Skip to content

← Money

Betting

Chapter 31 · Traps

In this chapter 6 sections
  1. The house’s bet
  2. A thousand bets
  3. What the law says here
  4. What you hear, and what nearly happens
  5. When it stops being a game
  6. Try this

The message arrives in Sabina’s class group on the evening of a big cricket match. Someone has forwarded a link with a picture of a trophy and a line in capitals: predict tonight’s winner, double your money. Under it a boy from another section has posted a screenshot of his account, Rs. 1,900 where there had been Rs. 1,000. Three people have answered with fire emojis. Sabina, who has read Chapter 29 and is suspicious of anything this easy, does what that chapter taught her. She asks who pays.

The answer this time is not a trick, or not only one. A betting site really can pay a winner, and the boy in the screenshot may really have won. What the screenshot can’t show is the number that decides everything, the one the site sets before anyone bets: how much it pays for a win compared with what the win is really worth. This chapter is about that number, why it means the house wins, how Nepal’s law treats the whole business, and what happens in the heads of the people who keep playing anyway. Sabina is made up, as in Chapter 27; the arithmetic, the laws and the research are real.

The house’s bet

Take a match between two sides so even that each really has the same chance of winning, one in two. A fair bet on it would pay you Rs. 100 for every Rs. 100 you risk: win, and you get your Rs. 100 back with Rs. 100 more; lose, and you lose your Rs. 100. Over many such bets you would come out about level, as with tossing a coin with a friend.

A bookmaker doesn’t offer that. It offers odds of 1.90 on each side. Odds of 1.90 mean that a winning Rs. 100 comes back as Rs. 190: your own Rs. 100 and Rs. 90 more. A loss still costs the whole Rs. 100. The difference looks small, and it is meant to. You are risking Rs. 100 to win Rs. 90 on a toss that is fair.

Now do the sum Chapter 8 did for the mystery box. In two bets like this, on average, you win one and lose one: +Rs. 90 and −Rs. 100, a loss of Rs. 10 over the two, or Rs. 5 a bet. That average, the expected value, is −Rs. 5 on every Rs. 100, and the 5% is called the house’s edge. Nobody loses exactly Rs. 5 on any one bet, of course. Each bet either wins Rs. 90 or loses Rs. 100. The Rs. 5 is what the bets come to when there are enough of them.

Chapter 21 told apart two kinds of bet: a fair one, where the average is zero and you are only paying for excitement, and a good one, where the average is on your side. A bet against a bookmaker is a third kind. Its average is set, in advance, against you.

Look at it from the bookmaker’s side, and something stranger shows. Suppose a thousand people put Rs. 100 on one side and a thousand on the other. The bookmaker takes in Rs. 2 lakh. Whichever side wins, it pays a thousand winners Rs. 190 each, Rs. 1.9 lakh, and keeps Rs. 10,000. It doesn’t need to know who will win. It doesn’t even need to care. The 5% is in the price, the way a shop’s margin is in its price, and the money comes from the people who lost. Real books are never balanced as neatly as that, and real matches are rarely even, so a bookmaker shades the odds on each side to keep its edge whatever people back. But the principle is the same, and so is the result: what it pays out is always a little less than what the chances are worth.

Every form of gambling that runs as a business works this way, only the size of the edge changes. A roulette wheel’s edge comes from its green zero, as the Going Further topic on fair games works out, and a lottery’s from paying out in prizes much less than it takes in: Britain’s National Lottery, for one, pays only part of each pound back as prizes, and the rest goes to good causes, the government and running costs. If you want to see how a game’s average is worked out when the prizes are many and the chances tiny, the Going Further topic on a game’s average does it for one of the world’s biggest lotteries.

A thousand bets

An edge of 5% sounds like it shouldn’t matter much. Over a few bets, it hardly does. That is exactly why it works.

Below are twenty made-up players. Each makes the bet above, Rs. 100 at odds of 1.90 on a true even chance, a thousand times, one bet after another. Their wins and losses are drawn at random, by a computer, but the same draws every time you open the page, so that the numbers here and in the drawing agree. Start with ten bets each, then a hundred, then a thousand.

Try it

Bet a thousand times against a 5% edge

Each line is a made-up player betting Rs. 100 at a time at odds of 1.90 on an even chance. Green lines end ahead.

−10,000−8,000−6,000−4,000−2,0000+2,000+4,000Rs.average: −Rs. 5 a bet05001,000bets made
Bets each
Ahead
2 of 20
Average player
−Rs. 3,993
The house kept
Rs. 79,860

After 1,000 bets each, 2 of the twenty are ahead. Together they have lost Rs. 79,860 of the Rs. 20,00,000 they bet, about 4%.

After ten bets, seven of the twenty are ahead. If you were one of those seven, you would be telling your friends you had a feel for it. After a hundred, six are ahead, some of them by more than Rs. 1,000. Luck, over a hundred bets, still swamps an edge of Rs. 5 a bet, and that is why a hundred bets can feel like proof of a system.

After a thousand, two of the twenty are ahead. The rest have spread out below zero, and they are not spread around nothing: they are spread around the dashed line, which has fallen steadily by Rs. 5 a bet to Rs. 5,000 below where they began. The average player is about Rs. 4,000 down. Together the twenty have bet Rs. 20 lakh, and the house has kept Rs. 79,860 of it, about 4%. Twenty players are still few: the edge says Rs. 1 lakh, and this twenty happened to be a little luckier than that.

Two things stand out. The first is that not everyone loses. Two players out of twenty finished ahead, and in another run of the dice it might have been one, or three. The exact chance of being ahead after a thousand of these bets can be worked out, and it is about one in twenty-one. Someone will be that one. The second is that the house doesn’t care which one. It doesn’t bet against any player; it bets on all of them at once, and across all of them the edge comes through almost exactly. A player who makes ten thousand bets has about one chance in fifteen million of being ahead. For the house, which sees millions of bets, the edge isn’t a chance at all. It is income.

You can see the same thing in a die game in Going Further’s fair-games topic, which follows one player at a stall, and the reason behind it, the law of large numbers, in Chapter 15 of Thinking Clearly. What the twenty lines add is the view from the counter: each player has a story, and the house has a sum.

There is no betting system that changes this. Doubling your bet after each loss, waiting for a side that is “due”, betting only on favorites: each changes how the losses arrive, in many small ones or a few big ones, but not their average, because every bet, big or small, still pays a little less than its chance is worth. Over a thousand Rs. 100 bets you expect about 500 wins and 500 losses: +Rs. 45,000 and −Rs. 50,000, a loss of Rs. 5,000, the same Rs. 5 a bet as before. However the bets are arranged, every Rs. 100 put through loses Rs. 5 on average, and a system that makes you bet more only puts more through.

What the law says here

In Nepal most of this is simply against the law. The Muluki Criminal Code, 2074, section 125, says that no one shall gamble or run gambling. It defines gambling broadly: any game or process in which money or property is staked on chance, and it says in so many words that this includes betting on someone else’s winning or losing. Two things are excepted. Small games of chance the government allows for entertainment at festivals and fairs are not gambling, and neither is a lottery approved by an official with the power to approve it. Separately, the same section bans सट्टाबाजी (sattabaji), betting on whether a side will win or lose in a game that is itself lawfully played, such as a cricket match.

The punishments climb. The first time, gambling can bring up to three months in prison or a fine of up to Rs. 30,000; the second time, up to a year and up to Rs. 50,000, and each time after that more again. The money staked and the tools of the game are seized. Betting on a lawful game can bring up to a year and Rs. 10,000, and the money is seized too. The Children’s Act, 2075, goes further for anyone under eighteen: involving a child in gambling or cards, or letting a child into a casino, is an offense against the child, by the adult who does it.

The Code says nothing about apps, and nothing about the word “prediction”. What it says is broad. Staking money on which side will win a lawful match is सट्टाबाजी in its own words, and staking money on something that comes by chance, the color of the next spin or the number that turns up, is gambling by its definition. Neither definition depends on what a game calls itself. Whether a particular app falls under them is for a court to decide, but the words make no exception for “prediction”, “quiz” or “earn”.

Casinos are the one kind of gambling house the state licenses, and only for visitors. They are licensed under the Tourism Act, 2035, and in 2025 the government issued new Casino Rules under it. As Nepal’s newspapers reported them in July 2025, those rules forbid Nepali citizens from entering casinos or playing in them, and in November 2025 the Department of Tourism sent the casinos a twelve-point directive to keep to the rules. The same papers have reported, more than once, that Nepalis get in anyway. A casino’s edge is the house’s, wherever the player comes from.

Online betting has no law of its own. The Electronic Transactions Act, 2063, which covers crimes done through computers, doesn’t name gambling or betting at all. In the cases against betting sites we read about, the money was the way in. In 2023, newspapers reported, a group accused of promoting an online betting site to Nepalis, with hundreds of millions of rupees said to have passed through it, was charged under the Nepal Rastra Bank Act and the Advertisement (Regulation) Act. In January 2025, newspapers reported, the Kathmandu Valley Crime Investigation Office arrested two people it said were running an online betting site, and seized dozens of SIM cards and ATM cards, chequebooks and phones. A betting site needs bank accounts and phone wallets to take money in, and those accounts are where it can be found.

The law matters for a plainer reason too. An app that promises to double your money is asking you to do something the law forbids, and a business that already breaks the law has little reason to pay you what it owes.

What you hear, and what nearly happens

Go back to the screenshot in Sabina’s group. It was a real win. But think about who posts screenshots. The boy who turned Rs. 1,000 into Rs. 1,900 posted his. Whoever lost that evening didn’t post theirs, and nor would the eighteen of the twenty in the drawing above who were behind after a thousand bets. Wins are told, losses are kept quiet, and so the stories you hear are a sample chosen by luck. It is the same mistake Chapter 10 of Thinking Clearly is about, and the same one Chapter 24 found among people boasting about shares.

Like the watch in the mystery box, the big win sells the game. A screenshot of one is the best advertisement a betting site could ask for, and it costs the site nothing, because the win can only be as common as the edge allows.

There is one more trick in many games of chance, and it works on the brain rather than on the arithmetic. It is the near miss.

In 2009 a team of researchers at the University of Cambridge, led by Luke Clark, built a simple slot machine on a computer. It had two reels, each with six pictures. When the two reels stopped on the same picture, the player won 50 pence. Sometimes the second reel stopped one place away from a win, just above or just below the line, and the researchers called those near misses. Forty students who did not gamble regularly played it, rating as they went how pleased they felt and how much they wanted to keep playing.

A near miss is a loss, exactly like any other. The players knew this, and they rated near misses as less pleasant than other losses. But the same near misses made them want to play on more. The pull to play on appeared only when the player had chosen the picture to aim for, as if, having nearly done it, they could do it next time. Fifteen more volunteers played it in a brain scanner, and in them near misses lit up parts of the brain that respond to winning. Something in the players was counting the loss as nearly a win.

A machine or an app can make near misses as often as it likes, since it decides what to show. A scratch card can be printed with two of the three symbols you need; a spinning wheel on a screen can slow down just past the prize. The players were not regular gamblers, which is the unsettling part: the pull of a near miss doesn’t come from a weak character. It comes from the way brains learn from what nearly happened, and a game can be built to use it.

When it stops being a game

Betting can stop being a small thing done now and then, and the change is hard to see from inside. Britain’s National Health Service gives a short set of questions to ask yourself, and they are worth knowing even if you never bet. Do you bet more than you can afford to lose? Do you need to bet more and more to feel the same excitement? Do you go back to try to win back what you lost? Have you borrowed money, or sold something, to bet? Have you ever wondered whether you have a problem? Has it caused you stress or anxiety, or money problems at home? Do other people criticize your betting, and do you feel guilty about it? The more of these that are true, and the more often, the more likely it is to be doing harm.

The one in the middle, going back to win back what you lost, is called chasing losses, and it is where the drawing above becomes dangerous. A player who has fallen behind needs luck to get back, and the only way to try is more bets, each with the same Rs. 5 edge against it. Chasing feels like fixing the problem. In the arithmetic, it is making it bigger.

If this sounds like you, or a friend, the most useful thing is also the hardest: tell someone. A parent, an older brother or sister, a teacher you trust. The National Health Service’s advice is plain: talk to someone, don’t think of betting as a way to make money, and ask the bank whether it can block payments to betting sites. Beyond that, hand the card or phone wallet to someone you trust for a while, delete the apps and leave the groups that share links. The Your Mind, in Practice tips list general helplines in Nepal for when something feels too heavy to carry alone; none is for betting alone, but they are there for anyone who needs to talk.

There is nothing to be ashamed of in having been pulled in. The games are designed by people whose business is to keep a player playing: the posted wins, the near miss, the small edge that hides itself for a hundred bets. Being caught by something built to catch you is not a failing. Staying quiet about it is the only part that makes it worse.

So the plain thing to take from this chapter is that a betting business pays a little less than the chances are worth, on every bet, and that little is what it lives on. Over a few bets luck hides it, and some people win; over many, it comes through, and the house, which takes everyone’s bets, collects it almost exactly. In Nepal, staking money on chance is a crime under section 125 of the Criminal Code, whatever an app calls it, apart from approved lotteries and small games at fairs. The wins you hear of are a chosen few, near misses are built to feel like nearly winning, and when betting stops being fun, the thing to do is to tell someone.

Try this

  • Next time you see odds on a match, two numbers like 1.85 and 1.95, work out what they say each side’s chance is: divide 1 by each. Add the two answers. If the match is fairly priced they would add up to exactly 1; how much more than 1 they add up to is roughly the bookmaker’s edge. (For 1.90 and 1.90: 0.526 + 0.526 = 1.053, an edge of about 5%.)
  • On paper, follow one player through ten bets by tossing a coin: heads +90, tails −100. How many heads out of ten do you need to finish ahead? (Six: five heads and five tails leave you Rs. 50 down.) Toss the ten and see where you end.
  • Look through the apps on your phone and the groups you are in. Delete any app that pays for “predictions”, and leave any group whose links keep sending you to one.
  • If a friend shows you a win, ask them, kindly, what they have lost in all since they started. If they don’t know, that is worth knowing too.

Found a mistake? Please tell us.