Coming back to TFT after a year away

You came back at an awkward moment. Set 17 has about three weeks left.

That is not a reason to put the game down, but it does change what is worth learning right now, and it makes one thing you have probably already noticed temporary in a way that is easy to miss.

What you missed, and why almost none of it matters

TFT is not a game you fall behind in. It is rebuilt on a schedule.

Riot ships a new set roughly every four months, and a set is a full replacement: new champions, new traits, new items, new core mechanic. In 2023 Riot ended the old mid-set update model, writing that the final mid-set would make way for “our new three-sets-a-year model.” That has held since. There is no half-refresh to study up on and no partial carryover.

So if your last games were before 30 July 2025, here is the whole of what you missed, and the point is how little you need to do about it:

Set	Name	Went live
14	Cyber City	2 April 2025
15	K.O. Coliseum	30 July 2025
16	Lore & Legends	3 December 2025
17	Space Gods	15 April 2026

Two complete sets came and went. Neither is in the game any more. Everything in Set 15 and Set 16 was deleted the day the next set shipped, which means the correct amount of time to spend catching up on them is zero.

What did carry over is everything that is not content: managing gold, knowing when to take damage to save economy, positioning, itemising a carry, reading what the shop is offering you. Those skills are the game. If they felt rusty coming back, that is the actual gap, and it is the one worth closing.

One piece of housekeeping that will save you confusion when you read patch notes. Riot’s patch numbers are now set-aligned, so the first number is the set: “16.3 is the third patch of Set 16,” and there are “generally eight patches per set.” The current patch is 17.7. When you see 18.1, that is not a small update. That is a new set.

The carousel is gone, and it is coming back

This is the change you spotted, and your read of it was right in substance.

Set 17 removed the carousel, the round where every player’s avatar was dropped onto a rotating ring of champions holding items and you ran your little guy at whichever one you wanted. Riot’s own framing in the patch 17.1 notes: “Replacing the Carousel, The Realm of the Gods offers you two gods to choose an offering from, then a less powerful, generic one from Pengu.”

One correction to the mental model, because it changes how you should value the picks. It is not a menu of champions holding items. It is a menu of offerings from gods, which is closer to an augment choice than to a draft. Every god offering does come with an item component attached, so items are still flowing, but you are choosing a reward rather than choosing a unit.

What actually changed, in the order it will affect you:

Nobody can take your pick any more. The old carousel was contested. Eight players raced at nine champions and you could get blocked, sniped, or forced onto your third choice by someone faster. That interaction is gone entirely. Your three options are yours.

Health changed jobs. It used to decide when you picked, releasing the lowest-health players first so a losing player got first grab. Now it decides what Pengu offers you. Riot’s wording: “If you’re lower in the standing, HP wise, Pengu will offer you higher cost units and you will receive component anvils instead of random components during PvE.” The comeback mechanic survived. It just stopped being about turn order.

Your picks compound. The god you choose most becomes your aligned god, and at stage 4-7 that god hands you a Boon in the form of an armory, then keeps dropping loot for the rest of the game. The old carousel had no memory. This one does, so a pick is partly a vote for what your late game looks like.

Now the part that matters for how much you invest in liking it. Riot flagged the removal as temporary from the start, writing in the Set 17 overview that the “Carousel might be on vacation this set, but that doesn’t mean it’s gone forever.” In the Set 18 dev drop published on 11 July 2026, the narrator puts it in two words: “Carousel’s back.”

Set 18: Enchanted Wilds goes to PBE on 28 July, and Riot’s patch schedule has 18.1 landing on 12 August 2026. So the thing you came back and liked is a one-set experiment that ends in about three weeks.

Worth being straight about the limits here: Riot has said three words about the returning carousel and nothing else. Whether it comes back unchanged, whether the health-based comeback tuning from the Realm of the Gods carries into it, whether the running is still there at all, none of that has been described. Anyone telling you what Set 18’s carousel does mechanically is guessing. That gets answered on PBE on 28 July.

Reading a comp page without getting burned

This is where returning players lose the most games, and it has nothing to do with mechanics. You open a tier list, read the letter and the name, and both of them mislead you.

The two-character label is the whole strategy. A card reading “lvl 6” is not telling you the comp is easy or that you should be level 6. It is telling you this is a reroll comp: stop levelling at 6, then spend gold rolling the shop over and over to three-star a cheap unit, because low-cost units appear most often in the level 6 shop. “Fast 8” is the exact opposite instruction: do not reroll at all, save aggressively, buy experience, and get to level 8 fast so the higher-cost units start showing up. Two comps, two incompatible relationships with your gold. Play a reroll comp with a Fast 8 economy and you will lose with a board that looks correct.

The name is marketing, not a description. More on this below, because the clearest example is one of the comps you flagged.

Pick rate deserves as much attention as average placement. A comp showing a strong average on a very low pick rate is often noise rather than a signal. If only a sliver of players are running something, the ones who are tend to be specialists forcing it when the game hands it to them, and the numbers flatter it accordingly.

Different sites will grade the same comp differently, and none of them is lying. They filter by rank and region, and the filter is usually not on the card. That difference is worth more to you than the letter itself.

Last thing, and it applies well beyond TFT: the AI summaries and SEO pages that sit above the real sources are frequently wrong about patch specifics. While writing this I hit a confident claim that patch 17.7 buffed N.O.V.A. Strike damage with exact before-and-after numbers. Riot’s actual 17.7 notes contain no N.O.V.A. change at all. Read the patch notes.

The two comps you were looking at

Taking these as worked examples, since they show both failure modes.

N.O.V.A. Akali

The plan, per TFT Academy: “Level to 6 and roll for Akali 3 and Jax 3. Always play 5 NOVA if you highroll early Kindred.” So it is a reroll comp built on three-starring Akali, and the level 6 label is the instruction, not a difficulty note.

Five N.O.V.A. is the payoff and it is worth understanding why. Riot describes the trait as giving each N.O.V.A. unit a power surge a few seconds into combat, and says that investing up to five “will reward you with a Striker selector at (5), with the chosen N.O.V.A activating their Strike during the power surge.” You are not just stacking a bonus. You are picking which unit gets to fire the big mid-fight effect.

On grading: the card on MetaTFT showed S tier, a 4.00 average placement and a 60.3% top-four rate. Both TFT Academy and BunnyMuffins grade the same comp A on the same patch. That is one notch down from two independent sources, and it is a good illustration of why the letter is the least portable part of a comp page. Treat those specific numbers as one site’s snapshot on one day, at whatever rank filter it defaults to.

Worth noting that patch 17.7 barely touched this board. Of its eight units, the only one that appears in the notes at all is Urgot, whose transformed passive damage went from 225% to 250% AD. The trait itself was untouched.

Meeple Corki, which is not really a Meeple comp

Here is the name problem in its clearest form.

Riot’s Meeple trait does something distinctive at scale: Meeps empower Meeple abilities, at (7) they start cloning champions on your bench, and at the prismatic tier you summon four MeepLords. That is the thing worth reading about, and it is presumably what drew your eye.

That board does not do any of it. Count the units against the trait: of the eight on the card, only five are Meeple, which lands on the Silver breakpoint. It never reaches the (7) cloning slot and gets nowhere near prismatic. It is a Corki carry board taking a health bonus in passing, running a Fast 8 economy. Calling it “Meeple Corki” is accurate in the way a street name is accurate.

The other reason to be careful: it showed a 0.18% pick rate. That is nearly nobody. A second stats site my research could actually read has it at a 0.06% pick rate with an 88% top-four rate, which is not evidence that the comp is excellent, it is evidence that the sample is too small to mean anything. There is no cross-source consensus that this comp is strong. One site did not rank it at all, another put a related Corki board in B.

So: your read on N.O.V.A. is well supported and matches your own results. Your read on Meeple is a stats-page artifact, and worth dropping.

What to do with three weeks

If you are playing casually with friends, none of this needs to change. Keep forcing N.O.V.A. if it is working and you enjoy it. The comps expire on 12 August either way, and there is no ladder position to protect.

If you want to come out of this actually better at the game, spend the time on the parts that survive the set rollover: gold thresholds, when to break economy, positioning against the specific boards beating you. Then arrive at Set 18 on 12 August level with everyone else, which is the best moment a returning player ever gets. Everyone is new that day.

One page-level caveat, in keeping with how this site works. Nothing here was verified in game by me. The mechanics and dates come from Riot’s own patch notes, set overview, dev drop and patch schedule; the comp grades come from community stats sites and are labelled as such. Riot’s schedule page carries its own hedge that “patch dates can change at any time for any number of reasons,” so treat 12 August as scheduled rather than certain.