The 10 openings that lose the most rating at 1200–1600
A list of “worst openings” is easy to fake. You pick ten names that sound dubious, invent percentages, and hope nobody checks. Chess Wizard does not yet have its own opening book — the insights page stays silent on repertoires rather than guess — so this post is not a report on our users.
It is a report on a public dataset: 15,013 named lines from a 20 million game subset of rated Lichess games played in 2024. Each line is a specific sequence of five moves. The rows below are the ones whose players averaged 1200–1600, with enough games that a percentage is not a streak. Expected score is wins plus half the draws, for the side that chose the idea.
That last clause matters. A trap that only appears in the data once someone has already fallen into it will look horrific, which is the point of several of these. A line you can decline will look milder. Read the moves, not just the name.
The ten
| Line | Who is leaking | Expected score | Games | Typical rating |
|---|---|---|---|---|
| Blackburne Shilling, White takes on e5 | White | 7% | 1,641 | 1356 |
| Damiano Defense | Black | 25% | 843 | 1352 |
| Fried Liver, Black takes on d5 | Black | 31% | 11,183 | 1387 |
| Anti-Fried Liver (3…h6) | Black | 40% | 7,036 | 1438 |
| Ruy Lopez Exchange, White grabs e5 | White | 41% | 4,008 | 1477 |
| Four Knights / Italian, the fork trick mishandled | White | 42% | 4,252 | 1511 |
| Hillbilly Attack vs the Caro-Kann | White | 42% | 3,288 | 1479 |
| Philidor with …Bg4 | Black | 43% | 5,929 | 1550 |
| Center Game | White | 45% | 2,511 | 1418 |
| Wayward Queen Attack | White | 50% | 3,453 | 1393 |
White’s baseline in this band is a little over 50%. Anything near a coin flip as White is already a leak. Anything in the 20s as Black is a crisis.
1. Blackburne Shilling, White takes on e5
1.e4 e5 2.Nf3 Nc6 3.Bc4 Nd4 4.Nxe5?? Qg5. White’s expected score in the games that reached 4…Qg5 and 5.Nxf7 Qxg2 is 7%. That is not a style. That is a trap with a name.
3…Nd4 is not a great move if White declines it. The main alternative in the same dataset, 4.c3, scores about 51% for White. Take the knight, or don’t play 3.Bc4 without knowing this exists.
2. Damiano Defense
1.e4 e5 2.Nf3 f6. The games that continue 3.Nxe5 fxe5 4.Qh5+ Ke7 5.Qxe5+ Kf7 give Black 25%. The move 2…f6 takes away the knight’s best square and opens the h5–e8 diagonal. Both things are free for White.
Play 2…Nc6. If you have been playing 2…f6 because it “protects the pawn,” it does not.
3. Fried Liver, Black takes on d5
1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 d5 5.exd5 Nxd5. Eleven thousand games in this band. Black’s expected score: 31%.
5…Nxd5 is the natural recapture and it is the one the Fried Liver is waiting for. 5…Na5, the Polerio, is the move that makes 4.Ng5 a lot less fun for White. You do not need the rest of the theory. You need not to take with the knight.
This is the highest-volume leak on the list. It is not rare. It is the Italian, which is what this rating band actually plays.
4. Anti-Fried Liver (3…h6)
1.e4 e5 2.Nf3 Nc6 3.Bc4 h6. Black’s idea is to stop Ng5. White’s idea is to play d4 and develop while Black has spent a tempo on a pawn move that does not develop. Expected score for Black after the typical 4.d4 exd4 5.Nxd4 Nxd4: 40%, over 7,000 games.
If you are scared of the Fried Liver, learn 5…Na5. Do not give up a developing move on move three to avoid a line you can just not enter.
5. Ruy Lopez Exchange, White grabs e5
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6 dxc6 5.Nxe5. Black plays 5…Qd4, forks the knight and the e-pawn, and White’s expected score is 41%. The Exchange Lopez with 5.d3, in the same band, is a much more ordinary 48%.
The extra pawn is not free. If you play the Exchange, play it as a structure, not as a snatch.
6. The fork trick, mishandled
1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Nc3 Nxe4. Black’s 4…Nxe4 is the fork trick: after 5.Nxe4 comes 5…d5, hitting bishop and knight. White’s expected score in the games that got there: 42%.
The trick is not a refutation of the Four Knights. It is a tactic you have to have seen once. After 5…d5 you give a piece back and you are fine. The leak is White who has not seen it, hanging the piece or panicking into a worse structure.
7. Hillbilly Attack vs the Caro-Kann
1.e4 c6 2.Bc4 d5. White’s bishop is looking at a pawn chain that is about to open onto it. The popular continuation 3.exd5 cxd5 4.Bb5+ Nc6 5.Bxc6+ bxc6 gives White 42% — you doubled the pawns and still scored worse than a coin flip.
Against the Caro, 2.d4 is the move. 2.Nc3 and 2.Nf3 are also moves. 2.Bc4 is a bishop you will have to move again.
8. Philidor with …Bg4
1.e4 e5 2.Nf3 d6 3.d4 Bg4 4.dxe5 Bxf3 5.Qxf3 dxe5. Black gives up the bishop pair and a lot of the light squares to keep a pawn. Expected score 43%, and this is a common Philidor — almost 6,000 games.
The Philidor is playable if you keep the bishop and fight for …e4 or …f5 later. The version that trades it on f3 for a pawn that was not in danger is the version that leaks.
9. Center Game
1.e4 e5 2.d4 exd4 3.Qxd4 Nc6 4.Qd1 Nf6 5.Nc3 Bb4. The queen came out, got kicked, and went home. White’s expected score: 45%. You have lost two tempi and Black is developing with threats.
If you want an open centre, the Scotch (2.Nf3 Nc6 3.d4) does the same job without parking the queen on d4.
10. Wayward Queen Attack
1.e4 e5 2.Qh5 Nc6 3.Bc4 g6 4.Qf3 Nf6 5.c3 Bg7. White’s expected score in the most-played continuation: 50%, over 3,400 games, average rating 1393. That is the first-move advantage, gone, in a line whose whole idea is to mate on f7 before Black knows how to stop it.
At 800 it scores. By 1400 the people who used to hang f7 have stopped hanging f7, and you are a tempo down in an open game. Retire it the way you retire Scholar’s Mate: fondly, and then not at all.
A line that is not on the list
The Stafford Gambit — 1.e4 e5 2.Nf3 Nf6 3.Nxe5 Nc6 — scores 57% for Black in the main 5.Nc3 Bc5 continuation at average rating 1553, over 5,600 games. YouTube will tell you it is a beginner trap. The database, in this band, will tell you it is printing rating for the player who chose it.
That is the whole argument for looking at numbers. Reputation and result diverge, in both directions.
How to read this if you play these lines
One game proves nothing. Fifteen is the floor before a personal score is worth acting on. If two or three of these ten show up in your last twenty games, that is the leak, and it is cheaper to stop playing them than it is to study a new middlegame.
A league-wide table is a map of where rating goes in general. Your own games are the map of where yours goes. Use this list to know what to look for. Do not use it as a personality.
Data: Varshney, S. (2026), Chess Multiverse Opening Intelligence Database (20M Lichess subset), doi:10.5281/zenodo.19307100, CC BY-SA 4.0. Lines are five-move sequences, not opening families from move one.