MLB · Current experiment

The only sport VigPulse currently runs

MLB moneyline is the entire current scope of the experiment, and it runs on paper. Baseball was chosen first because it produces a large number of independent, well-priced events on a daily cadence, which is the fastest way to accumulate evidence about a method.

Current capabilityThis page describes what VigPulse actually does today: MLB moneyline, paper-only, with a public ledger.

Cadence: Daily, roughly 15 games
Primary market: Moneyline
Main driver: Starting pitcher and lineup
Liquidity: High on major markets

Current status

What is actually running

This is the whole of it. One sport, one market, on paper, with the ledger published in full.

Sport: MLB
Market: Moneyline
Stake type: Paper only
Graded entries: 33

Paper only, and early

No real money has been staked. The published record is 33 graded paper entries (9W-24L, +2,417.5u), and the pre-game closing-line-value sample is 6 entries. That is not enough to demonstrate anything, and it is not presented as though it were.

Read every row of the ledger →

The record

Where the units actually came from

The headline unit total is real, and on its own it is misleading. Splitting the ledger by phase is the fastest way to see why, so the split is published here rather than left for a reader to reconstruct.

Every graded MLB entry, grouped by whether it was logged before first pitch or during the game. Derived from the published ledger.
Phase	Entries	Record	Paper units	CLV reported
Pre-game entries	6	0W-6L	-600u	Yes
In-play entries	27	9W-18L	+3,017.5u	No — n/a*

The pre-game arm lost every bet

Uncomfortable

All 6 pre-game entries graded as losses, for -600u. Those same 6 entries produced positive closing line value on every one, averaging +4.93%.

Both statements are true at once, and that is precisely why CLV is measured separately from profit. Across six bets, profit is close to pure noise. Six positive CLV readings are a weak hint, not a result.

Every unit came from in-play

Read carefully

The 27 in-play entries account for +3,017.5u, which is the entire positive total. VigPulse offers no in-play capability; those rows are a record of what the experiment did, not a product.

CLV is never reported on them. An entry logged after first pitch is priced against a game already in progress, so the gap to the close measures innings elapsed rather than price discovery.

Three winners carry the total

Concentration

The three largest winners contribute +2,725u of +4,217.5u in winning units, or 64.6% of all upside. The single largest is +1,100u.

A total that concentrated is not a steady edge. Remove those three rows and the experiment is comfortably negative.

What the ledger does not contain

Two schema fields were never captured, and they render as Not recorded on every row rather than being reconstructed: the sportsbook each price came from (0 of 33 recorded) and the model version behind each entry (0 of 33 recorded). Both gaps materially limit what this record can support, which is why they appear as open gates further down the page.

Market structure

Why MLB was chosen first

Sports are not interchangeable inputs to one model. The structure of the sport determines what can be modelled, how fast evidence accumulates, and where a market is likely to be wrong. Baseball was picked for the second of those reasons above all.

Volume is the point

A daily fifteen-game slate produces far more gradeable events per week than a weekly sport. Since the limiting factor on learning anything about a betting method is sample size, a sport that generates more samples per unit of calendar time is worth more than one with better-understood teams.

Moneyline is the cleanest market

Baseball has no spread in the football sense, and the run line is a fixed -1.5 rather than a moving number. That makes the moneyline the primary market and the one where book disagreement is most directly comparable.

Scheduled information dominates

Lineups post at a known time before first pitch, and probable starters are published in advance. Because the biggest information events are scheduled, the repricing they cause is observable rather than random, which makes movement easier to attribute.

Weather matters more than in most sports

Wind and temperature move totals substantially and moneylines somewhat. Outdoor stadiums with distinctive geometry amplify this, which is a genuine source of disagreement between books that model it differently.

Method

Reading a baseball market

The sequence below is the one described on the methodology page, written for this sport. It explains how a price is read. No proprietary formula is published, and none of this asserts that the reading works.

Market to decision, in baseball terms

6 steps

Market
Collect the moneyline on both sides of a game. Baseball helps here: there is no moving spread to track alongside the price, so a game is two numbers rather than a number and a line.
Consensus
Remove the vig from each two-sided price so the implied probabilities sum to one, then combine across books. The de-vigged consensus is the reference point; a raw posted price never is.
Movement
Track how that consensus travels between the opener and first pitch. Because lineups and probable starters post on a schedule, most large baseball moves can be attributed to a known event rather than written off as noise.
Model
Form an independent estimate of the same probability. In baseball the dominant inputs are the starting pitcher, the posted lineup, the park, and the weather.
Opportunity
Compare the model estimate against the de-vigged consensus. A gap is only interesting if it survives de-vigging, clears the uncertainty band, and does not rest on a single disagreeing book.
Risk
Size to the uncertainty rather than to the conviction. A daily fifteen-game slate makes correlated exposure easy to accumulate without noticing, which is a bankroll problem before it is a modelling one.

Where to see it laid out

Illustrative

The Pulse board and the opportunity board render the same sequence on screen. Both run on invented sample data at a fixed snapshot and carry a demo label on every screen.

Check the math yourself

The de-vigging, expected value, and staking steps above are standard and free to verify. The calculators run entirely in your browser, and the academy explains each concept in plain language.

Difficulties

What makes baseball hard

Known problems

4 listed

Individual game outcomes are extremely noisy. Even a strong team loses roughly four games in ten, so results carry very little information per game.
Bullpen usage is partially unobservable before the game and can matter more than the starter.
Late scratches can invalidate an entry entirely, and the void rules differ between books.
The daily volume that makes baseball attractive also means a slow grind of small decisions rather than a few high-conviction ones.

What would make the record meaningful

3 listed

A far larger sample than the current published ledger before any claim about edge would be meaningful.
Consistent capture of which book each entry came from, which the current record does not have.
Pre-game entries as the majority of the sample rather than the minority.

Readiness

Open gates on the MLB experiment

Two of these are held. Four are not. None of them carries a committed date, and the experiment stays on paper until they do.

Conditions the MLB paper experiment has and has not met. Status is read from the published ledger, not asserted.
Gate	Status	Evidence
Publish every graded entry, including the bad ones	Held	All 33 graded rows are published in full, with no filtering and no selection.
Stay on paper until the record means something	Held	No real money has been staked on any entry in the published record.
Record which book each entry was taken at	Not met	Book is unrecorded on all 33 rows and renders as "Not recorded" rather than being filled in after the fact.
Record a model version against each entry	Not met	Model version is unrecorded on all 33 rows, so no result can be attributed to a specific version.
Make pre-game entries the majority of the sample	Not met	6 of 33 entries are pre-game. The other 27 are in-play, where closing line value is not a meaningful reading.
Accumulate a sample large enough to test a claim	Not met	6 pre-game CLV readings cannot separate an edge from chance, and are not presented as though they could.

Comparison

How the other sports compare

Every other sport on this site is research roadmap with no model, no data pipeline, and no published record. The table below is the short version of why baseball came first: it is the row that produces the most gradeable events per week.

Structural comparison across the five sport pages. Only MLB is current, and MLB is paper-only.
Sport	Status	Cadence	Primary market	Liquidity
MLB	Current · paper	Daily, roughly 15 games	Moneyline	High on major markets
NFL	Roadmap	Weekly, roughly 16 games	Point spread	Highest of any US sport
NBA	Roadmap	Daily, roughly 8 games	Point spread and total	High
WNBA	Roadmap	Seasonal, several games per week	Point spread and total	Lower than NBA
Tennis	Roadmap	Near-continuous, year-round	Match winner	Varies enormously by event tier

Current: paper experimentNo picks soldNo wagering offeredNo book integration