I Used the No-Vig Tool for a Week and Lost Several Arguments

The week I brought the no-vig tool into every conversation #

I committed, on a Sunday night and against the advice of several friends, to running every two-sided market I bet through the no-vig conversion tool for one week. The tool is small. It takes the two prices the book is offering — say -110 and -110 on a spread, or -130 and +110 on a moneyline — and returns the implied probabilities with the hold removed. The point of the tool is to produce a fair probability against which the bettor s own estimate can be compared. The point of running it for a week was to discover, in public and with a tracking sheet open, how often my estimates beat the tool s and by how much.

The first thing the tool did was make me less popular. The second thing it did was end several arguments earlier than they would otherwise have ended. The third thing it did was reduce my average claimed edge from a healthy-sounding nine percent to a less photogenic four percent, and then, once I forced myself to add an honest uncertainty band, to a barely-noticeable one percent. The arc of the week was, in retrospect, a tutorial on the difference between thinking I had an edge and being able to defend that I had an edge in a way that did not involve raising my voice.

The market mechanic the tool exists to expose #

A sportsbook does not post the fair price. It posts the fair price plus the rent. The rent is the vig, and the vig is the reason the book can survive a week in which bettors are right about most of the action. On a standard -110/-110 spread the vig is about 4.5% per side, which sounds modest until you realize that the bettor needs to win about 52.4% of bets just to break even ¹. The arithmetic is well known and rarely felt by the bettor whose handicapping voice describes a fifty-five percent estimate as a solid edge. The fifty-five percent estimate beats the market by about three percentage points. The book is keeping more than that in hold on the worst-case spread.

The conversion itself is mechanical. Given two American odds, the function noVigPair returns the pair of implied probabilities after the hold has been split off proportionally ²:

function americanToImplied(odds: number): number {
	return odds > 0 ? 100 / (odds + 100) : -odds / (-odds + 100);
}

// No-vig conversion: scale both implied probabilities so they sum to 1.
function noVigPair(home: number, away: number): [number, number] {
	const h = americanToImplied(home);
	const a = americanToImplied(away);
	const total = h + a;
	return [h / total, a / total];
}

The mechanic gets steeper as the markets get more exotic. A player prop carries roughly eight percent vig per side; a same-game parlay carries more; a PrizePicks-style flex parlay carries hold in the double digits before the legs are even considered. The bettor who places these without converting to no-vig is implicitly assuming her estimates are accurate enough to overcome the hold, which is a much taller order than the postable two-sided format makes it look. The tool exists to make the height of the order visible at placement time.

What the diary recorded that my arguments did not #

I kept the no-vig diary as a six-column sheet: market, posted prices, no-vig fair, my estimate, uncertainty band, and a decision column with one of three values — take the bet, pass, or revisit. The most consistent finding across the week was that my estimates were closer to the posted price than I had been claiming in arguments. On about half of the lines my estimate beat the no-vig fair by less than two percentage points; on most of those, my uncertainty band was wider than the gap. The diary made the gap visible. The arguments had been hiding it.

The bet I most wanted to take that week was a college spread on a team I had been watching closely for a month. My estimate said the fair line was about a point and a half better than the posted spread. The no-vig tool agreed with the posted line within a quarter point. The uncertainty band on my estimate was, when I forced myself to write it down honestly, about two points wide. The bet survived the conversation. The bet did not survive the conversion. I passed, the team covered comfortably, and the bet would have been a positive-CLV win. The diary did not care. The diary said, correctly, that the bet was a coin flip wearing a confidence rosette and the pass was the right action regardless of the outcome.

The uncertainty band that ended the most arguments #

The single piece of the workflow that most consistently changed my behavior was the uncertainty band. The tool returns a no-vig fair number. My estimate returns my own number. The uncertainty band turns my number into a range, and the rule I committed to was that the bet was only allowed to exist if my range did not include the no-vig fair. The rule sounds restrictive. It is restrictive. It is also the only rule I have used that prevents me from arguing my way into bets where my point estimate beats the market by a number smaller than my own honest uncertainty about my own estimate.

The first effect of the rule was to roughly halve the number of bets I was willing to place. The second effect, which I noticed by the end of the week, was that the bets that survived the rule were materially better than the bets that had not. I cannot prove this from a sample of twenty-four, but I can claim it directionally: the surviving bets had wider gaps between my estimate and the no-vig fair, the surviving bets generally moved my way in the market between placement and close, and the surviving bets composed a noticeably smaller share of the slate. The discipline traded volume for defensibility, which is the trade I would not have made without the diary in front of me.

The tool did not make betting simple. It made pretending simpler things were edges a little harder. That counts as progress in this household.

The protocol that survived contact with my friends #

The protocol I am keeping past the experiment is the protocol that emerged on Thursday afternoon, when one of my friends sent me a screenshot of a prop he wanted me to like and I ran it through the tool before responding. The prop had about an eight percent edge by his estimate, which deflated to about negative one percent after no-vig conversion. The conversation that followed was shorter than usual and ended without either of us placing the bet. The protocol is, in three steps: convert any two-sided market before discussing it, write my estimate alongside the no-vig fair, and apply the uncertainty band before deciding. The protocol takes about thirty seconds per market. It has, so far, prevented at least one bet per day that I would have placed without it.

This is, like the rest of these experiments, a discipline that produces no photographs and few stories. It produces a slightly smaller bet count, a slightly cleaner argument, and a slightly more honest description of the bettor s own edge. The bankroll will catch up to these effects, slowly, over a season. The bettor catches up to them faster, because the diary forces the catching-up to happen at the moment of placement rather than at the moment of settlement. The moment of placement is the only moment the bettor can change. Settlement is the moment everything else has already happened.

The closing argument #

The tool did not make me a sharper bettor in any cinematic sense. It made me a slower one, and a more accurate one in the way that flossing makes a person s gums more accurate. The week ended with fewer bets, more arguments declined, and a bet log in which the surviving entries had wider edge gaps and shorter uncertainty bands than the entries from the week before. The bankroll did not move much in either direction; the workflow improved more than the result did, which is exactly the order in which these experiments are supposed to produce their effects.

I will keep the protocol. I will keep running every two-sided line through the conversion before discussing it. I will keep being slightly insufferable when a friend wants to argue about a posted price without first converting it. The insufferability is the workflow s side effect, not its point. The point is the discipline of not believing the bettor s own edge claim before the math has been done, which is the smallest and most reliable favor the bettor can do for the bankroll across the noisy, opinionated, well-marketed bottom of a long week.