Results methodology

How the ledger is recorded and graded

A published record is only worth as much as the method behind it. This page documents exactly how each of the entries on the results page was captured, sized, settled, and measured — including the parts of the method that are weak, and the two fields that were never recorded at all.

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

Reviewed: 23 August 2026
Rows documented: 33
Fields in schema: 14
Fields never filled: 2

Principle

Record first, explain second, never backfill

Three rules govern the file behind the results page, and the third is the one that costs something.

One source of truth

No hand-typed stats

Every figure published anywhere on this site is computed from the ledger file at build time. No aggregate is typed into a page. If the file changes, the headline numbers change with it, including in the wrong direction.

Every graded row is published

No selection

All 33 graded entries appear on the ledger. There is no filter, no featured subset, and no archive of entries that did not work out.

Gaps stay gaps

Not recorded

A field that was not captured at the time is never reconstructed afterwards. A value recovered from memory is a guess wearing the costume of a record, and it would contaminate the only asset this project has.

Capture

What is recorded, and when

An entry is written when the process flags a price. It is not revisited, and the row is not edited after the fact except to add the settlement.

Every field stored per entry, what it means, and how consistently it was actually populated across the 33 rows.
Field	Meaning	Populated
timestamp	When the entry was logged, in UTC.	Every row
settledAt	When the result was graded, in UTC. Drives the display order on the ledger.	Every graded row; null while open
sport / market	Competition and market type.	Every row
bet	The selection, written as the fixture and the side taken.	Every row
entry	Decimal price at the moment the entry was logged.	Every row
close	Decimal price the market settled at before the event started.	Every row
phase	PRE if logged before first pitch, INPLAY if logged after it.	Every row
result / score	Win or loss, and the final score used to grade it.	Every graded row; null while open
units	Profit or loss in paper units, positive for a win, negative for a loss.	Every row
clv	Percentage gap between entry price and closing price. Published for pre-game rows only.	Stored on every row, published on PRE rows
paper	Whether real money was staked. It never was.	Every row, always true
arm	An internal grouping label carried through from the process that generated the entry.	Every row
book	Which venue the price was observed at.	Never — 0 rows
modelVersion	Which version of the process produced the entry.	Never — 0 rows

Two fields were never recorded, on any row

The venue and the model version are in the schema and are empty for all 33 entries. That means you cannot check the entry prices against a named book, and you cannot tell whether the process changed halfway through the sample. Both are real weaknesses in the record. They render as Not recorded on the ledger and will not be filled in later.

Sizing

How stake size worked

Stake is not stored as a column. It is recoverable from the payouts, so it is reconstructed here rather than described from memory.

Stake sizes used: 50 / 100
Total staked: 2,400u
Net result: +2,417.5u
Return on turnover: +100.7%

That return figure is not a performance claim

A return on turnover above 100% across 33 bets is not evidence of skill; it is evidence of longshots. Entry prices in this file range from 1.22 to 38.00 in decimal terms, and 65% of all winning units came from three entries. At these prices a sample this size has enormous variance in both directions, and this one happens to have landed high.

Entries grouped by the arm label stored in the file. What each label means internally is not published, because the observable facts below are all that the record actually supports.
Arm	Entries	Phase	Stake	Paper units
A	1	INPLAY	100 units	-100u
B	18	INPLAY	50 units	+1,142.5u
C	14	INPLAY and PRE	100 units	+1,375u

Settlement

How entries were graded

Grading is the least interesting part of the method and the easiest place to quietly cheat, so it is spelled out.

Each entry is settled against the final score of the fixture and marked W or L. The score used is stored on the row, so any grading decision can be checked against the actual result rather than taken on trust. There are currently 0 open positions; graded rows are marked W or L and retain the score used for settlement.

Profit is calculated at the entry price, not the closing price. A winner returns the stake multiplied by the decimal price less one; a loser gives up the stake. Pushes, voids, and dead heats do not appear in this sample — the market is moneyline only, over a window of 2026-08-21 to 2026-08-24.

What settlement does not model

No commission or fee is deducted, because no venue is recorded and therefore no fee schedule applies.
No limit check. Every entry assumes the full stake was available at the recorded price. At the longer prices in this file that assumption is doing significant work.
No slippage or fill delay. The entry price is the price observed at the moment of logging, not a price that was actually accepted.
No bankroll dynamics. Stakes are flat within each arm and do not scale with the running balance.

Measurement

How closing line value is defined

CLV is the metric this experiment is actually trying to move, because profit over a few dozen bets is close to pure noise.

Closing line value measures the gap between the price you took and the price the market finished at. If you consistently take prices better than the close, you are finding something before the market does. It produces a reading on every entry, so it accumulates evidence far faster than win-loss does.

CLV % = (entry_decimal / closing_decimal − 1) × 100

Worked example, from the file:
  entry price   3.13
  closing price 2.72
  CLV %  = (3.13 / 2.72 − 1) × 100
         = +15.07%

A positive figure means the price shortened after the entry was logged — the market moved towards the selection. A negative figure means it drifted, and the entry was on the wrong side of the move.

Why in-play CLV is published as n/a*

27 of the 33 entries were logged after first pitch. Comparing an in-play price to the pre-game closing price measures how much of the game has already happened, not whether the entry read the market well. The largest such gap in this file is +1839%.

Publishing numbers of that size as closing line value would inflate the headline by an enormous factor, and it would be the single most misleading thing this site could do. So in-play rows show n/a* with the reason stated beneath the table, and the average CLV figure is computed from pre-game rows only.

Known limits

Everything wrong with this record

Stated in one place, so nobody has to work it out from the footnotes.

Sample and scope

Severe

33 entries over roughly four days is a working log, not a track record.
One sport and one market: MLB moneyline.
Only 6 entries carry a meaningful CLV reading.
The unit total rests on three winners at long prices. Variance dominates everything at this size.

Method and recording

Material

Venue never recorded, so entry prices cannot be independently verified.
Model version never recorded, so a mid-sample change to the process would be invisible.
Stake availability is assumed rather than tested; no limit was ever checked.
Paper only. No real fill, no account restriction, and no closed account — the three things that most often destroy a strategy that looks fine on paper.

The same split published on the results page, repeated here because it is the single most important thing to understand before reading the totals.
Phase	Entries	Wins	Paper units	CLV published
Pre-game entries	6	0	-600.0u	Yes
In-play entries	27	9	+3017.5u	No — n/a*

Audit and corrections

How to check this, and how errors get fixed

A record nobody can audit is a marketing asset, not evidence.

Auditing it yourself

What is checkable

Every row is published with its fixture, score, entry price, and closing price. Scores can be checked against any public box score.
Unit figures can be recomputed from the entry price and the reconstructed stake, and the arithmetic on this page shows how.
CLV can be recomputed from the entry and closing prices using the formula above.
Entry prices cannot be verified, because no venue was recorded. That is the limit of what this record can prove.

The correction policy

Public, not quiet

Report an error and it gets checked against the source, whichever direction it moves the numbers.
Confirmed corrections are republished with the correction noted, not edited silently.
No row is ever removed because it looks bad. Removing losers is the failure mode this whole page exists to make impossible.
If a correction changes a headline figure, it changes everywhere, because every figure is derived from the same file.