Last Updated: October 17, 2026
"Change is annoying, but certainty is absurd." — Voltaire
CRISPI, you had a good run
Today, CRISPI officially retires. Its replacement is the DeckCheck Threat Index (DTI), a new framework for evaluating Commander decks and helping you find the right bracket. I covered the philosophy of the new system and the flaws of CRISPI in I Was Wrong About CRISPI. If you want that deeper dive into this change, start there.
This post is reference material on how the finished system works and what your results mean.
Overview
When you scan a deck, you get three connected results: twelve benchmark ratings that translate into a Threat Index score and a recommended bracket. Together, they describe what your deck can do and the Bracket you likely need to play it in.
The twelve benchmarks measure resources, access to needed cards, threats, interaction, and resilience. Each benchmark receives a rating from S down through A, B, C, and F, with an explanation tied to the rating.
Importantly, not all benchmarks are equal. Speed and the ability to execute your plan carry the most weight. Finding cards, answering opponents, and protecting the plan also contribute substantially. Recovery and backup plans contribute less.
DeckCheck assigns your deck a bracket using these three things:
- Construction requirements. The rules of the official Bracket system. A low score cannot erase those requirements.
- Overall capability. Score thresholds help place decks according to the combined strength of their twelve ratings.
- Specific checks, called Gates. These catch decks whose speed or ability to restrict opponents are dialed way up relative to their other stats, often referred to as "min-maxing."
You can inspect and understand the reasoning behind your result. Open any benchmark to see its explanation and Card Ledger, which lists the cards supporting that rating. The same card can appear in several places when it serves several roles.
If a result seems off, the explanations let you identify what needs attention: a missed interaction, an unrealistic setup, or a recovery plan the deck cannot actually afford. You can use those details when submitting feedback—and when explaining your deck to the people you are about to play with.
What a DTI number means
In almost every game you've played, a bigger number wins. One more point of attack, one more level, one more star. So when your deck scans a 38 and your friend's deck scans a 35, it's fair to assume yours is stronger.
It might not be. The Threat Index can't tell you that it is. And if you assumed it could, that's not on you. Almost every other number in games works that way. The DTI doesn't.
Close scores are a tie
The Threat Index is twelve tier ratings added together. On the heavily weighted benchmarks, an S is worth 10 points and an A is worth 5. So if one scan reads one tier differently, the score can move up to 5 points. Nothing about the deck changed. That's one judgment call.
On top of that, AI doesn't read a deck exactly the same way twice (more on that in the Omni-Score section below). Put those together, and a 38 and a 35 are the same deck as far as the DTI can tell.
Big gaps usually mean something real
A 20-point gap is a different story. To get from 35 to 55, a deck needs several benchmarks a full tier higher or more: a faster clock, more interaction, more ways to find its pieces, more ways to recover. A 55 is very likely stronger than a 35.
Very likely. Not always. A 35 packed with graveyard hate can wreck a 55 that wins from the graveyard. Every deck archetype has its respective counter-deck. And even when the 55 really is the stronger deck, that doesn't mean it wins more. That's the part most players miss.
Threat is not the same as winning
The Threat Index measures threat: how fast your deck can kill, how hard it is to stop, and how much it can shut down everyone else. In other words, how capable it is.
In a 1v1 game, more capable usually means more wins. Commander isn't 1v1. It's four players, any three of them can team up against you, and everyone at the table is reading threat in real time.
So the deck that shows its strength early (or obviously), with its commander out on turn 2 and swinging on turn 3, eats the first removal spells and generally becomes the archenemy of the rest of the table. Meanwhile, the quiet deck in the corner with a value engine and a couple of mana rocks keeps accumulating resources. By the time anyone looks over, it's too late.
Seasoned players know this. They don't just watch the scariest board. They look for the deck that's flying under the radar, or slowly building an engine that turns lethal if no one answers it. That skill is threat assessment, and no score can teach it. The DTI reads your 100 cards. It can't read the table or solve for politics.
Threat assessment is only one of the things that decide games. A lower-scoring deck can beat a higher-scoring one because of:
- Seat order. Going fourth against a fast deck means you get the fewest turns before it goes off.
- Bad threat assessment. The table spends its removal on the wrong player.
- Politics and grudges. Someone at the table still hasn't forgiven you for last week's Armageddon.
- The pilot. A player who knows their deck inside out beats a stronger list played on autopilot.
- Luck. Sometimes, the cards are just not in your favor.
A higher score doesn't mean you'll win against decks with lower scores. It means your deck is more of a threat. Sometimes that's exactly why it loses.
A better deck can keep the same score
The DTI rates each benchmark in tiers, and a tier is a wide bucket. A deck can get a lot better inside its tiers without crossing into the next one. Smoother mana, a cleaner curve, better cards doing the same jobs: none of that moves the score unless a benchmark changes tier.
So you can swap 20 cards, watch the score barely move, and still end up with a deck that's more consistent and wins more often. The DTI isn't missing your upgrade. Your deck got better at what it already does. It just didn't move up a tier of threat.
How to use it
- Start with the bracket. That's the main job of the score: finding a table where every deck has a fair shot.
- Treat close scores as equal. A few points apart means the same kind of deck.
- Watch the bracket lines. A 51 and a 53 land on different sides of the Bracket 4 line, but they're about as close as two decks get. The panel tells you when a deck sits in the high end or the low end of its bracket. A high 3 and a low 4 are neighbors.
- Judge upgrades by your games. If a swap makes the deck play better, trust that over a score that didn't move.
- Use a big gap to start the Rule 0 talk. If your deck scores 20 points above everyone else's, say so before the game.
- Rescan if you want a steadier number. Every scan of the same list feeds the Omni-Score, which takes the middle tier of each benchmark. It can't remove every difference between scans, but it narrows them.
The players who've spent the most time with power levels already think this way. A score is where the conversation at the table starts, not where it ends.
The Omni-Score
AI doesn't read a deck exactly the same way twice. Scan the same 100 cards twice and a benchmark can come back an A one time and an S the next. Until now, the only answer was "keep the scan you like," and that's a bad answer. So I built a better one.
When you scan the exact same list two or more times, DeckCheck combines those scans into one result: the Omni-Score.
- Each benchmark takes the middle tier of your scans. Three scans that say S, A, S give an S.
- Ties go up. With an even number of scans, a tie between two tiers goes to the higher tier, and a tie between two turn estimates goes to the faster turn.
- Then the normal rules run. The Omni-Score is scored exactly like one scan: same points, same thresholds, same Gates.
- More scans, steadier score. From three scans on, the middle tier outvotes one unusual scan.
Each benchmark is decided on its own, so the Omni-Score can land where no single scan did, even above all of them. A Gate can fire on the combined tiers when it fired on no single scan. That's on purpose: the Omni-Score is the best read of each benchmark, and the bracket follows from that.
Construction requirements don't get a vote. Game Changers, mass land denial, extra turns, and combos are facts about your cards, so the highest card requirement always applies.
What you'll see:
- Once a list has two scans, the Threat Index pill glows and the panel says Omni-Score. That score is your deck's Threat Index and bracket everywhere on DeckCheck: the builder, your deck page, My Decks, and Search.
- The switch at the top of the panel opens any single scan, so you can compare.
- Open a benchmark to see the tier each scan gave. The reason and the Card Ledger come from a scan that gave the Omni-Score tier, so the explanation always matches the tier you see.
- The written synopsis comes from your newest scan.
- Edit the list and the count starts over for the new version. The deck timeline keeps every version and every scan.
You never have to rescan, and each scan still costs 5 credits. But if a tier looks off on a deck you haven't changed, or you just want to be absolutely sure about a deck, now your rescans mean something. It adds to what you already have, and nothing gets thrown away. Click Rescan at the bottom of the Threat Index panel.
Current DeckCheck thresholds and gates
Overall score thresholds
| Threat Index | Starting bracket | Additional requirement |
|---|---|---|
| Below 16 | 1 — Exhibition, otherwise 2 — Core / Precon | Bracket 1 requires a creative theme or restriction that deliberately sacrifices effectiveness. |
| 16–31 | 2 — Core / Precon | Gates or construction requirements can raise placement. |
| 32–51 | 3 — Upgraded Casual | Gates or construction requirements can raise placement. |
| 52–67 | 4 — Optimized | The cEDH gate can raise placement to Bracket 5. |
| 68–96 | 5 — cEDH, otherwise 4 — Optimized | Bracket 5 requires supported competitive construction. |
A deck scoring 29 is 3 points below the Bracket 3 score threshold of 32. It may already qualify for Bracket 3 or higher through another rule. Likewise, 51 is 1 point below the Bracket 4 score threshold of 52.
Gates that raise placement
DeckCheck checks the Gates in this order, and the first one that matches sets the bracket. A Gate can only raise a deck above its score threshold, never lower it.
| Gate | What it catches | Current requirement | Minimum bracket |
|---|---|---|---|
| cEDH | A deck built for the competitive table that also plays like one. It's fast or locks hard, and once it moves, nobody gets a turn to stop it. | Velocity or Suppression is 40+, P1 is S or A, and P2 is S. | 5 |
| Early finish (aka Glass Cannon Gate) | A deck that ends the game early. It might be weak at everything else and fold to interaction, but it can kill the whole table or lock everyone out by turn 4. | Reliably eliminates every opponent, or locks the game completely, by turn 4. | 4 |
| Turbo onset | A deck that forces answers before the table is ready. It doesn't have to win early, but on turn 3 it puts out something that must be dealt with immediately. | P1 is S, and R1 or A2 is also S. | 4 |
| Early oppression | A deck that doesn't win fast at all, but stops everyone else from playing early in the game. Heavy taxes, locks, or relentless attrition that start early leave opponents without a real game. | I2 is A, or I2 is S with it by turn 6. | 4 |
- Velocity = mana (R1) + assembly (A2) + threat onset (P1) + completion (P2) + protection (S1).
- Suppression = mana (R1) + restriction (I2) + reactive answers (I1) + threat onset (P1) + backup plans (S3).
Points per benchmark
| Benchmark group | S | A | B | C | F |
|---|---|---|---|---|---|
| R1, A2, P1, P2 | 10 | 5 | 2 | 1 | 0 |
| R2, A1, P3, I1, I2, S1 | 8 | 4 | 2 | 1 | 0 |
| S2, S3 | 4 | 2 | 1 | 0 | 0 |
All twelve weighted values add up to the Threat Index, out of 96. 96 is S-tier on every benchmark.
Important Update Note: September 20, 2026
It's clear to me now from the feedback I received from multiple sources that Eco mode for the DTI scan itself just can't exist. It's not smart enough and it's confusing and problematic and so I've disabled Eco mode for the Threat Index scans.
Comments
…