How Bold Picks Work - Dirty-Dawg.com

The short version

Every pick on every game is compared with everyone else's pick on that same game.
If yours is far enough from the pack (a "z-score" of 3.5 or more), it's a Bold Pick. Only a few percent of picks make it.
Once the game is scored, a Bold Pick becomes a Top Dawg if you were right and backed it, or a Dawg Pile if you were wrong, backed it, and put real confidence on it.
Calling a rare upset correctly can earn a Top Dawg even if the pick wasn't bold.
Nothing is shown until the week's picks have locked.

1

Turn each pick into one number

A pick has two parts: a team and a confidence. To compare picks we fold them into a single number. One team in the game counts as minus and the other as plus, and the size is your confidence.

Notre Dame at BYU, 12 on Notre Dame

−12

Notre Dame at BYU, 3 on BYU

+3

That puts the whole field on one line, from "all in on one team" to "all in on the other." A 1 on either side lands near the middle, because a 1 says "I'm not sure."

2

Find the field's center and spread

For each game we take every player's number and find two things:

The center: the median. Line everyone up and take the middle pick.
The spread: the MAD (median absolute deviation). Measure how far each pick sits from the center, then take the middle of those distances. It's a typical "how far from the pack" distance.

Why the median and not the average? An average gets dragged by the very picks we're trying to spot. One wild pick pulls the average and inflates the spread, which makes that same pick look less unusual. The median and MAD barely move when one extreme pick is added, so a bold pick can't hide behind its own effect on the numbers.

When the MAD is zero. On a lopsided game, more than half the field can land on the exact same number, and then the MAD is 0. We can't divide by zero, so we use the average distance from the median instead, times 1.2533. That standard substitute keeps the scale the same. A game where literally everyone made the identical pick can't produce a Bold Pick.

3

Score how far out your pick is

Your distance from the center, measured in units of the spread, is the modified z-score:

z = 0.6745 × (your number − median) ÷ MAD

The 0.6745 puts it on the same scale as an ordinary standard deviation. A z of 1 is a normal pick, a z of 2 is unusual, and a z of 3.5 is rare.

|z| ≥ 3.5

Bold Pick (regular season)

|z| ≥ 2.0

Bold Pick (playoff / bracket weeks, see below)

8+ picks

Needed on a game before anything can be flagged

The |z| means the sign doesn't matter: way out on either side counts. That leads to something people don't expect: a Bold Pick can be low confidence. If nearly everyone puts 12 or 13 on a big favorite and you put 4 on it, you've told the field "I don't trust this one," and that's an outlier too.

A real one: Notre Dame at BYU

CFCC Week 7, 2026 · 368 picks · BYU won

Players on Notre Dame	341 (93%), averaging 11.3 confidence
Players on BYU	27 (7%), averaging 3.9 confidence
Median (the center)	−12 (12 on Notre Dame)
MAD (the spread)	1

One player put 14 on BYU, which is +14:

0.6745 × (14 − (−12)) ÷ 1 = 17.54

That's five times the 3.5 line, so it's a Bold Pick. BYU won, so it became a Top Dawg. A MAD of 1 means the field was packed tight around "12 on Notre Dame," so stepping away from it scores big fast.

Try it yourself

Same game, same field. Pick a side and a confidence (1–15).

9

4

The game ends: Top Dawg or Dawg Pile?

Being bold isn't enough. The badge has to match what you meant. Three terms:

The field's favorite is the side the whole field leaned toward, weighted by confidence. It's the side the average pick lands on.
Backed it means your confidence was at or above the average confidence of the other players who took the same team you did.
Real confidence means at least a third of the way up the week's confidence range: 5 or more in a 15-game week (the top confidence divided by 3).

Your Bold Pick…	Badge
Picked the winner, and it was not the field's favorite (you called the upset)	🏆 Top Dawg
Picked the winner, it was the field's favorite, and you backed it	🏆 Top Dawg
Picked the winner, it was the field's favorite, but you were light on it	no badge
Picked the loser, backed it, and put real confidence on it (5+ in a 15-game week)	💩 Dawg Pile
Picked the loser, but you were light on it, or your confidence was 4 or lower	no badge

A tied game counts as a win for everyone, same as the scoring.

Why the real-confidence rule for a Dawg Pile? Without it, a 1 could land in the pile just because everyone else on that side also used a 1. A Dawg Pile is for going hard on the loser, and a 1 through 4 says "I'm not sure."

Why "no badge" for a timid pick on the favorite? Picking the winner should never be a Dawg Pile. But quietly hedging on the favorite isn't a Top Dawg either. You were bold in the "I don't trust this" direction, and the game proved you wrong.

The upset path. Some correct upset calls don't clear the 3.5 bar because they carried modest confidence. So there's a second way in: if fewer than 20% of the field took your side, it wasn't the field's favorite, and it won, you get a Top Dawg at any confidence. There is deliberately no matching "rare side that lost" path. A small, hopeful underdog flyer that doesn't hit isn't a Dawg Pile. Only a truly bold pick can land in the pile.

Two more from Week 7

CFCC Week 7, 2026 · 368 picks on each game

🏆 12 on Florida, at Texas

Only 9 of 368 (2%) took Florida. Median: 13 on Texas. MAD: 1.

0.6745 × (−12 − 13) ÷ 1 = −16.86

Bold. Florida won, and Florida was not the field's favorite, so it's a Top Dawg: the upset call.

💩 15 on Mississippi State, at LSU

11 of 368 took Mississippi State, averaging just 4.1. Median: 13 on LSU. MAD: 1.

0.6745 × (−15 − 13) ÷ 1 = −18.89

Bold. LSU won, 15 is well above the 4.1 that Mississippi State's other backers averaged, and well past the 5 needed for real confidence. That's backing the loser: Dawg Pile.

5,520

picks in Week 7

361

flagged Bold (6.5%)

103

became Top Dawg

81

became Dawg Pile

The other 177 Bold Picks ended with no badge: right but light on the favorite, or wrong without real confidence behind it.

Playoff and bracket weeks

In the playoff format there's no "other side." You put confidence on every team, and each team earns you points round by round. So:

Your number is just your confidence on that team, with no plus or minus.
The bar for Bold is |z| ≥ 2.0. This format doesn't have the lopsided games that made the regular season need a higher bar.
Top Dawg: your conviction matched the outcome. You went bigger than the field and the team scored points for you, or you went smaller than the field and the team went out without scoring.
Dawg Pile: the opposite. You loaded up and the team flopped, or you faded them and they delivered.

Where you'll see them

Homepage, after picks lock: the 5 boldest picks of the week (largest |z|) in each contest, as 🔥 Bold Picks.
Homepage, after the week is scored: for 48 hours, a recap with up to 5 Top Dawgs and 3 Dawg Piles per contest. It's a highlight reel, not a shame list.
The Bold Picks Board: every Bold Pick of the week once picks lock, then every Top Dawg and Dawg Pile once it's scored, with search and filters.
Each player's picks page: every badge they earned that week, next to the pick.
The weekly recap email: the week's best Top Dawg and worst Dawg Pile.

Tap any badge for the numbers. It shows the pick's confidence, its vs. field average (how far your confidence sat from the field's average on this game, from your side's point of view), the field size, and the z-score that flagged it.

FAQ

Can anyone see Bold Picks before the deadline?+

No. Bold Picks are built from the whole field's picks, so nothing appears until the week's picks have locked.

My pick was low confidence. Why was it flagged?+

Bold works in both directions. If nearly everyone loaded up on a team and you put a small number on it, you were far from the pack, the same as someone who loaded up on the other side.

I picked the winner. Why no Top Dawg?+

If the winner was the field's favorite, you also had to back it, at or above the average confidence of the others who took that team. A light pick on a favorite that won gets no badge either way.

Why 3.5?+

Lower cut-offs flagged far too much. At 2.0, a lopsided game made almost any minority pick look bold, about 1 in 7 picks. At 3.5 it lands on the genuinely unusual ones, a few percent of picks in a typical week.

Why do some games produce lots of Bold Picks and others none?+

It depends on how tightly the field agrees. When almost everyone takes the same team at similar confidence, the spread is tiny and stepping away from the pack scores big. On a true toss-up the field is spread out, and it takes a lot more to stand out.

Do the house accounts count?+

Johnny Vegas and the other house and test accounts are part of the field (their picks count toward the median), but they're never featured with a badge on the homepage or in the recaps.

I lost a Bold Pick at low confidence. Why no Dawg Pile?+

A Dawg Pile needs real confidence behind it: at or above your side's average AND at least 5 in a 15-game week (a third of the confidence range). A 1 through 4 on the loser gets no badge.

Why not just use the z-score for Dawg Piles too?+

Every Dawg Pile already IS past the 3.5 z-score; that's what makes it Bold. But on a lopsided game, taking the other side at any confidence is far from the pack. When 364 of 368 players put 14 or 15 on Miami, a 1 on Florida State scored z = -17.7. The z-score tells us you went against the field. The confidence check tells us you actually went hard on it.

Does this affect my score?+

No. Badges are just for bragging rights (or the opposite). Your points come from the regular scoring.

Something look off? Let us know.