How to Calculate EV on Player Props (Step by Step)

If you are betting player props without calculating expected value, you are gambling. If you are calculating expected value, you are investing. That is not hyperbole. Expected value (EV) is the single most important concept in profitable sports betting, and understanding how to calculate it on player props is what separates the 3% of bettors who win long-term from the 97% who do not.

In this guide, we will walk through the EV calculation step by step using real player prop examples. By the end, you will understand implied probability, true probability, edge, and EV percentage, and you will know why you need an independent probability model to do this correctly.

The EV Formula

Expected value tells you how much you expect to profit (or lose) per dollar wagered over many bets. The formula for a standard player prop is:

EV% = (True Probability x Payout) - 1

That is it. Three inputs: your estimated true probability of the bet winning, the payout you receive if it wins (expressed as a decimal multiplier), and the number 1 (representing your original stake). If the result is positive, the bet has positive expected value. If it is negative, the book has the edge. Let us break down each component.

Step 1: Convert American Odds to Implied Probability

Before you can calculate EV, you need to understand what the book thinks the probability is. This is called implied probability, and it is derived from the odds the book offers.

For American odds, the conversion formulas are:

Negative odds (favorites): Implied Probability = |Odds| / (|Odds| + 100)

Positive odds (underdogs): Implied Probability = 100 / (Odds + 100)

Let us work through a real example. Say you see the following prop on DraftKings:

Luka Doncic Over 27.5 Points: -115

The odds are -115, so we use the negative odds formula:

Implied Probability = 115 / (115 + 100) = 115 / 215 = 53.49%

This means DraftKings is pricing this prop as if Luka has a 53.49% chance of scoring over 27.5 points. But that 53.49% includes the book's vig (their margin). The actual break-even probability is 53.49%, but the true probability of the event could be higher or lower.

Let us also convert the payout. At -115 odds, if you bet $115 to win $100, your total return on a win is $215 on a $115 bet. As a decimal multiplier:

Payout = (100 / 115) + 1 = 1.8696, or approximately 1.87x

For DFS books like PrizePicks, the payout is simpler: a standard pick pays 1.84x regardless of the line. Underdog Fantasy pays 1.86x. Some DFS books like ParlayPlay and Sleeper offer variable payouts ranging from 1.58x to 2.05x depending on the prop.

Step 2: Estimate True Probability

This is where the hard work happens, and where most bettors fall short. The implied probability from the book is not the true probability. It is the book's pricing, which includes vig and is influenced by where the money is flowing. Your job is to estimate the actual probability independently.

Let us continue with our Luka Doncic example. To estimate his true probability of scoring over 27.5 points, you might consider:

Recent form: Luka has averaged 31.2 points over his last 5 games. His season average is 28.8. Weighting recent performance more heavily (say 50% last 3 games, 30% last 5, 20% season) gives you a weighted average around 30.1 points.

Matchup: Tonight's opponent ranks 25th in points allowed to opposing point guards. They play at the 6th-fastest pace in the league. Both factors push Luka's projection up.

Game context: Is this a back-to-back? Is Kyrie Irving out, meaning more usage for Luka? Is the game expected to be competitive (keeping starters in) or a blowout (early bench time)?

After incorporating all of these factors, suppose your model projects Luka at 30.4 points with a standard deviation of 7.2. Using a probability distribution (the math behind converting a projection and variance into a probability), you calculate that Luka has a57.5% true probability of going over 27.5 points.

Step 3: Calculate the Edge

The edge is simply the difference between your estimated true probability and the book's implied probability:

Edge = True Probability - Implied Probability = 57.5% - 53.49% = +4.01%

A positive edge means you believe the event is more likely to happen than the book thinks. But edge alone does not tell you how profitable the bet is. For that, you need to factor in the payout.

Step 4: Calculate EV%

Now we plug into the EV formula:

EV% = (True Probability x Payout) - 1

EV% = (0.575 x 1.87) - 1 = 1.0753 - 1 = +7.53%

This means that for every dollar you bet on this prop, you expect to profit 7.53 cents over the long run. On a $100 bet, that is $7.53 of expected profit. Over 1,000 similar bets at this EV level, you would expect to profit roughly $7,530 on $100,000 in total wagers.

Is +7.53% EV good? At Turtle +EV Labs, we use a 6% EV threshold for NBA picks (8% for NHL, which has higher variance). Anything above 6% clears our bar. This Luka prop at +7.53% would make it onto our dashboard.

A Second Example: NHL Shots on Goal

Let us run through another example in a different sport to reinforce the concept.

Auston Matthews Over 4.5 Shots on Goal: +110 (PrizePicks, 1.84x payout)

First, the implied probability from +110 traditional odds: 100 / (110 + 100) = 47.62%. But on PrizePicks, the payout is fixed at 1.84x regardless of the line, so the implied break-even probability is 1 / 1.84 = 54.35%.

This is an important distinction. On PrizePicks, the vig is baked into the fixed payout, not the line itself. Every prop on PrizePicks requires a 54.35% win rate just to break even.

Now for the true probability. Matthews has averaged 4.9 SOG over his last 5 games, with a season average of 4.4. He is facing a team that allows the 5th-most shots per game. Your model projects him at 5.1 SOG with a standard deviation of 2.3. Running that through a probability calculation, you get a 58.2% true probability of going over 4.5.

EV% = (0.582 x 1.84) - 1 = 1.0709 - 1 = +7.09%

Another profitable play. Notice how different the payout structure is between a traditional sportsbook (-115 odds = 1.87x) and PrizePicks (flat 1.84x). That three-cent difference in payout changes the EV calculation. This is why scanning multiple books matters. The same prop might be +EV on one book and -EV on another purely because of the payout.

Why You Need an Independent Probability Model

Some bettors try to shortcut the EV calculation by using one sportsbook's line as a proxy for "true probability" and betting wherever they find a better price at another book. The classic version of this is using Pinnacle's lines (widely considered the sharpest) as your true probability and betting whenever another book offers a better price.

This approach has two problems:

Pinnacle is not truth. Pinnacle's lines are sharp, but they still include vig, they still have biases, and they still misprice props. Using Pinnacle as your probability source means you are assuming their line perfectly reflects reality. It does not. Pinnacle's edge is that they are less wrong than other books, not that they are right.

Pinnacle does not cover most player prop markets. For DFS platforms like PrizePicks and Underdog Fantasy, there is no Pinnacle line to compare against. These are completely separate markets with their own pricing. You need a model that can independently assess the probability of a player going over or under a given line, regardless of what any book says.

The right approach is to build or use a model that starts with player data, not book data. You want a model that looks at a player's historical performance, recent form, matchup context, pace, rest, and injury status, and produces a probability estimate from scratch. Then you compare that probability against every available book's line to find where the edge exists.

This is exactly what Turtle +EV Labs does. Our models are sport-specific. Our NBA model uses different calibration parameters than our NHL model, which uses different parameters than our MLB model. Each sport has its own probability function tuned to the actual distribution of outcomes in that sport. We do not borrow a generic model and slap it on every market.

The Calibration Problem

Here is something most EV guides will not tell you: getting the raw probability right is only half the battle. The other half is calibration.

Calibration means that when your model says something has a 60% probability, it actually happens 60% of the time. Most models are poorly calibrated. They might identify the right direction (correctly favoring the OVER or UNDER) but assign probabilities that are too extreme or too conservative.

A model that says a prop has an 80% probability when the true probability is 65% will calculate a huge EV and bet aggressively. But when the results come in, the win rate will be 65%, and the overbetting will have destroyed the bankroll.

At Turtle +EV Labs, we have found that OVER predictions are systematically overconfident across every sport we model. Our NBA model, for example, applies direction compression: OVER probabilities are scaled by 0.55x and UNDER probabilities by 0.85x, with a hard cap at 58%. Without these adjustments, our OVER win rate was significantly below what the model predicted. With them, our calibration is tight and our 57.2% overall win rate closely matches what the model projects.

This kind of calibration work requires thousands of graded picks to validate. You cannot do it after 50 bets. You need 5,000+. We have graded over 50,000 picks across all sports, which gives us the statistical power to detect and correct these biases.

Putting It All Together: A Decision Framework

Here is the step-by-step process for evaluating any player prop:

1. Identify the prop: Player, stat type, line, direction (OVER/UNDER).

2. Get the payout: What does the book pay on a win? Convert to decimal multiplier if needed. For DFS books, this is usually fixed (PrizePicks 1.84x, Underdog 1.86x). For sportsbooks, convert from American odds.

3. Estimate true probability: Use a model based on player data, matchup context, and recent form. Do not use another book's line as a proxy.

4. Calculate EV: EV% = (True Probability x Payout) - 1. If positive, the bet has positive expected value.

5. Apply a threshold: Not every +EV bet is worth taking. Variance is real. A bet with +1% EV will take thousands of trials to realize. We use a 6% minimum (8% for NHL) to ensure our picks have enough edge to overcome short-term variance.

6. Track everything: Log every bet, every result, every payout. Calculate your actual ROI by sport, by stat type, by direction. If something is not working, cut it.

Why Manual EV Calculation Is Not Enough

You can absolutely calculate EV by hand using the formula above. The problem is scale. On any given NBA night, there might be 400 player props across 8 games and 40 books. That is 16,000 individual EV calculations. And the lines move constantly. A prop that was +EV at 7:00 PM might be -EV by 7:15 PM because the line shifted.

This is why automated tools exist. Turtle +EV scans 40+ sportsbooks every 2 minutes, runs sport-specific probability models, calculates EV on every prop, and surfaces only the picks that clear the threshold. You get a curated feed of +EV props that updates in near real-time, backed by a model that has been calibrated on 50,000+ graded picks.

Understanding the math is critical. You should know exactly why a pick is +EV and what the underlying assumptions are. But executing at scale requires automation.

Next Steps

If you want to go deeper on expected value, read our guide on What Is Expected Value in Sports Betting, which covers the foundational concepts in more detail. If you are ready to start finding +EV player props without doing the math yourself, Turtle +EV Labs does the heavy lifting. Our 57.2% win rate and +5.3% ROI across 50,000+ graded picks are the result of exactly the process described in this guide, running 24/7 across every major sport.