How I Use Lichess Puzzles to Work on Mistakes

I saw a video by this lady talking about collecting mistakes from your chess games.

---

Credit: Kamryn: This Document Changed My Chess

---

I started thinking: this sounds useful, but I wonder how one could find categories for their chess mistakes? Then it dawned on me, Lichess puzzles. An amazing resource of chess puzzles that are tagged with categories. The user also gets bespoke puzzles based on their rating, so beginners and experts can work on the same theme matched to their level by clicking the same link. Perfect.

So, I ranked the Lichess puzzles by frequency, removed very rare ones, ones likely to have lots of crossover, and merged others I could group. Like, I took out mate in 5 and above. I will forgive myself for missing mate in 5, 6, 7, 8!

I was left with a list of puzzles to use when I work on my mistakes. I can ask myself the questions: where did I go wrong? Was it the endgame, middlegame or opening? What went wrong? What did I miss? It makes it easier to see what I need to work on, and now I know what mistakes to look for! Though I don’t think I’m at the level to spot the German-named tactics, ‘Zwischenzug’ and ‘Zugzwang’, even with review!

Something I notice in the data. We should be seeing every ‘fork’, ‘mate in 1’ and ‘mate in 2’. They are so common that they need to be seen instantly. Everything less frequent, like sacrifices, we might not see as much as they could require some hesitancy and more calculation, and pins might not be seen as often because they are not always forcing moves.

The data could also be used for training by normalising the numbers. For example, dividing them by the least common tactic and rounding down, we can see the most common puzzle, ‘forks’, are 310x more likely than the least common, a ‘castling’ puzzle. So, in theory, they are 310x more useful to be practised. There will be other ways of normalising, but I will leave it up to the reader to do what they like with the raw data.

Here they are listed for ease of use. I won't write what each one means to keep this short, and they will be self-describing when you do a puzzle. They all link to the puzzle theme on Lichess!

---

Mates (998,120)

1. Mate in 1: See Below
2. Mate in 2: 780,261
3. Mate in 3: 190,010
4. Mate in 4: 27,849

Endgames (733,508)

1. Rook: 319,009
2. Pawn: 215,419
3. Bishop: 81,099
4. Queen: 68,791
5. Knight: 49,190

Basic Middlegames (2,270,287)

1. Fork: 805,355
2. Pin: 370,704
3. Discovered Attack: 316,846
   • (Discovered Check): 108,808
4. Deflection (Overload): 262,671
5. Attraction: 216,468
6. Skewer: 136,488
7. Double Check: 31,635
8. X-Ray: 21,312

Advanced Middlegames (1,570,523)

1. Sacrifice: 447,552
2. Defensive Move: 365,461
3. Quiet Move: 249,668
4. Promotion: 143,618
   • (Underpromotion): 1,132
5. Clearance: 79,136
6. In-Between Move (Zwischenzug): 76,344
7. Trapped Piece: 71,863
8. Forced Move (Zugzwang): 61,052
9. Capture the Defender (Undermine): 42,108
10. Interference: 22,580
11. En Passant: 8,546
12. Castling: 2,595
13. *Collinear Move: 8,395 (Newly added! Thanks, Lichess!)*

---

Openings (96,556)
You can also use Lichess puzzles for openings and only select the ones you use. Why would we work on the King’s Pawn opening if we never play it? These openings should, of course, be tailored to you, but here is my repertoire to share. I think it’s also useful for a beginner like me, after completing the practise section on Lichess, to use vibrant examples from real-life games.

1. Queens Pawn: 75,959
2. Modern: 18,729
3. St George: 1,378
4. Carr: 254
5. Kangaroo: 236

---

Common Mate in 1’s (376,561)
Here are the mates in ones. These should be second nature, but everyone needs a refresher.

1. Back Rank: 199,223
2. Pillsbury's: 67,648
3. Opera: 63,942
4. Smothered: 23,091
5. Epaulette: 22,657

Uncommon Mate in 1’s (74,362)
I’ve split these mates from the others, not because they are difficult, but because they are considerably rarer than the others and, as such, might not need as much practise. They still make the list, though, for posterity.

1. Corner: 10,792
2. Hook: 10,169
3. Swallow Tail: 8,398
4. Triangle: 7,854
5. Anastasia's: 7,153
6. Morphy’s: 7,135
7. Arabian: 7,091

---

Total: 6,119,917
More than there are on the database! That will be because some double up or more in certain categories. I wonder how long it would take to complete 6 million puzzles? Seriously though, it is amazing what Lichess have provided, and all for free, long live Lichess!

P.S And yes, I used AI to compile the data, but I wrote it myself. A great use of a great tool!