How to Spot Fake Picks
Most "picks" sold online are not backed by anything you can check. The playbook is always the same: post the winners, bury the losers, quote a giant win rate, and hope nobody asks for the receipts. This is a plain-English guide to the red flags, and to the one thing that separates a real record from a highlight reel: a complete, public, timestamped log of every pick, wins and losses included. We are not here to name anyone. We are here to teach you how to read the behavior.
Red flags of a fake capper
None of these is proof on its own, but they cluster. The more of them you see, the more you are looking at marketing dressed up as a track record.
1The losers disappear
Watch what happens after a bad night. A real record keeps the losses on the page forever. A fake one quietly deletes the losing posts, archives the bad days, or "starts fresh" every week. If a slip can vanish, the whole record is fiction, you are only seeing a curated highlight reel.
2Only green-slip screenshots
A cashed-bet screenshot proves exactly one thing: that one bet hit. It says nothing about the hundred bets around it. Anyone can place ten bets, screenshot the two winners, and delete the rest. A pile of green slips is the cheapest thing to fake in sports betting, treat it as marketing, not evidence.
3No public, timestamped record
The single most important question: can I see every pick, posted before the game started, with the date and time attached? If the only "record" is a DM, a pinned message that gets edited, or a number with no underlying log, there is nothing to audit. A pick logged after the result is known is worthless.
4"70%+ win rate" with no losing streaks
At a standard -110 price you only need to win about 52.4% of the time to break even. A real, sustained edge usually lives a few points above that line, not at 70% forever. A permanent 70% win rate with no documented cold stretches is a statistical fantasy: even genuinely sharp bettors hit losing weeks. Round, suspiciously high numbers that never dip are a tell, not a flex.
5ROI and win rate used interchangeably
These are different things, and the confusion is often deliberate. Win rate is the share of bets that win. ROI (return on investment) is profit divided by the total amount risked. You can have a 60% win rate and still lose money if you keep laying heavy favorites, or a 48% win rate and profit by betting plus-money underdogs. When someone blends the two into one big percentage, slow down, clarity is cheap, and vague math hides bad results.
6No sample size
A "hot" capper who is 9-1 has told you almost nothing. Ten bets is noise; a coin can land heads nine of ten times by luck. Edges only become visible over hundreds and thousands of graded outcomes. If there is no count, no "X picks graded", there is no way to know whether you are looking at skill or a small, lucky run.
The math behind the "70%" claim
Here is the number worth memorizing. At a standard -110 price , the typical line on a point spread or a two-way prop, you need to win about 52.4% of your bets just to break even, because the book takes its cut (the "vig") on every wager. So a real edge does not look like 70%. It looks like grinding out a few points above that break-even line, over a large sample, with losing stretches still on the page.
That reframes the sales pitch. A permanent, advertised 70%+ win rate with no documented cold runs is not a sign of a sharper handicapper, it is a sign the record has been curated. Variance is real. Even a genuine, long-run edge produces losing weeks. A record that never shows one has had the losses removed.
We are honest about this with our own numbers, too. We studied our older graded history and published the real results, including the cuts that did not beat the break-even line. That is what "in-sample, not a promise" looks like in practice.
How to verify before you pay
You do not need to trust anyone. You need to be able to check. Before money changes hands, demand all four of these, and walk if you cannot get them:
- The full graded record, wins AND losses. Not a win count. The complete log, with every losing pick still on it. If the losses are not there, the record is incomplete by design.
- Dates and timestamps, posted before each game. Every pick should be provable to have existed before the result was known. A pick "logged" after the fact proves nothing.
- A real sample size you can count. Hundreds or thousands of graded picks, not a dozen lucky ones. Ask for the number, and make sure it is large enough to mean something.
- Grading tied to official results you can audit. You should be able to take any pick, look up the box score yourself, and confirm the grade. Transparency that you cannot independently verify is just a nicer-looking screenshot.
What an honest record looks like
Turtle separates newer Verified Results, tracked from posting through settlement, from older Historical Results. The older count is 138,000+ model and tool rows, not unique customer recommendations. Some older rows have incomplete posting, price, or model-version detail, and official corrections may update a result. Performance states the relevant sample and method.
We frame those numbers the way you should expect anyone honest to: as a historical, in-sample record, not a promise of future profit. That distinction is the whole point, verifiability over hype. See how our models and grading work, or read the record yourself below.
Frequently Asked Questions
Is a 70% win rate possible in sports betting?
Over a small sample, sure, anyone can run hot for a week or two. As a sustained, long-run number it is extremely implausible. At a standard -110 price you only need about 52.4% to break even, and a genuine edge typically sits a few points above that, not at 70% with no cold stretches. When someone advertises a permanent 70%+ win rate and never shows a losing run, treat the number as a marketing claim, not a measured result.
What is the difference between win rate and ROI?
Win rate is the percentage of your bets that win. ROI (return on investment) is your profit divided by the total amount you risked. They answer different questions, and one can look great while the other is negative. You can win 60% of your bets and still lose money laying heavy favorites, or win under 50% and profit on plus-money underdogs. Anyone who blends the two into a single impressive-sounding percentage is making it harder for you to check their math, which is the point.
How do I verify a pick seller before I pay?
Ask for the full graded record, every pick, wins and losses, with the date and timestamp each was posted before the game. Look for a real sample size (hundreds or thousands of graded picks, not a dozen), and confirm the losing days are still visible. If the only "proof" is screenshots of winners, a number with no underlying log, or a record you cannot audit yourself, you have not been shown anything verifiable. No verifiable record, no payment.
Why do screenshots of winning bets not prove anything?
A screenshot of a cashed bet only proves that one specific bet won. It tells you nothing about how many bets surrounded it or how many lost. It is trivial to place many bets, screenshot the few that hit, and delete the rest. A wall of green slips is among the easiest things to fake in this space, it is a marketing artifact, not evidence of an edge.
What does a trustworthy track record actually look like?
A trustworthy record states what was counted, when it was posted, the price used, the dates, and the sample. Turtle separates newer Verified Results from more than 100,000 older Historical model and tool rows. Older rows remain useful, but publication and model-version detail is incomplete for part of that history.
Stop trusting screenshots. Start checking the record.
Audit our full graded history, try the free calculators, or start a $1 first week and review current modeled +EV picks across dozens of sportsbooks. Newer Verified Results and older Historical Results are labeled separately.
21+. Always bet responsibly. Past results don’t guarantee future performance.