlegame accuracy
elohell.app/blog/blitz-study · 61,000 blitz games
Under 600: drawn endgames held, queen 24%, rook 20%, minor-piece 18%. Winning endgames held, queen 70%, rook 57%, minor-piece 69%.
600–900: drawn endgames held, queen 24%, rook 23%, minor-piece 25%. Winning endgames held, queen 66%, rook 59%, minor-piece 66%.
900–1200: drawn endgames held, queen 28%, rook 25%, minor-piece 29%. Winning endgames held, queen 66%, rook 59%, minor-piece 61%.
1200–1500: drawn endgames held, queen 31%, rook 26%, minor-piece 26%. Winning endgames held, queen 66%, rook 60%, minor-piece 70%.
1500–1800: drawn endgames held, queen 31%, rook 29%, minor-piece 31%. Winning endgames held, queen 63%, rook 59%, minor-piece 67%.
1800–2100: drawn endgames held, queen 37%, rook 29%, minor-piece 31%. Winning endgames held, queen 67%, rook 63%, minor-piece 72%.
2100–2400: drawn endgames held, queen 37%, rook 36%, minor-piece 35%. Winning endgames held, queen 66%, rook 61%, minor-piece 72%.
FMs and NMs: drawn endgames held, queen 43%, rook 41%, minor-piece 43%. Winning endgames held, queen 76%, rook 67%, minor-piece 76%.
IMs: drawn endgames held, queen 49%, rook 42%, minor-piece 45%. Winning endgames held, queen 72%, rook 68%, minor-piece 80%.
GMs: drawn endgames held, queen 48%, rook 44%, minor-piece 59%. Winning endgames held, queen 73%, rook 74%, minor-piece 77%.
The top 10: drawn endgames held, queen 57%, rook 55%, minor-piece 58%. Winning endgames held, queen 76%, rook 79%, minor-piece –.
Ratings swing about 110 points on luck alone
Elo hell is the feeling of being stuck at a rating. Over their newest 97 blitz games, the median player's rating moved 110 points between its high and its low; a coin flip with the same 8-point steps swings 112. A 100-point slide or climb says little about whether you got better or worse.
Playing above your rating barely changed that. We ranked players by how their accuracy in their first 50 games compared with what's typical at their rating: over the next 50, the best third gained a median of 12 points and the worst third 8, with almost the same ranges.
Rating change over the next 49 games
Players without a title, split into thirds by how their accuracy in their earlier games compared with what's typical at their rating. Median change in rating points; the thin line is the 95% range.
elohell.app/blog/blitz-study · 61,000 blitz games
Players don't seem to tilt that much after a single loss
Within a sitting, with games no more than 30 minutes apart, the median player's blunder rate after a loss is within 0.4 percentage points of their rate after a win or a draw, in every group.
Pooled over all players, losses do raise it a little: +2.2% after one loss and +4.5% after three in a row, against each player's own rate. For someone who blunders on 7% of moves, that is about 7.3% after three. Small, but both are outside chance: their 95% ranges stay above zero.
Blunders in the next game, same sitting
Share of moves, median player
elohell.app/blog/blitz-study · 61,000 blitz games
Blunders after losses in a row, same sitting
All 618 players with enough games: relative change in blunder rate compared with each player's own rate after a win or a draw. The thin line is the 95% range.
elohell.app/blog/blitz-study · 61,000 blitz games
Under 600: 13.1% of moves blundered in the next game of a sitting after a loss, 12.4% after a win or draw.
600–900: 10.1% of moves blundered in the next game of a sitting after a loss, 10.0% after a win or draw.
900–1200: 8.2% of moves blundered in the next game of a sitting after a loss, 8.1% after a win or draw.
1200–1500: 7.8% of moves blundered in the next game of a sitting after a loss, 7.9% after a win or draw.
1500–1800: 6.9% of moves blundered in the next game of a sitting after a loss, 7.2% after a win or draw.
1800–2100: 6.4% of moves blundered in the next game of a sitting after a loss, 6.1% after a win or draw.
2100–2400: 5.6% of moves blundered in the next game of a sitting after a loss, 5.3% after a win or draw.
FMs and NMs: 4.4% of moves blundered in the next game of a sitting after a loss, 4.0% after a win or draw.
IMs: 3.9% of moves blundered in the next game of a sitting after a loss, 3.4% after a win or draw.
GMs: 3.1% of moves blundered in the next game of a sitting after a loss, 3.0% after a win or draw.
The top 10: 2.6% of moves blundered in the next game of a sitting after a loss, 2.0% after a win or draw.
How we did this
Players. We included 277 Chess.com players who shared their username in our Reddit thread in October 2026 and had at least 50 blitz games. To fill out the groups we drew random players: on October 15, 25 from each of 1500–1800, 1800–2100 and 2100–2400 and 25 each of GMs, IMs, and FMs or NMs from Chess.com's lists of titled players; on October 16, 40 more in each of 900–1200, 1200–1500, 1500–1800, 1800–2100 and 2100–2400. The untitled ones were picked from the opponents in these games, and every one of them had played in the last 30 days and had at least 100 blitz games. A few had too few recent blitz games and dropped out; the table below has the final counts. Last, the top 10 of Chess.com's blitz leaderboard on October 15, 2026.
Games. Each player's newest 100 blitz games, or fewer when their newest 1,000 games of any kind held fewer. A game is blitz when the base time plus 40 times the increment comes to at least 3 minutes and under 8, the rule Lichess uses. Stockfish 17.1 evaluated every move at depth 16.
Groups. Players without a title are grouped by the median of their own Chess.com blitz rating across those games. Titled players are grouped by title, whatever their rating. Every figure is the median player in the group, and a player counts toward a figure only with enough of what it measures, for example 100 moves with more than half the clock left or 10 winning positions. Endgames are the exception: once split by type and start they are too rare per player, so each endgame figure pools all of the group's endgames of that type and start, at least 40. How we decide an endgame was held is under Endgames below.
Endgames. An endgame starts once six or fewer queens, rooks, bishops and knights are left, counting both sides, and is grouped by the strongest piece left, so a rook ending can still have bishops or knights on it. Each one is sorted by where it started, using the engine's evaluation from the player's side: winning (two pawns or more ahead), losing (two pawns or more behind) or drawn. It counts as held unless one of the player's moves cost a pawn or more and moved it into a worse band. Endgames that started lost are left out, since there was nothing left to lose: 33% of rook endgames under 600 start lost, against 9% for the top 10. Held uses the evaluation, not the result, so a later loss on time doesn't count against it. A player needs at least four moves in an endgame type for it to count, and pawn endings are too rare to split this way.
Limits. The thread players chose to take part. To check that, we compared them with the randomly chosen players at the same level: in 900–1200, 1200–1500, 1500–1800 and 1800–2100, the two sets of players were within 4 percentage points of each other on blunders, winning positions converted, forks found and rook endgames held. The highest groups are small: the top 10 is ten players. Rating and these measures move together, which doesn't show that changing one changes the other.
| Group | Players | Games | Median rating |
|---|---|---|---|
| Under 600 | 86 | 8,166 | 403 |
| 600–900 | 48 | 4,693 | 756 |
| 900–1200 | 108 | 10,580 | 1058 |
| 1200–1500 | 76 | 7,501 | 1325 |
| 1500–1800 | 87 | 8,602 | 1686 |
| 1800–2100 | 78 | 7,769 | 1921 |
| 2100–2400 | 62 | 6,179 | 2258 |
| FMs and NMs | 23 | 2,265 | 2474 |
| IMs | 23 | 2,279 | 2670 |
| GMs | 23 | 2,248 | 2820 |
| The top 10 | 10 | 1,000 | 3306 |
See these numbers for your own games
Elo Hell runs the same analysis on your games. Connect your Chess.com or Lichess account and Insights shows your blunders by clock, how often you convert winning positions, which tactics you find and miss, and how your games end.
