Why your unit size is too large.

Pick a number. The amount of money you would bet on a single signal if the model fired it tonight. Hold that number in your head.

For most people, that number is too large. Not by a little. By a meaningful multiple. The unit size you have set is almost certainly the unit size of a bankroll that is two or three times what you can actually afford to lose without changing your behavior.

This article exists because the most common reason disciplined bettors blow up is not bad picks. It is correct picks with stakes that were too large to survive normal variance.

What a unit is actually supposed to be.

A unit is a fixed fraction of your bankroll. The standard recommendation is 1 percent to 2 percent. If your bankroll is 5,000 dollars, a unit is 50 to 100 dollars. If your bankroll is 1,000 dollars, a unit is 10 to 20 dollars.

That last number is where the gap between theory and practice opens. A bettor with a 1,000-dollar bankroll betting 10 dollars per signal sees a year of work produce maybe 400 dollars of expected profit. Compared to the time spent watching games, reading models, and tracking results, the return is small. The temptation is to size up to make the work feel worth the return.

Sizing up does not make the return feel better. It makes the variance feel worse. A 10-loss losing streak at 10 dollars per bet costs 100 dollars. The same streak at 50 dollars per bet costs 500 dollars. The first one is a bad month. The second one is half the bankroll.

The bankroll test.

There is a simple test for whether your unit size is correct. Imagine the model goes <span class="stat negative">3-10</span> over the next two weeks. Not a catastrophic stretch. A normal bad cycle within the variance of a real edge.

If that stretch would change how you sleep, your unit size is too large. If it would change which bets you take next, your unit size is too large. If it would tempt you to size up to recover, your unit size is too large.

A correct unit size lets you watch a 3-10 stretch with mild irritation and complete confidence that the next 20 bets will probably be fine. If you cannot picture watching that stretch calmly, the size is wrong.

Why people get this wrong.

Three reasons recur.

The bankroll is a fantasy number.

The bettor mentally counts their bankroll as the money they could theoretically deposit, not the money they have already separated from rent, food, and savings. Bankroll has to be money you have actually parked for this specific purpose. Anything else is a number on paper.

The stake feels too small to matter.

A 10-dollar unit feels trivial. The bettor sizes up so the wins feel meaningful. This trades a small psychological gain for a large variance increase. The wins do feel better at 50 dollars per bet. The losses also feel much worse.

The wins came first.

Early wins inflate confidence. The bettor sizes up after a hot stretch because the math feels conquered. The hot stretch ends. The new larger stake meets a normal variance cycle. The bankroll takes damage that the original unit size would have absorbed.

How to fix it without changing your routine.

Cut your unit in half. Not as a temporary precaution. As your new permanent size. Run it for 60 days. If your bankroll grows during those 60 days, raise your unit proportional to the growth, not back to the old size.

The boring version of bet sizing is the only one that compounds. The exciting version compounds until it does not, and then it ends the entire effort.