Raising the favourite to −120 pushes the dog to +100. A dog at +100 carries zero margin — we keep nothing on it, win or lose. And the dog is the side we are trying to attract: the recreational money that arrives after the early money is already down, typically 80 to 95 percent of everything we write on the game.

So the trade is this. We collect extra juice on the small slice that is already in, most of which was written before we juiced, and in exchange we write the overwhelming majority of our handle at no margin at all. Move the number instead and both sides sit at −110, and we keep 4.55% of every dollar that comes in.

Even the light version leaks. Going to −115 puts the dog at −105, where we keep 2.38% instead of 4.55% — we have halved the margin on the majority of the handle to add a point and a bit to the minority.

At a 15/85 split, moving holds 4.55 per hundred. Juicing to −115 holds 3.00. Juicing to −120 holds 1.25.

This is the part that gets missed. Juice is a charge on the next customer. It does not touch the position we already hold. If we are long the favourite, we are long the favourite at −110 and we are still long the favourite at −120 — identical exposure, we have simply made the bet more expensive for someone who was never our problem.

Moving the number changes what we need to happen for us to win. That is the job. Juice is a pricing tool; the number is the position.

Say the early money took the favourite at −5.5 and we now need dog money. If we write it at +6 and the game lands on exactly 6, the favourite ticket wins and the dog ticket pushes — we pay one side. If we write it at +6.5 instead, both tickets win and we pay both. Same game, same number, twice the damage.

That is a middle, and a middle costs roughly 2.1× an ordinary losing side because we pay two winners and keep nothing. Games land exactly on a whole number 3.83% of the time. Stopping on the number takes that risk to zero; stepping over it accepts it in full, for nothing in return.

Whole numbers keep working as you go. +7 is no worse than +6.5, because the 7 pushes and stops the window again.

Which is why the rule is to walk rather than jump. Writing at +6, then +6.5, then +7 spreads the exposure across three prices, and only the tickets at +6.5 and beyond can be middled at all. Jumping straight from −5.5 to −7 puts every dog ticket inside the same window.

The instinct against moving is fear of getting caught on the wrong number. The data does not support the fear.

Three games in four are settled nowhere near our number. Refusing to move protects us against a three percent event while leaving us carrying a one-sided position that is certain. The exposure we are avoiding is small and occasional; the exposure we are accepting is large and guaranteed.