Lakeshore Edge Create Free Account
Field Guide · Reading a Model

Why 49.4% Wins More Money Than 53%

Every sports-betting "expert" online brags about their win rate. Most of them are losing money. Our model is currently at 49.4% on 310 settled picks, under a coin flip, and up +$2,273. Those two numbers aren't a contradiction. Here's the math, and the one metric that actually predicts whether a model makes money.

By Lakeshore Edge · 9 min read

TL;DR Win rate without odds context is theater. A bettor hitting 65% on -200 favorites loses money. A bettor hitting 49% on +120 dogs makes money. What you actually want is positive expected value, and the cleanest leading indicator of that is closing-line value (CLV) — not hit rate.

The 53% breakeven myth

You'll see the number 53% in almost every "how to bet sports" article. The idea: at standard -110 juice, you need to win 52.4% of your bets to break even, so 53% means you're profitable. Round it up and call it the magic number.

That's true. But it only applies if you bet -110 on every single game. Nobody does that. Real breakeven changes with every line. Here's the full picture:

Breakeven win rate by American odds
30% 40% 50% 60% 70% 80% +300 +150 +100 -110 -150 -200 -300 +150 @ 40% -110 @ 52.4% -200 @ 66.7% American odds
Read this carefully: a 65% win rate on -200 favorites means you're losing money. You needed 66.7% to break even. The tipster reporting "65% on premium picks" without telling you the average odds is reporting nothing.
Odds you got You need to win... Real-world context
+20033.3%Underdog. One in three is enough.
+12045.5%Small dog. Way below a coin flip.
-11052.4%Standard spread / total juice.
-15060.0%Moderate favorite.
-20066.7%Heavy favorite. Two out of three.
-30075.0%Huge favorite. Three out of four.

Same bettor, same skill level. Different odds. Wildly different breakeven requirements. This is the only chart that matters when you read someone's win rate.

The tipster's trick

Want a 65% win rate? Bet only -200 favorites. They win ~67% of the time naturally, because the price reflects the win probability. Now run a Twitter account, post the wins, and ignore the math.

High Win Rate / Losing Money
"Vegas Lock King"
Picks100
Avg odds-200
Win rate65%
Needed66.7%
-$250
Low Win Rate / Making Money
Lakeshore Edge (actual)
Picks310
Avg odds+105
Win rate49.4%
Needed48.8%
+$2,273

The lock king hit 65% and lost money. We hit 49.4% and made money. The difference isn't skill or luck. It's which prices we bought. He was paying premium for favorites everyone already knew would win. We're picking value where the line is mispriced.

Reality check Our 49.4% number is real and audited. Every settled pick is in the Performance journal on this site, with the model's probability, the book's odds, the closing price, and the outcome. You can sort by loss and read every mistake we've made. That's the only way to know a number is real.

The one number you actually want: CLV

Hit rate is a lagging indicator. You need 200+ resolved bets before it stabilizes, because outcomes are noisy. A model picking 55% can post 45% over 50 games purely from variance. By the time hit rate tells you the truth, you've already bet 200 games.

Closing-line value (CLV) stabilizes around 30 bets. It's the difference between the price you locked in and the price the market settled at by game time. If you got the Yankees at -150 and the line closed at -180, you "beat the close" — you bought at a better price than the consensus eventually agreed on.

Why does this matter so much? Because the closing line is the most accurate prediction the market produces. By game time, every sharp has weighed in, every late info movement is baked in, and the market has settled. If you can consistently buy before the line moves your way, you're picking up signal the market doesn't have yet.

Our CLV on the same 310 picks:

+4.5
Avg CLV (pts)
87%
Beat the close
109
CLV-graded picks
+$2,273
Est. P/L
87%
of our bets beat the closing line
In other words: 87 out of every 100 picks, our price was sharper than the market's at game time.

That 87% is the number that says we're not just lucky. Variance can push hit rate around. Variance can't fake CLV at this rate over 109 samples. To consistently beat closing lines, the model has to be finding information faster than the market.

How to read any sports-betting model in 5 seconds

Put any model into one of four boxes. The horizontal axis is hit rate relative to breakeven. The vertical axis is CLV.

Hit rate vs CLV: the only matrix you need
Hit rate < breakeven
Hit rate > breakeven
CLV > 0
SHARP (us) Beating the close, lower hit rate. Edge is real; results haven't caught up. Stick with it.
SHARP + LUCKY Beating the close AND winning more than breakeven. Rare; usually small samples that revert.
CLV < 0
BAD Losing the close AND losing bets. Avoid. This is most random tipsters once you check.
LUCKY Hitting picks but getting worse prices than the close. Will revert. This is most "hot streak" tipsters.

Most tipsters who advertise high win rates live in the LUCKY quadrant. The bigger the win-rate brag and the more silence about CLV, the more confident you can be they're there.

What to look at when evaluating any model or service

Open audit Lakeshore's /model page shows every one of these numbers, updated as picks settle. The Performance journal lists every pick we've ever made — wins, losses, stale, the works. We're betting against ourselves by being this transparent; we think it's the only way to earn the trust the rest of this industry spent 20 years burning.

FAQ

What win rate do I actually need to be profitable?

It depends entirely on the odds you're betting. At -110 you need 52.4%. At -200 you need 66.7%. At +120 you need 45.5%. The right question isn't "what win rate" but "what win rate relative to your average price". If you can't compute that from a tipster's reported numbers, the numbers aren't useful.

Can I have a low win rate and still make money?

Yes, easily. Picking +180 dogs that win 40% of the time has positive expected value. The mistake casual bettors make is treating win rate as the goal. Pros treat EV as the goal, and EV depends on both win probability and price.

Why is CLV more reliable than hit rate?

Hit rate is contaminated by variance. A 55% model can post 45% over 50 games just by bad luck. CLV is measured on every snapshot regardless of outcome, so it accumulates evidence faster. CLV stabilizes around n=30; hit rate doesn't stabilize until n=200.

Can I have positive CLV and still lose money?

Short-term, yes. Variance can hurt over 20 or 30 picks even when you're consistently beating the close. Long-term, no — positive CLV sustained over hundreds of picks is the strongest signal there is that an edge is real and will eventually convert to profit.

Why doesn't every site report CLV?

Two reasons. One: most don't measure it because it requires snapshotting closing prices on every game, which is operationally annoying. Two: when they do measure it, the number is usually bad, so it's quietly omitted from marketing. Most "experts" online are hitting at high rates on heavy favorites with negative CLV — the LUCKY quadrant — and selling subscriptions to the win rate.

How does Lakeshore measure CLV exactly?

Every pick is snapshotted at publish time with the FanDuel decimal odds and our model's win probability. We then snapshot the FanDuel line again within 15 minutes of game start. CLV = our model's probability minus FanDuel's closing implied probability. We also now track this against Pinnacle (the sharp book), which is the gold-standard benchmark professionals use. Read more on why we use Pinnacle.

See the audit yourself
Every pick, every loss, every closing line. The /model page is the proof; the Performance journal is the receipts.
Open the audit
Sports betting carries real financial risk. Past model performance does not guarantee future results. This article describes how to read performance metrics; it is not betting advice. Bet responsibly and only with money you can afford to lose. If gambling is causing harm, visit ncpgambling.org or call 1-800-GAMBLER.