To improve your chess rating, stop losing games you shouldn't lose before you try to win games you can't yet win. At most club and online levels that means three things, in this order: stop hanging pieces, solve tactics in volume, and review every serious game you lose. Play slower time controls so you can actually use what you train. Openings and deep theory come much later than most players think.
None of this guarantees a number. Rating progress is uneven, it varies a lot from person to person, and plateaus of months are normal. What follows is a priority order, not a promise.
What actually moves your rating
A rating is just a running score of results. Points come from games, and games below master level are mostly decided by mistakes, not by brilliant plans. So the fastest gains come from reducing the size and frequency of your mistakes.
- Not hanging pieces. Losing a piece for nothing ends most games at lower ratings. A simple check before every move fixes more games than any opening book.
- Tactics volume. Pattern recognition comes from repetition. Solving many puzzles at a level where you get most of them right trains you to see forks, pins and mates during a game.
- Slower time controls. You can't apply a checklist with 10 seconds on the clock. Longer games give you time to think, and they produce games worth reviewing.
- Game review. Every loss contains the lesson you need next. Finding the move where the game turned is more useful than another hour of random study.
Everything else (opening repertoire, positional theory, endgame technique) matters, but its weight grows as your rating rises.
The training plan by rating band
The bands below use rough online-style ratings. They are a guide to emphasis, not hard boundaries, and different sites rate differently, so a 1200 on one site is not a 1200 everywhere. For what each band says about your level, see what is a good chess Elo?
| Rating band | Main problem | Top priority | Weekly focus | Skip for now |
|---|---|---|---|---|
| Under 1000 | Pieces left undefended; missed one-move threats | Blunder check before every move | Easy puzzles daily, rapid games, review each loss for the first hung piece | Opening memorisation |
| 1000 to 1500 | Two-move tactics missed for both sides | Tactics volume: forks, pins, skewers, back-rank mates | Daily themed puzzles, slower rapid or classical games, basic king-and-pawn endgames | Long opening lines past move 8 to 10 |
| 1500 to 2000 | Calculation errors and drifting plans in quiet positions | Deeper calculation and honest game analysis | Harder multi-move puzzles, annotate your own games before using an engine, rook endgames, a stable repertoire | Switching openings every week |
Under 1000: stop giving material away
At this level, games are usually lost by the player who hangs the last piece. Before every move, ask three questions: what did my opponent's last move threaten, are any of my pieces undefended, and what does my move allow? It feels slow at first. That is the point.
Pair this with short daily sessions of easy puzzles. Pick a difficulty where you solve most of them, because the goal is seeing common patterns fast, not grinding through problems you can't solve.
Play rapid rather than bullet. After each loss, find the first move where you gave away material and note why you missed it.
1000 to 1500: build tactical vision
Players here have stopped hanging pieces in one move but still miss two-move ideas: a knight fork after a check, a pin that wins a piece, a back-rank mate. This is where tactics volume pays off most.
Train by theme. A week on fork puzzles teaches you to spot the knight landing squares that hit king and rook together, and you will start seeing them in your own games. Add basic endgames from the endgame guide, especially king-and-pawn positions, since many won games are thrown away there.
For openings, learn principles and the first few moves of one line with White and one defence for each of 1.e4 and 1.d4. That is enough.
1500 to 2000: calculate and analyse honestly
By now tactics alone aren't enough, because your opponents see the same short tricks. Games are decided by calculation errors three or four moves deep and by drifting in quiet positions without a plan.
Solve harder puzzles where you must see the whole line before moving. Analyse your serious games yourself first: write down what you thought at each critical moment, then compare with the engine. Our guide to analysing a chess game walks through that process. A stable opening repertoire helps now, because you'll face the same positions repeatedly and can learn from them.
How rating maths works
Knowing the maths helps you set realistic expectations and choose your opponents. Under the Elo system, your expected score against an opponent is:
E = 1 / (1 + 10^((Opponent rating − Your rating) / 400))
After the game your rating changes by K × (S − E), where S is 1 for a win, 0.5 for a draw and 0 for a loss. The full explanation is in how Elo rating works.
K is the speed of change. Under the FIDE Rating Regulations (section 8.3.3), K is 40 for a player new to the list until they have completed events with at least 30 games, and for players until the end of the year of their 18th birthday while rated under 2300. It is 20 for other players under 2400, and 10 once a player has reached 2400.
Worked example: what a win against a stronger player is worth
Take a 1200-rated player with K = 20. We computed each value with the formula above in a short Node.js script, so you can reproduce it exactly:
const expected = (me, opp) => 1 / (1 + 10 ** ((opp - me) / 400));
const change = (me, opp, score, K) => K * (score - expected(me, opp));
// change(1200, 1400, 1, 20) -> 15.19
| Opponent | Expected score | Win | Draw | Loss |
|---|---|---|---|---|
| 1000 | 0.760 | +4.8 | −5.2 | −15.2 |
| 1200 | 0.500 | +10.0 | 0.0 | −10.0 |
| 1400 | 0.240 | +15.2 | +5.2 | −4.8 |
| 1600 | 0.091 | +18.2 | +8.2 | −1.8 |
With K = 40, every number doubles: a win against a 1400 is worth about +30.4.
Three lessons come out of this table:
- One upset win against a player 200 points stronger is worth about three wins against a player 200 points weaker (+15.2 against +4.8).
- Losing to weaker players is expensive. A loss to a 1000 costs −15.2, wiping out an upset win. This is why cutting blunders matters more than chasing brilliancies.
- To gain rating against a pool, you need to beat the expected score. Against opponents 200 points higher you only need to score more than about 24% to gain. Against equal opponents you need more than 50%.
Try your own numbers in the Elo calculator. It uses the same generic formula, so it won't apply FIDE's rounded lookup table or its 400-point rule, under which rating gaps over 400 are counted as 400 for players rated below 2650.
Online sites work differently. Lichess, for example, uses the Glicko-2 system, not plain Elo, so your exact gains and losses there won't match this table, but the principle is the same: results above expectation raise your rating.
A simple weekly routine
If you have about five hours a week, this split follows the priorities above:
- Puzzles, 20 minutes a day. Themed at first, mixed later. Accuracy before speed.
- Two or three slower games a week. Rapid 15+10 or longer online, or a club game. Use the blunder check every move.
- Review every loss. Find the turning point yourself, then confirm it in Game Review. Write one sentence about what to do differently.
- One endgame session a week. Start with king and pawn, then rook endgames.
Keep a short log of the lesson from each loss. Patterns in your own mistakes tell you what to train next more reliably than any generic plan, including this one.
Why progress isn't guaranteed
Ratings measure results against a pool, and the pool changes. Ratings rise in jumps and stall for long stretches, and a bad week can cost a month of gains. Age, available time, how you train and how often you play all affect the pace, and nobody can honestly promise a gain of a fixed number of points by a date.
What you can control is the process: fewer blunders, more patterns, slower games and honest review. If your rating is flat but your losses come from harder mistakes than before, you are improving.
Frequently asked questions
How long does it take to improve your chess rating?
There is no reliable fixed timeline. Some players gain hundreds of points in their first months because they stop hanging pieces; others plateau for a year at the same level. Consistent training and review make progress more likely, but the pace varies widely.
Do puzzles really improve chess rating?
Puzzles train pattern recognition, which is what decides most games below master level. They help most when you solve at a difficulty where you get most right and then actually look for the same patterns in your games. Puzzles without playing and reviewing games transfer less.
Should I play blitz or rapid to improve?
For improvement, slower games are better. Rapid and classical give you time to apply a blunder check and calculate, and they produce games worth reviewing. Blitz is fine for fun, but it rewards fast habits, including bad ones.
Why does my rating go down when I study?
Trying new ideas, openings or a slower thinking process often costs results at first, before it becomes habit. Normal variance also produces losing streaks. Judge progress over months and look at the quality of your mistakes, not only the number.
How many points do I get for beating a higher-rated player?
Under the Elo formula with K = 20, a 1200 player gains about 15.2 points for beating a 1400 and about 18.2 for beating a 1600. With K = 40, those numbers double. Online sites using Glicko-2 calculate changes differently.