Sports
Research MLB player props across every book
Every batter prop is a rate multiplied by a number of plate appearances. Every pitcher prop is a rate multiplied by a number of batters faced. Get the count right before you argue about the rate.
Analytics only. quant+ never places bets or holds funds.
- player. The name on the prop, and his role tonight.
- family. Strikeouts, hits, total bases, runs or runs batted in.
- line. The posted number and whether it is a .5 or a whole number.
- over / under. Both sides, priced separately, never assumed symmetric.
- best side price. The best available price on the side you would take.
- book. Where that price is posted, for the count of books below.
- fair price. Margin removed, so a book price and a model price are comparable.
- projection. The model's own expected count or probability for the same prop.
- movement. Direction and size of the change since the market opened.
- book count. How many books quote this exact prop and line.
The column names a live prop table would carry. No current line, price or book is shown anywhere on this page, and none is claimed.
Baseball props are not one distribution with five labels
The five common families share a scoring structure and almost nothing else. Each has its own unit of opportunity, its own shape, and its own characteristic mistake, and a research process that treats them as interchangeable will be confidently wrong in at least three of them.
Strikeouts
Unit: Batters faced, capped by pitch count
The line is set against a workload nobody has published
A strikeout prop is a count of successes over a number of trials, and the number of trials is the pitcher's batters faced, which is not fixed. It depends on how deep he goes, which depends on his pitch count, his effectiveness and the score. A model that assumes a full six innings when the market assumes five and a third has an error in the trial count rather than in the rate, and that error is larger than most rate disagreements.
Hits
Unit: At-bats, minus walks and sacrifices
The base rate is high and the price is close to even
A hit prop is close to a coin flip for most regulars, because the average outcome sits near the posted line. When the two sides are nearly symmetric, the book's margin is a large share of the edge available, and a small error in the probability estimate is enough to flip the sign of the expectation. This is the family where the price comparison matters more than the projection.
Total bases
Unit: A bounded weighted count
Zero, one, two, three and four are not equally likely
Total bases weight each hit by its value, so the outcome is a small integer with a long right tail from home runs. The distribution is strongly skewed and overdispersed relative to a simple model, and a mean is a poor summary of it. Pricing the over on a whole number line versus a half number line is a genuinely different proposition, because one of them can push and the other cannot.
Runs
Unit: Team run context, then individual share
It depends on teammates you are not modelling
A run is scored by the player but produced by the lineup around him. Getting on base is necessary and not sufficient, and the conversion from on-base to run scored depends on who bats behind and how the game is managed. It is the most context-dependent of the five families and the one least well served by an isolated per-player rate.
Runs batted in
Unit: Opportunity-driven, high variance
Opportunities are the whole story
Runs batted in require runners on base at the moment the batter comes up, which is a function of everyone batting ahead of him, the score, and the inning. It is the family with the largest gap between a player's talent and his opportunity, and therefore the one where a model built on opportunity rather than on hitting skill has the most to add.
Batting order is a plate appearance forecast in disguise
A batter prop is an attempt count multiplied by a per-attempt rate, and the attempt count is not a property of the player. It is a property of where he bats, how often the team turns the lineup over, and whether the game goes long enough for him to come up again. Two hitters with identical skill can have different fair prices for the same line purely because one bats second and the other bats eighth.
This is why a research surface should carry the batting spot as a first-class field rather than as a note. It is the single input that most often explains an apparently surprising line, and it is the input most often left implicit. When a line moves after a lineup card is posted, the move is frequently about the count rather than about the player.
The removed-at-bats problem is the other half of it. A pinch hitter, a defensive substitution or an ejection can take an at-bat away, and a lopsided score can take a whole plate appearance away by ending the game before the spot comes around. Those are not rare events. They are a systematic downward pull on the projected count for players near the bottom of the order, and a model that ignores them will be biased high on exactly those players.
- Spot in the order. First, second, third, cleanup and so on.
- Expected plate appearances. Roughly four, occasionally five, sometimes three.
- What can remove one. Pinch hitter, defensive substitution, ejection, or the game ending.
- What can add one. Extra innings, or a high-scoring game with many base runners.
- What it does to a rate prop. Scales the expected count directly.
A strikeout prop is a fraction with two uncertain numbers
The rate is reasonably stable. The denominator is not. Most disagreements about a strikeout line are really disagreements about how long the pitcher will work.
Trials, not innings
The relevant denominator is batters faced rather than innings pitched, because a pitcher who allows two base runners in an inning and escapes it has faced extra batters without recording extra outs. Modelling by innings quietly assumes a clean inning every time.
The cap is real
Managers remove starters on pitch count regardless of how well they are throwing, so the distribution of batters faced has a soft ceiling rather than a free tail. A model that lets a hot pitcher continue indefinitely is pricing a game that will not be managed that way.
Opponent, not just pitcher
Strikeout rate is a pitcher attribute and also an opponent attribute. The lineup across from him carries its own strikeout tendency, and the same pitcher against two different lineups is two different props. Treating a season rate as opponent-neutral is the most common single error in this family.
Nothing here is a claim about any particular pitcher, lineup, park or price. These are structural statements about how the market is built.
A fair price and a book price are different objects
A book price carries a margin. A model price does not. Comparing them without removing the margin will report an edge where there is only a cost.
Convert first
American odds, decimal odds and implied probability are three encodings of one number. Convert everything to probability before comparing anything, because prices in different formats are not otherwise comparable.
Remove the margin
The two sides of a prop sum to more than one hundred percent. The excess is the book's margin, and it has to be apportioned before a fair price exists. Whichever method is used, name it, because different methods produce different fair prices from the same market.
Then compare like with like
A model projection becomes an edge only after it has been expressed as a probability and placed next to a fair price. A projection that beats the raw book price but not the fair price is not an edge at all.
Weigh the sample
A prop quoted by three books and a prop quoted by twenty are different research problems. Thin markets move further and settle less precisely, and a large apparent edge on a thin market is more likely to be a stale quote than a discovery.
For the arithmetic of implied probability and margin removal, see the fair odds calculator and the de-vig calculator.
What a prop research pass checks, in order
A repeatable order matters more than any individual step, because it stops a projection from being built to justify a price that was noticed first.
- 01. Confirm the family and the unit. Establish what is being counted and over what opportunity. This decides everything downstream and is the step most often skipped.
- 02. Establish the opportunity count. Batting spot and expected plate appearances, or expected batters faced with a workload ceiling. This is the denominator.
- 03. Build the rate from the right sample. Component rates with the sample weighted sensibly, adjusted for the opposing pitcher or lineup. Not a season average carried forward unchanged.
- 04. Apply the environment. Ballpark and weather where the family is sensitive to them. Total bases and runs are the most sensitive; strikeouts are the least.
- 05. Convert to a fair price. Turn the projection into a probability, remove the margin from the market, and only then form an opinion.
- 06. Check the movement and the book count. A line that has already moved has already absorbed information. A line offered by three books has not been arbitraged to the same degree.
What this cannot tell you
Research reduces uncertainty. It does not remove it. Baseball generates a very large amount of randomness per game, and a prop that a model priced correctly will still lose a substantial fraction of the time. A single result carries almost no information about whether the process was sound, which is why the evaluation has to be over many props rather than over one good night.
Some inputs are genuinely unavailable before the market moves. Betting lineups are posted close to the game, weather shifts, and a late scratch changes the count. The research surface can only work with what is public at the time, and a model that assumes it will always see the lineup first is assuming something that is not true.
And thin markets are thin for a reason. Where a prop is quoted by few books, the price is a less reliable estimate of anything, including of itself. The right reading of a large apparent edge in a narrow market is usually that the market is uncertain, not that the model has found something.
quant+ publishes research and comparison surfaces. It does not place bets, does not hold funds, and has no account connection to any sportsbook. The game-level side of baseball markets is covered in MLB betting models.
Scan prop markets · Compare prop prices
Count the opportunity, then price the skill
Put the posted line, the book count and your own projection side by side, with the margin removed so the comparison is honest.
Questions people ask
What does an MLB player prop research tool actually do?
It collects the posted line, both sides of the price, the number of books offering the prop and any movement, then places that next to a model projection that has been converted into the same units. The value is in putting a fair price beside a book price so the difference is visible, rather than in producing a single number.
Why does batting order matter so much for hitter props?
Because a hitter prop is a rate multiplied by a number of plate appearances, and batting order is the main determinant of that number. A spot near the top of the order means more trips to the plate over a season and a higher chance of a fifth appearance in a given game. Two hitters of equal skill in different spots do not have the same fair price for the same line.
How are strikeout props different from hitter props?
Strikeout props are capped by workload rather than by lineup position. A manager removes a starter on pitch count regardless of results, so the number of batters faced has a soft ceiling and a much shorter tail than a hitter's plate appearances. The rate also depends heavily on the opposing lineup's own strikeout tendency, which a season rate does not capture.
What is the difference between a whole number line and a half number line?
A half number line cannot push, so one side must win. A whole number line can land exactly on the number, and depending on the market that either pushes and returns the stake or is resolved by a stated rule. That makes the two lines different problems, and the price difference between them is not arbitrary.
Why do runs and runs batted in deserve extra caution?
Both depend on what the rest of the lineup does rather than only on the player being researched. A run requires getting on base and then being driven in or advancing around; a run batted in requires runners on base at the right moment. Those are opportunity inputs that a per-player rate does not contain, and they are the largest source of error in the two families.
Can MLB props be beaten consistently?
The widely quoted prop markets are priced competently, and the margin on a near-even prop is a large share of any edge available. Any genuine advantage is more likely to come from correct handling of the opportunity count and the environment than from a better estimate of a player's skill, and it has to be demonstrated against fair prices over a large sample rather than inferred from results.
Sources
- SlipScore | Player Prop Research Tool for Sports Betting, slipscore.io
- Starting pitcher | Wikipedia, en.wikipedia.org
- How to build an MLB betting model | Rithmm, rithmm.com
