Teamfight Manager 2 BP strategy: a process that survives the Auto Patch System

1. First-bans: target patterns, not stars

The first-ban slot is the one most new managers waste. The instinct is “ban the best champion.” That's almost always wrong in TFM2 because the meta is broad — the AI ban algorithm and human opponents both leave many tier-1 picks open. Better mental model: ban what the enemy team is best at, not what is best overall.

The Personal Tier List overlay during draft and the Analyst's Pre-Match Report tell you which champions the opposing players have the highest individual scores on. That is the cut, not the top of your own tier list. A C-tier champion that the enemy mid has 95 mastery on is more dangerous than an S-tier they've never picked.

Second pattern: ban flex enablers, not flex picks themselves. Champions deliberately designed to be multi-role draw their value from the combination with a specific partner. If you ban the partner, the flex collapses to whichever role is now worse for them. The partner is usually a single-role specialist with no clean replacement, which makes that ban structurally cheaper than banning the flex pick directly.

Third: don't first-ban your own counter. Tempting trap — “I want to play X, X loses to Y, ban Y.” But this telegraphs your draft. Save the Y ban for last-ban after you've actually locked X.

Fearless Ban/Pick inverts everything above. In Fearless mode, a champion banned in game 1 stays available; a champion picked in game 1 cannot be re-used. The enemy's pocket picks — deep on 1-2 players, shallow on the bench — become the priority target, because those are the ones they'll struggle to replace across the series.

Treat first-ban as cutting the enemy's strongest tool, not as removing the game's strongest tool.

2. Last-bans: the lock-in stage

By the time you're at last-ban, the comp shape is mostly visible. Both teams have made 2 picks each, and the lanes that are still open tell you what each side is trying to build. The last ban's job is to close the door on the enemy's planned final pick, not to grab a small statistical edge.

Look at the unfilled lanes on both sides. If the enemy is missing a clear engage tool, their last pick almost certainly is one — ban the best engage tool left in that role's pool. The Analyst's role tags help here: champions are tagged for the role they realistically play in the current patch, not their nominal role.

Watch their staff archetype too. A team with a strong Analyst drafts more reactively; a team relying on Head-Coach defaults tends to draft to script. The latter is more predictable, the former needs you to ban broader.

Counter the engage timing, not the engage champion. Ultimates unlock at level 5, so if the enemy is shaping a level-5-spike engage comp, denying their ult-spike enabler — a hard-CC second pick — is more disruptive than banning the engage core. The core can still pick into your bans; the enabler often can't.

One mode-specific note: side selection matters. The losing team picks side for the next game in a series. If you're going into game 2 from a loss, your last-ban in game 1 should already be modelled around the side you'll force next, not the side you currently hold.

3. Flex picks: the value of a question mark

A flex pick is a champion that can credibly play two or more roles. Their value isn't the champion — it's the information asymmetry they create. Until you actually lock the role with a subsequent pick (or by the time the loading screen resolves), the enemy can't fully counter-draft.

Distinguish designed flex from accidental flex. The dev team commits to deliberately keeping some champions multi-role and to nerfing role identity if too many become universally viable. Designed flex is stable across patches. Accidental flex — a top laner the Auto Patch System briefly buffed into the mid range — appears and disappears patch to patch. Don't build a series strategy around it.

Flex needs deep player pools, not just champion pools. A jungler with one champion in their pool can't actually be flexed into top, even if the champion is technically flexible. Check the per-player tier-list overlay before announcing a flex to yourself.

If you're playing with Hide Attributes mode on, flex picks gain even more — the enemy can't immediately see your players' alternative pools, so a flex pick stays ambiguous longer.

Final rule: flex is a tempo tool, not a power tool. Most flex picks are slightly weaker than role specialists. You pick them because they let you delay committing — pick last in a phase, then drop the specialist into the lane the enemy didn't counter. If you're picking flex and then revealing the role immediately with the next pick, you got zero benefit.

4. Counter-picks: the last-pick advantage

Last pick (P5 in standard ban/pick) is the most powerful slot in the draft because it's the only one with full information — every enemy pick is locked. The mistake new managers make is wasting it on “the best champion left” instead of the champion that breaks the enemy's win condition.

There are three useful counter-pick types, and they don't go in the same slot:

The Pre-Match Analysis Report classifies enemy players by tier before draft — use it to spot which enemy player is the comp's load-bearing column. The win-condition counter goes at that player, not at their lane opposite. A comp that lives off one carry dies when that carry can't farm; you don't need to beat them in lane to do that.

Specifically for TFM2: the jungle smite-style execute (epic monsters at ≤700 HP) and the Serpen objective are pivotal. A jungler who can win the smite contest at Serpen is a comp-level swing, not just a lane matchup — so jungle counter-picks usually pay back more than lane counter-picks in the same slot.

5. Priority by lane

Not all lanes are equal in BP priority. The order in which you should pick a role (or force the enemy to pick it) varies by what your comp wants. The default frame:

• Jungle first when the comp has a smite-dependent win condition. Serpen-first comps need a jungler who can win the smite. Lock jungle before mid, or you'll get squeezed out of the smite-winner pool.
• Support second when running engage. Engage comps live or die on the support's first-pick CC. If your engage win-con sits on the support, lock them before the enemy bans the top engage supports.
• Mid third by default. Mid is the lane with the deepest pool in most patches, so it has the most replacements — picking it earlier than necessary is wasted information.
• Top fourth or last for comps that need a counter. Top lane is the most isolated, which makes it the best counter-pick slot — saving it lets you read the enemy comp first.
• ADC last when scaling, second when bullying. A scaling carry doesn't need to be picked early. A lane-bullying ADC + roam support combo needs to lock the support-ADC duo together early.

A side note on AI difficulty: in CPU-vs-CPU testing, an Easy-vs-Hard mismatch produced roughly a 2:8 win-rate split from draft skill alone, before any in-match decisions. Translation: draft is genuinely worth this much process. If you're losing matches to a tier-equivalent opponent, the first thing to audit is the draft sequence above, not the tactics screen.

6. Comp synergy: the final check

Before you confirm the last pick, run a 30-second synergy check. The five picks need to add up to a comp that can answer one question: how do we actually win this game? TFM2 comps generally land in one of the archetypes the comps page catalogues — Engage, Poke, Split-push, Scaling, Counter, or deliberate Flex. If you can't put your draft into one of those (or a deliberate hybrid), you've drafted five strong champions, not a team.

The three-question check:

1. Who initiates? Name the champion. If no one on the team has a hard engage tool and your comp isn't deliberately poke or scaling, you'll get permanently kited.
2. Who finishes? Name the carry. If your comp has two designated carries that both need farm, one will be starved.
3. Who survives the enemy win-con? Re-read their comp and name your answer. If you don't have one, the last-pick slot was wasted on tier-list value instead of role function.

That third question is the one the Auto Patch System keeps alive across seasons. Champions, items, and tier rankings shift; the “do we have an answer to their win-con” test does not.

FAQ