Free Picks: Who Pays for Them and What 20 Picks Prove
A free picks feed costs you nothing to read. Someone pays for every pick anyway. Find the payer first. Then run a test with numbers, because the record on the page does not settle the question.
Here is the number to hold onto. A feed with zero skill, picking at -110 and winning half its picks, shows a profit after 20 picks 41% of the time. A feed with a real 55% win rate, worth a 5% return on every dollar risked, shows a loss after 20 picks 41% of the time. Both outcomes are common. A week of free picks sorts neither feed.
This guide covers three things. The guide names who pays for the free feed. The guide shows what a short record proves, with exact probabilities. The guide ends with a one-week test built on a different question from the one most bettors ask. The record math for longer samples sits in our tipster record guide, and this piece does not repeat the math.
Who Pays for the Free Pick
A pick has a cost. The writer spends time, the site spends hosting money, and someone covers both. Three payers fund free feeds. Each one wants something different from you.
Payer One: The Sportsbook
Affiliates earn money when you open and fund an account. Better Collective, a large publicly listed sports betting affiliate, reports two pay models in its filings. A fixed fee per new depositor is called CPA. A share of the player's net losses is called revenue share. In its Q2 2025 report, Better Collective said CPA revenue fell 31%, which Better Collective tied to lower partner activity in the US. The same report put new depositing customers at about 300,000 for the quarter, with 86% coming from revenue share agreements.
Read the second figure twice. On revenue share, the affiliate keeps earning while you keep betting and keep losing. A feed paid this way earns more when your account loses. The feed does not need you to win. The feed needs you to stay active.
What a revenue share link earns when you lose
The 30% default is an assumption. Programs keep their rate cards private. Set your own number.
Payer Two: Your Own Subscription
The free pick also works as an ad for the paid pick. In 2015, Covers.com launched an affiliate program through Income Access, reported by SBC News and Gaming Intelligence. Affiliates earned 20% of subscription and individual pick sales, 100% of a referred customer's first sale, and a $1 bounty on a free account sign-up. The terms are old and the program has since changed. Look at the structure instead of the figures. Every step of the funnel carries a price: the free sign-up, the first pick purchase, and each renewal.
Payer Three: The Victim
Some feeds exist to sell fiction. In December 2024, Cory Zeidman pleaded guilty in the Eastern District of New York to conspiracy to commit mail and wire fraud. Prosecutors said his group advertised nationally, used fake names, and pressed prospective bettors to pay large fees for non-public information on injuries, referees, and fixed games. The Justice Department said the information was fictitious or came from open-source internet research. Zeidman faces up to 20 years in prison and approximately $3.6 million in forfeiture and restitution.
Most feeds are not fraud. The case shows the ceiling on the harm, and the method. A confident voice and a stream of picks look the same whether research or invention produced them.
The Conflict Has a Name
Roman Inderst and Marco Ottaviani modeled this setup in the American Economic Review in 2009. A seller of advice also hunts for customers. The firm sets pay by weighing the expected cost of misselling against the cost of giving marketing incentives, and the transparency of commissions changes how much misselling occurs. Apply the model to a pick feed. The writer you read is also the salesperson. Pay tied to sign-ups pushes toward volume and confidence, not toward your results.
Regulators treat the money link as material. The FTC's revised Endorsement Guides, 16 CFR Part 255, treat an affiliate commission as a material connection, and law firm summaries of the 2023 update say the connection needs a clear and conspicuous disclosure. Massachusetts goes at the book instead of the writer. Its sports wagering advertising rule, 205 CMR 256, holds the operator answerable for advertising done on its behalf by affiliates and other third parties under contract. Check any feed you follow for a disclosure next to the pick, not in a footer.
What 20 Picks Prove
Take a feed posting picks at -110. Each win pays $0.909 per $1 risked. Each loss costs $1. Break-even sits at 52.38% wins, because 110 divided by 210 equals 0.5238. At 20 picks, a bettor needs 11 wins to show a profit. Exact binomial probabilities for the chance a feed shows a profit after a given number of picks:
| Picks posted | True win rate 50% | True win rate 52.38% | True win rate 55% | True win rate 58% |
|---|---|---|---|---|
| 7 (one week) | 50.0% | 55.2% | 60.8% | 67.1% |
| 20 | 41.2% | 49.7% | 59.1% | 69.4% |
| 50 | 33.6% | 46.6% | 61.3% | 76.4% |
| 100 | 30.9% | 49.1% | 69.3% | 86.7% |
| 300 | 19.3% | 48.4% | 80.8% | 97.3% |
Read the 20-pick row. The 50% feed shows a profit 41.2% of the time. The 55% feed shows no profit 40.9% of the time. A feed at 52.38% wins, the break-even rate, shows a profit roughly half the time at every length. At 7 picks, a coin-flip feed shows a profit half the time.
Proving a 55% win rate takes volume. At -110 the 55% bettor earns 0.05 per unit risked, and a single bet has a standard deviation near 0.95 units. A one-sided 95% test needs n = (1.645 × 0.95 / 0.05)², which is about 980 picks. A feed posting 7 a week needs nearly three years.
Odds a feed shows a profit by luck or by skill
What the Research Says About Tipsters
The academic record agrees with the table. Martin Spann and Bernd Skiera compared tipsters, betting odds, and prediction markets on three seasons of German Bundesliga matches, between 678 and 837 games. Prediction markets and betting odds scored about equally well and clearly beat the tipsters. None of the forecasts produced systematic profit once the state bookmaker took its 25% fee.
Egon Franck, Erwin Verbeek, and Stephan Nüesch tested the market itself on 5,478 matches from Europe's five biggest leagues over three seasons. The betting exchange predicted more accurately than the bookmakers. Prices carry the information. A tipster must beat a price, not a coin.
Crowds of tipsters tell a different story, and the difference matters. Alasdair Brown and J. James Reade studied Oddsportal, where tipsters rank by betting return and no prize rewards accuracy. A strategy of backing the outcome a majority predicted earned an average return of 1.317% across 68,339 events. Limiting the crowd to experienced or accurate tipsters did not improve returns. Gruettner, Wambsganss, and Back analyzed 1,534,041 tips from 3,484 Blogabet tipsters. The group won 3.29% more tips than bookmaker-implied probabilities predicted, with an average yield of 3.97%.
Both studies measure the crowd. Neither one picks out a single feed you should follow. Skill, if present, shows up in aggregates and across thousands of events. One account with 20 posted picks gives you none of the proof.
The One-Week Test
You cannot test skill in a week. You test whether the feed is worth tracking for a year. Use paper bets, with no money at risk. Do not open an account through the feed's link while you test.
- Log every pick in a sheet at the moment of posting. Record the time, the pick, the posted price, and the stake. Take a screenshot. Check for edited or deleted posts at the end of the week.
- Record the price you get at your best book. Shop two or three books at posting time and write down the best available. A pick posted at +110 when the market has moved to -105 is not a pick you bet.
- Record the closing price, the last price before the game starts. A sharp book's no-vig close is the benchmark, as our closing line value guide explains.
- Count the picks where your logged price beat the close.
- Check disclosure. Find the affiliate language next to the picks and the sign-up button. No disclosure ends the test.
- Check losses. A feed must show its losing picks, with the same detail as its winners.
Why the closing price? A win or loss settles the bet and says little about the price. Beating the close settles before the game. If the feed has no information beyond the market, its picks beat the close about half the time. Under this assumption, 15 or more beats out of 20 happens by chance 2.1% of the time. Fourteen or more happens 5.8% of the time. Twelve or more happens 25.2% of the time.
| Picks beating the close, out of 20 (this many or more) | Chance by luck alone (50% each) | Verdict |
|---|---|---|
| 12 | 25.2% | Noise. Drop the feed. |
| 13 | 13.2% | Noise. Drop the feed. |
| 14 | 5.8% | Borderline. Track one more week. |
| 15 | 2.1% | Keep tracking on paper. |
| 16 or more | 0.6% | Keep tracking on paper. |
Two limits apply. First, a feed passing the test still proves little. The pass earns a longer paper trial, not your money. Second, testing many feeds finds winners by chance. Run ten feeds through a 15-of-20 threshold and you expect about 0.2 false passes, so a single pass among ten feeds is thin evidence. Raise the bar as the number of feeds you test grows.
Free Picks Rules
- Treat every free feed as an ad until the disclosure and the price log say otherwise.
- Never fund an account through a feed's link because of a record under 300 graded picks.
- Read a revenue share link as a cost. Whoever holds the link earns on your net loss.
- Shop the price yourself at DraftKings, FanDuel, and a third book. The posted price rarely stays available.
- Judge the feed on closing line value first and on win-loss second.
- Pay for picks only after a long paper record on timestamped, unedited posts. Even then, price the subscription against the edge you measured.
The verdict: a free feed is a sales channel. A coin-flip feed shows a profit after 20 picks 41% of the time, and a feed with a 5% edge shows a loss 41% of the time, so a week of results tells you nothing about skill. Test the process instead. Log the pick time and price, compare to the close, check the disclosure, and trial the feed on paper. Fifteen closing-line beats out of 20 earns another week of tracking. Anything less earns an unsubscribe.
Apply the same test to us. DawgHousePicks earns referral pay from the partners on our free month page: ProphetX, Rebet, BetR, Outlier, Chalkboard, Pikkit, Polymarket, Optimal Bet and Pickfinder. Our DraftKings, FanDuel, Caesars, BetMGM, Kalshi and PrizePicks links are refer-a-friend links, which can also pay us a referral bonus. The conflict in this article applies to us.
Sources and Method
Better Collective's Q2 2025 interim report supplies the CPA, revenue share, and new depositing customer figures. SBC News and Gaming Intelligence supply the 2015 Covers affiliate terms. The Justice Department's Eastern District of New York release supplies the Zeidman plea. Inderst and Ottaviani supply the misselling model. Manatt and InfoLaw Group summarize the FTC Endorsement Guides, and the Massachusetts Gaming Commission supplies 205 CMR 256. Spann and Skiera, Franck and coauthors, Brown and Reade, and Gruettner and coauthors supply the tipster and market studies.
Every probability above comes from the exact binomial formula: the chance of k or more wins in n picks at win rate p, where a pick wins 100 / 110 units at -110 and loses 1 unit. The calculator recomputes each cell. The 980-pick figure uses a win payout of 0.909, a win rate of 55%, a return of 0.05 per unit, a per-bet standard deviation of 0.95, and z = 1.645. The closing-line table assumes each pick beats the close with probability one half under no information, ignores ties, and uses the same binomial formula.
Limits: the revenue share rate is an assumption you set, because programs do not publish rate cards. The Covers terms date from 2015 and have changed. The closing-line test fits markets with a clean closing price and does not prove skill. Studies of tipsters use European soccer and online communities, and the results for any single US feed need their own data.