Skip to content

Push-Adjusted EV

Some bets have three outcomes, not two. This prices the one where your stake simply comes back.

%
%

On whole-number NFL spreads, key numbers like 3 and 7 carry real push mass.

$

Expected value

+$2.45

2.45% of stake. Ignoring the push and treating this as a two-way bet would give −$4.55 — so the 7.0% push chance is worth $7.00 here. Pushes pay nothing, but they take probability away from the loss branch, and that is worth real money.

Chance it wins
50.00%
Chance it pushes
7.00%

Stake returned, no profit either way.

Chance it loses
43.00%

Derived — whatever is left over.

EV as ROI
+2.45%
Win rate among decided bets
53.76%

Pushes excluded. Compare this with the 52.38% break-even rate — it is the like-for-like figure.

What the push is worth
$7.00

EV here minus EV if the same win chance had no push mass behind it.

Where the expected value comes from
OutcomeChanceP/LContribution
Wins50.0%+$90.91+$45.45
Pushes7.0%$0.00$0.00
Loses43.0%−$100.00−$43.00

With the push chance set to zero this gives exactly the same answer as the ordinary expected value calculator — the two are the same formula, and the third outcome simply has no mass. Push probability is yours to estimate: on NFL spreads it comes from the scoring distribution around key numbers, and 3 and 7 carry far more of it than the numbers beside them.

Runs entirely in your browser · nothing is sent or saved

What is a push in betting?

A push is a bet that settles level — the result lands exactly on the number — so your stake comes back and nobody wins.

It happens on whole-number spreads and totals: back −3 and win by exactly three, and the bet pushes. Half-point lines cannot push, which is precisely what the half point is for.

A push pays nothing, so it is easy to treat as irrelevant. It is not. Every point of push chance is a point taken away from losing, which is worth the full stake — and on NFL key numbers like 3 and 7 that mass is large enough to change whether a bet is worth taking.

How it works

A push returns your stake. You do not win and you do not lose, so its direct contribution to expected value is exactly zero. That makes it sound irrelevant, and it is not — because probability mass has to come from somewhere, and every point of push chance is a point taken away from the loss branch.

That is the whole insight. A bet with a 50% win chance and a 7% push chance loses only 43% of the time. It is worth meaningfully more than an identically priced bet that wins 50% and loses 50%, even though the push itself pays nothing.

This matters most on whole-number NFL spreads and totals, where key numbers carry serious push mass. A game landing exactly on 3 or 7 is not a rare event — those two margins account for a large share of NFL results — so a −3 with a real push chance behind it is a fundamentally different bet from a −3.5, and the half point costs what it costs for exactly this reason.

The other useful figure is the win rate among decided bets, because that is what compares like with like. A price of −110 needs 52.38% to break even, and that figure refers to bets that actually get decided. Comparing it against a raw win rate that includes push mass in the denominator is comparing two different quantities.

Set the push chance to zero and this returns exactly what the ordinary expected value calculator returns. That is not a coincidence — it is the same formula with an empty third branch, and the two are tested against each other to make sure they never drift apart.

The formula

loss chance = 1 − win chance − push chance

EV = win  × stake × (D − 1)
   + push × 0
   − loss × stake

EV ROI = EV / stake

win rate among decided bets
  = win / (win + loss)

with push = 0 this reduces exactly to
the standard two-outcome EV formula.

The push term is written out explicitly even though it contributes zero — it is the branch that takes mass away from the loss, which is where its value comes from.

A worked example

An NFL spread of −3 at −110, $100 stake. You make it 50% to cover, with a 7% chance the game lands exactly on 3 — which on a key number is realistic.

The bet loses 43% of the time. EV is 0.50 × $90.91 − 0.43 × $100 = +$2.45, or +2.45% ROI.

Ignore the push and treat it as 50/50 and you get −$4.55 — a losing bet. Same price, same win chance, opposite conclusion. The 7% push chance is worth $7 on a $100 stake, and it is the entire difference.

Among decided bets the win rate is 50 / 93 = 53.76%, which clears the 52.38% that −110 requires. That is the comparison that makes the bet make sense.

Common questions

Does a push help or hurt me?
It helps, relative to the loss it displaced. A push itself is worth zero, but every point of push chance is a point that is no longer a loss — which is worth the full stake in expectation.
Where do I get a push probability?
From the scoring distribution of the sport. In the NFL, margins of 3 and 7 are far more common than the numbers around them, and historical margin frequencies are the usual source. It is an estimate, and the calculator treats it as yours to make.
Why is my win rate different from the raw percentage?
Because break-even rates refer to decided bets. Excluding pushes from the denominator gives the figure that compares like with like against the 52.38% a −110 price requires.
Does this match the standard EV calculator?
Exactly, when the push chance is zero. They are the same formula with an empty third branch, and the two are tested against each other so they cannot silently drift apart.
Which markets can push?
Whole-number spreads and totals, moneylines in sports that can draw where the book voids rather than settles, and some props. Half-point lines cannot push under integer scoring, which is precisely what the half point buys.

The guide behind this calculator

Related calculators

For informational and analytical purposes only. These tools do not predict outcomes and do not recommend wagers.