How to read a TFT tier list (and when to ignore it)
The numbers come from high elo:
Set 18 Patch 18.5 77,232 high-elo games Games from 6 Oct to 20 Oct
A tier list is a summary of thousands of games, not a ranking of who is “best”. Used well, it tells you which comps have been finishing high lately. Used badly, it sends you into a comp that is strong in the data and weak in your hands. This guide explains what each number on our tier list means and where it can mislead you.
Where the numbers come from
Every figure here comes from 77,232 ranked games played by Platinum-to-Challenger players. For each game we take the final board of every player, group similar boards into comps (same carry, mostly the same units) and record the placement each one finished in. Nothing is estimated or copied from another site: if a comp has few games, its numbers are noisy and we say so.
What each metric means
- Average placement: the mean of the finishing positions, from 1 to 8. A random board averages 4.5, so anything clearly below that is doing well. It is the number we sort by (smoothed, see below).
- Top 4: how often the comp finishes in the top half. In ranked, that’s where you gain points, so it is the most useful metric if your goal is climbing steadily.
- Win rate (1st place): how often it wins the game. Random is 12.5 %. A comp with a high win rate and a mediocre top 4 is a high-risk, high-reward pick: it wins a lot when it works and collapses when it doesn’t.
- Play rate: how many boards ended in that comp. A comp that few people play can look great just because the few players who pick it are good at it.
How tiers are assigned
Tiers are relative, not absolute. We sort every comp by its smoothed average placement and cut the list: the best 15 % is S, the next 30 % is A, the next 30 % is B and the rest is C. That means there will always be S-tier comps, even in a flat patch where all of them are similar. The tier tells you where a comp stands against the others this week, not that it is guaranteed to win.
“Smoothed” means that a comp with few games is pulled towards the 4.5 average, as if it had 30 extra mid-table games. Without this, a comp that happened to win twice in ten games would top the list. It’s the same reason the order inside a tier uses a smoothed win rate instead of the raw one.
What the tiers look like this set
Weighted by games, for comps with at least 50 games:
| Tier | Comps | Games | Top 4 | Avg. placement | 1st place |
|---|---|---|---|---|---|
| S | 72 | 111,360 | 66% | 3.63 | 22% |
| A | 112 | 261,101 | 53% | 4.33 | 13% |
| B | 114 | 182,879 | 42% | 4.89 | 9% |
| C | 92 | 53,597 | 29% | 5.61 | 6% |
390 comps have enough games to appear here. The gap between S and C looks huge, and part of it is real, but part of it is an artifact of how final boards are recorded (see survivorship bias below): the table measures games where the board got completed, not what happens if you choose the comp at stage 2. In any single game you can still finish top 4 with a C-tier comp and bottom 4 with an S-tier one.
Five ways a tier list can mislead you
- Survivorship bias. The final board only exists if the player survived long enough to build it. Comps with expensive carries are only completed in games that were already going well, so their numbers look better than their real power when you attempt them from behind.
- Sample size. Under about 50 games the numbers are mostly noise. That’s why comp pages with fewer games stay on the site but are kept out of search results.
- Rank. The data is Platinum-to-Challenger. Comps that need precise positioning or fast decisions work better in the hands of players who do it every game. At lower ranks, forgiving comps usually outperform the table.
- Contestation. A comp is strong when you are alone on it. If three other players are on it too, there are not enough copies for anyone (see scouting). No tier list can show that for your lobby.
- Patch timing. The first days after a patch have little data and players are still experimenting. Wait a few days before trusting a fresh ranking, and check the patch notes to see why something moved.
How to actually use it
- Pick two or three comps from S and A that you are comfortable playing, not the single top one.
- Prefer comps with plenty of games and a solid top 4 over the ones with the best win rate if you want to climb consistently.
- Open the comp page and check the early game tab: a comp is only good if you can reach it. Look at the items it needs and whether your first components fit.
- Stay flexible. The best players don’t force a comp; they start from what the first stages give them and move towards the closest strong comp. The “related comps” on each page are exactly those transitions.
How TFT Coach helps
After a game, My matches compares what you played with the comps in the data, and the in-game coach suggests the closest strong comp for the units you already have. Neither replaces your own judgement, but they save you from guessing.