As chess players, we all know the feeling. You get paired against someone rated 200 points higher, and suddenly you sit up straighter. You calculate a little longer. You second-guess your opening choices. Conversely, when paired down 200 points, it is easy to play quickly, assuming your opponent will eventually blunder.
But does the data support these psychological shifts? Do we actually play more accurately when facing stronger opponents? Do we change our opening repertoires to be more solid or more aggressive? And if we do play better against stronger opponents, how often does it actually translate into an upset victory?
To answer these questions, we analyzed a sample of approximately 15,000 Blitz games (representing over 28,000 individual player performances) using Stockfish 12 per-ply evaluations. We segmented players into 200-point rating bands from 800 to 1600 (Chess.com Blitz equivalents) to see how behavior changes as players climb the rating ladder.
Here is what the data reveals about playing up versus playing down.
1. Accuracy: Do You Play Better Against Stronger Opponents?
The most striking finding in the data is that players are significantly more accurate when playing up.
When we measure accuracy using Average Centipawn Loss (ACPL)—where a lower number means the player's moves more closely matched the engine's top choices—we see a clear "V-shape" curve across the rating differential spectrum.
When a 1000–1200 rated player faces an equally rated opponent, their median ACPL is 118.8. But when that same player faces an opponent rated 200 points higher, their median ACPL drops to just 50.8. This massive improvement in move quality suggests that players concentrate harder, calculate more deeply, and play more solidly when they know they are the underdog.
Interestingly, accuracy also improves when playing down 200 points (median ACPL drops to 85.5 for the 1000–1200 band). This is likely because stronger players are rarely put under severe pressure by much weaker opponents, allowing them to easily find good moves and convert advantages without being forced into complex, error-prone defensive calculations.
Actionable Advice for 800–1200 Players: The data proves you are capable of playing much higher-quality chess than your average game suggests. The focus and calculation depth you summon when facing a 1400-rated player is your true ceiling. If you can learn to treat your 1000-rated peers with the same respect and caution you give to stronger players, your baseline accuracy will improve dramatically.
2. Opening Choices: Do We Play More Aggressively?
A common piece of chess wisdom is that when playing a stronger opponent, you should "complicate the game" and play sharply to create tactical chaos. Conversely, when playing a weaker opponent, you should play solidly and wait for them to self-destruct.
The data shows that players at these levels do exactly the opposite.
When playing up (+200 points), players do not significantly change their opening repertoires. They stick to what they know. However, when playing down (-200 points), players noticeably shift away from sharp, theoretical openings and toward classical, open games.
For example, in the 1200–1400 band:
- In even matchups, players use sharp 1.e4 openings (like the Sicilian, Caro-Kann, or Pirc) in 32.8% of their games.
- When playing down 200 points, their use of these sharp openings collapses to just 21.0%.
- Instead, they pivot toward Flank openings (1.c4, 1.Nf3) and classical 1.e4 e5 setups.
This suggests that stronger players prefer to avoid early tactical landmines when facing weaker opponents, opting instead for positional games where their superior strategic understanding will eventually win out.
Actionable Advice for 1200–1600 Players: If you are the underdog, do not assume you need to play a crazy gambit to win. The data shows that underdogs achieve their best accuracy when playing their standard repertoire. If you are the favorite, the data validates the strategy of shifting toward solid, positional openings (like 1.e4 e5 or 1.d4) to deny the weaker player cheap tactical tricks.
3. The Anatomy of an Upset
If lower-rated players play so much more accurately when facing stronger opponents, how often do they actually win?
The upset rate (the percentage of games where the player rated 200 points lower scores a win or a draw) climbs steadily as players move up the rating ladder.
At the 800–1000 level, a player facing an opponent 200 points stronger only manages to score 25.3% of the time. But by the time players reach the 1200–1400 band, the underdog scores in 33.2% of games—meaning the favorite fails to win one out of every three games.
Why do upsets happen? In our review of the game data, upsets at the 1200–1500 level rarely occur because the underdog slowly outplays the favorite. Instead, they happen because the favorite, perhaps playing too quickly or underestimating the opponent, commits a catastrophic one-move blunder in an otherwise equal position.
Example 1: The Favorite Blunders Early
In this game between a 1543-rated White and a 1222-rated Black, the position is roughly equal out of the opening. Black (the underdog) has played solidly.
Here, Black plays the disastrous ...Nxe4?? (red arrow), completely missing that the knight is adequately defended and the move simply hangs a full piece. The engine evaluates the position at +3.90 for White. The correct, solid move was simply ...c6 (green arrow) to challenge the center. The underdog's concentration lapsed for just one move, and the game was effectively over.
Example 2: The Favorite Overreaches
In this game, a 1215-rated White is facing a 1576-rated Black. White has survived the opening and the position is completely equal.
White, perhaps feeling the pressure of playing a much stronger opponent, lashes out with the aggressive but flawed Nxh7?? (red arrow), sacrificing a knight for a pawn without sufficient compensation. The evaluation swings to -5.61 in favor of Black. The engine prefers the calm, developing move Nc3 (green arrow).
Actionable Advice for All Players: Upsets are rarely the result of brilliant 10-move combinations by the underdog. They are almost always the result of a single, unforced blunder. When playing up, your goal is simply to stay in the game and wait; the pressure of being the favorite often causes the stronger player to overpress and blunder. When playing down, your primary job is blunder-checking. You do not need to play brilliantly to beat a lower-rated player; you only need to play safely.
Conclusion
The rating differential between you and your opponent absolutely changes how you play.
When you are the underdog, fear and respect drive you to play with significantly higher accuracy. When you are the favorite, you tend to steer the game toward calmer, more positional waters to avoid tactical accidents.
The ultimate lesson from the data is one of consistency. If an 1100-rated player can play with the accuracy of a 1300-rated player simply because they are intimidated by their opponent's rating, then the skill is already there. The key to climbing the rating ladder is learning to summon that "playing up" level of focus in every single game, regardless of the number next to your opponent's name.
Data and Methodology
This analysis is based on a sample of 15,000 Blitz games played in March 2025, sourced from the Lichess open database via the Grandmaster Guide MCP API.
Platform Calibration: Because the raw data originates from Lichess, we mapped the Lichess Blitz ratings to approximate Chess.com Blitz ratings to make the advice actionable for Chess.com users. The mapping used is as follows:
- Chess.com 800–1000 ≈ Lichess 1200–1420
- Chess.com 1000–1200 ≈ Lichess 1420–1565
- Chess.com 1200–1400 ≈ Lichess 1565–1705
- Chess.com 1400–1600 ≈ Lichess 1705–1850
Metrics:
- ACPL (Average Centipawn Loss): Calculated per-ply using Stockfish 12 evaluations. Blunders are defined as any move that worsens the evaluation by 300 centipawns or more.
- Rating Differential: Calculated as
Opponent Rating - Player Rating. "Playing Up (+200)" means the opponent was rated between 150 and 250 points higher than the player.
Data Files: The underlying aggregated data used to generate the charts in this article is available in the attached CSV files:
summary_acpl_by_band_diff.csvsummary_opening_aggression.csvsummary_upset_rates.csvsummary_game_length.csv
Chess Coach <Apr 20, 2026>