A winning pick and a good bet are not the same thing.
Almost everyone grades a wager by whether it cashed. That is the one input you never held: it arrives after the fact, from a contest you had no hand in, and it describes the sport rather than the choice. The choice was which price to accept, it is complete and fully described before the opening whistle, and it can be graded on its own terms. What follows is the arithmetic that pulls the two apart: a 60 percent record that loses money, a 40 percent record that makes it, and a genuine edge that is still underwater after 500 wagers.
A decision and a result are two different objects
A result is one sample drawn from a distribution. The game had a range of ways it could have gone, exactly one of them happened, and that is the whole of what a settled ticket tells you. A decision is a different object entirely. It is the choice to accept a particular number at a particular price for a particular stake, and it is finished and fully described before the opening tip.
Only one of those two is yours. You pick the price, the sport picks the result, and that asymmetry is the reason the decision is the only half that can be graded honestly: everything that determined whether it was sound was already knowable when you made it. A basketball total decided by a contested three in the closing seconds says nothing about whether the price on that total was generous. The shot fell or it did not. The price was fair or it was not. Those two facts never touched each other.
Because the two are independent, four pairings are possible and all four are ordinary: a sound price that wins, a sound price that loses, a bad price that wins, and a bad price that loses. The two mismatched pairings in the middle are not rare exceptions. A wager with a real edge at standard -110 pricing might win 55 percent of the time, which is another way of saying it loses 45 percent of the time, so close to half of your best decisions are supposed to lose. Reading backwards from the outcome to the quality of the choice has a name worth learning, which is resulting, and it is the single most expensive habit in betting because it teaches you the wrong lesson at exactly the moment you are paying attention.
Sixty percent can lose and forty percent can win
Two bettors each finish a year with 500 one-unit wagers. The first backs heavy basketball favorites priced around -300 and wins 60 percent of them. The second takes football underdogs at +200 and wins 40 percent. The win rates make the answer look obvious. The arithmetic reverses it.
A price of -300 risks 300 to win 100, so it asks for 75 percent and pays only a third of a unit when it lands. Three hundred wins at a third of a unit each is 100 units of profit. Two hundred losses at a full unit each is 200 units gone. The year closes at minus 100 units on 500 units staked, which is a 20 percent loss for someone who was right three times out of every five.
A price of +200 asks for only 33.33 percent, and 40 percent clears that comfortably. Two hundred wins at two units each is 400 units, 300 losses cost 300 units, and the year closes at plus 100 units on the identical 500 staked, a 20 percent profit. Same number of wagers, a win rate twenty points lower, opposite bank balances. Every price carries its own pass mark, and that pass mark is its implied probability, which the break-even calculator returns for any number you type in. What you do with the gap between that pass mark and your own estimate is the subject of positive expected value. This page is only about which of the two questions you were answering.
Grading by the scoreboard, in both directions
The error runs two ways. One is quoting a win rate as a report card, and the other is waving a loss away as bad luck, and both let the outcome stand in for the decision, once when the result flatters you and once when it runs against you. A record given without prices is not a record, it is a count. Sort the two bettors above by win rate and you place the one who lost a fifth of everything staked at the top of the table.
Even when every price is identical, results wander much further than people expect. A bettor with a genuine 55 percent win rate, staking one unit flat at -110, finishes 500 wagers down money roughly one time in eight, because the spread of plausible outcomes over that distance is almost as wide as the profit the edge expects to earn across it. The distribution behind that figure, and the staking rules that let you sit through the bad end of it, are worked out in bankroll management. What matters here is what it does to a record: one unchanging set of decisions produces seasons that read as proof of skill and seasons that read as proof of its absence, and the bettor was the same person in both.
Meanwhile the sample needed to settle the question is enormous. Telling a true 53 percent bettor from a true 50 percent bettor takes on the order of 1,000 wagers, and five NFL wagers a week works out at about a hundred a season, which puts that verdict a full decade away. A verdict that arrives ten years late is not a feedback loop, and that is the practical case for grading something other than the record: the quality of a price is knowable the moment you take it and needs no sample at all.
Grade the price you took, not the ticket you cashed
The practical version is a logging habit. At the moment you bet, record the market, the number, the price, and the stake. At the close, record the number and the price again. The distance between those two entries is a reading you get on every wager, the losers included, which is why a hundred logged prices say more about your process than a hundred results do. Putting the two prices on one comparable scale is the subject of closing line value, and the mechanics of keeping the log itself are covered in tracking bets and ROI.
How you read the log back matters just as much. Sort it by price band rather than by date, and compare each band to its own break-even instead of to 50 percent. A bucket of baseball run line wagers taken around -150 needs 60 percent just to stand still. A bucket of underdog moneylines taken around +150 needs only 40 percent. Averaged together with no regard for price, those two buckets produce a number that means nothing at all. Read separately, they tell you which part of the board you are actually good at, which is the question a season-long win rate can never answer.
Then there is what to do the night a wager loses. The question that produces information is whether you would accept that same price again knowing only what you knew at the time. If the answer is yes, the decision survived the result and nothing needs changing. If the answer is no, something was wrong before the game started and it was there for you to find, which is the only case that deserves to be called a mistake.
The winning weeks are the dangerous ones
This argument cuts both ways, and the pleasant direction is the one that costs people money. The days immediately after a winning streak are when stake sizes drift upward without anybody consciously deciding that they should, which is the same substitution of the result for the decision, only more expensive, because this time it is being acted on rather than merely believed. If the size of your next wager is being set by how the last three landed rather than by a rule you wrote down in advance, that is the signal to stop, and our page on responsible gambling is the right place to go next. Betting should stay something you size before you start and can walk away from mid-run.
Process and results, answered
Why can a winning bet still be a bad bet?
Because the price decides whether a wager was worth making and the game only decides whether it paid. If you accept a number that demands a 75 percent chance on an outcome that is really 60 percent to happen, the ticket can still cash, and you have been paid once for a decision that loses money over any long run of repetitions. The reverse holds just as firmly: a sound price can lose without anything having gone wrong.
Can you make money betting while losing most of your wagers?
Yes, and it is routine among people who bet underdogs. A price of +200 needs to land only 33.33 percent of the time to break even, so a bettor hitting 40 percent on numbers like that is profitable while losing three wagers out of five. What decides the outcome is the relationship between how often you win and how much you are paid when you do.
How many bets does it take before a win rate means anything?
Separating a genuine 53 percent bettor from a coin flip, at conventional confidence, takes on the order of a thousand wagers. Samples of fifty or a hundred are dominated by variance and can show almost anything, which is why a short record is the weakest evidence in betting. A record is also unreadable until it is grouped by price band, because a win rate blended across heavy favorites and long underdogs has no single break-even it can be measured against.
What does variance mean in sports betting?
Variance is the range of results that one identical set of decisions can produce by chance alone. The swing on a single wager is far larger than the profit a winning bettor expects to earn from it, so over any short run the noise is doing most of the work and the skill sits almost invisible underneath it. That is why two people running the identical process can finish a season on opposite sides of zero, and why a 60 percent record and a 40 percent record are both unreadable until you know the prices behind them.
Should I stop betting after a losing streak?
A losing streak on its own carries almost no information about whether the decisions behind it were sound, because a run of five or six losses is an ordinary event at any realistic win rate and arrives without anything having changed. The reason to stop is never the run itself, it is what the run is doing to how you decide, and the sequence of wins and losses is the part of a bad week least worth studying. If the size of your next wager is being set by the last few results, or you cannot look back over the prices you took calmly, then stopping is correct for reasons that have nothing to do with arithmetic.
Is a 60 percent win rate good in sports betting?
It depends entirely on the prices behind it, and on its own the number cannot be read at all. Sixty percent on wagers priced at -300 is a losing season, since that price asks for 75 percent, while 60 percent at -110 would be an extraordinary one against a break-even of 52.38 percent. Ask what the average price was before you call any win rate good or bad.
How do you judge a bet if the result does not tell you?
Judge it on the price you accepted, measured against the best estimate of the true chance that was available when you accepted it. That question has an answer whether the ticket cashed or not, which is exactly what makes it more useful than the scoreboard. When the honest answer is that the same number would not be worth taking again, the work is to name the thing that was knowable before kickoff and went unchecked, because that is the only part of the night you can do anything about.
Judge the decision, not the night
Our boards put a model probability next to the market price, because the comparison that decides whether a wager is worth making happens before the game starts rather than after it ends.