I paid $30 to join a tipster's group. Here is what I found

Why I joined

I built predictapp so people would not need to pay tipsters. But before saying we were better, I had to study the competition. So I paid to join one tipster's "exclusive" group, to see what they really offer. I am not naming him. This is one experiment, not a verdict on every tipster, but the business model is the point.

How the sale works

First, credit where it is due: this man is a sales genius. He has almost 170,000 people in a free Telegram channel, where he shows expensive restaurants, penthouses and videos by the pool. "Win €11,666 in a couple of hours and eat a good steak!" "I am not saying this to show off." People assume he lives that way because he wins bets.

Then he gives away free picks, mostly at very low odds, so they hit often and the channel sees a long line of green. Next to them sits a bookmaker's referral code: sign up with it, deposit €35, get a free bet. Then comes the offer: 3 "safe" picks for that night, only 300 spots, $30 to get in.

The three "safe" picks

I paid, a little nervous. I cover every match in 30+ leagues, but what if I was reading things wrong? What if my product was not as good as I thought?

Real Madrid made sense, and predictapp was already saying so. Yellow cards, we do not model because they are too volatile, so fair enough. Then the third: Oviedo to win or draw against a good Alavés. I opened predictapp, and we clearly disagreed. predictapp gave Oviedo 19% and the draw 24%, 43% between them. His odds of 1.72 needed about 58% just to break even.

What happened

He lost 2 of the 3 "safe" picks. It was never his fault: there is always a reason he was right and it was bad luck. Then came the "recovery" picks: six more slips over the next two days, across NBA, tennis and soccer. Three lost and three won, and one Spurs handicap at 3.30 did most of the work.

Count every pick as 1 unit, the usual way to judge a tipster: 4 won, 5 lost, and he finished 0.3 units down. In euros his screenshots end slightly up, but only for someone who stayed to the very end: by the last night they were about €133,000 down, and the way back was another €100,000 on a single NBA game. Most people would have been out long before. Whether those bets were real or not, the business never depended on them. The paid group had about 300 members at $30 each, around $9,000 collected before a ball was kicked. Win or lose, he has been paid. The result is your problem.

The math most people never do

Low odds produce a beautiful green screenshot, not necessarily a winning account. It takes 60 seconds to check.

Make 7 bets at "safe" odds of 1.40 and win 5 of them. You won more than twice as often as you lost, and your profit is zero. At 1.40 a win pays 0.40 per unit, so it takes 3 wins to cover 1 loss. Win 2 and lose 1 with 10 on each, and you are down 2.

So if you are in a tipster's group, add up every win and every loss at the real odds, not the hit rate.

predictapp vs a tipster

• ›What you get: a tipster gives you a pick. predictapp gives you the probability of every outcome and the fair price on every market.
• ›Coverage: a tipster covers a few games a night. predictapp covers every match in its leagues, every day.
• ›Price: this tipster charged $30 for 3 picks. predictapp Soccer is €4.99 a month, and a free account gets one reveal a day.
• ›Record: tipsters show screenshots of the wins. predictapp's record is public, misses included.
• ›How you get paid: a tipster is paid when you join. predictapp's subscribers only stay if the numbers are useful to them.
• ›Who decides: a tipster tells you what to bet. With predictapp you learn to read a price and decide, including deciding not to bet.

Do not look for the truffle

The tipster hit some picks. Everyone does. That proves nothing. There is a saying I love: even a blind pig can find a truffle, but it helps to know they grow in oak forests. Look for the forest: the probabilities, the fair prices, the real chance of your parlay, where your process leaks. A man selling truffles from a penthouse is not going to show you the forest, because then you would not need him.

What users tell us

“Hola que tal, es una app muy buena mejor a comparación con otras que he usado y bastante certera.”

Hi there, it is a very good app, better compared to others I have used, and quite accurate.

“una buena apuesta es la que se hace con valor esperado, no es la ganadora; puedes hacer malas apuestas y ganar (por suerte) o puedes hacer buenas apuestas de valor y perder (por mala suerte).”

A good bet is one made well, not the one that wins. You can make bad bets and win on luck, or make good bets and lose on bad luck.