Two losing games that win

Game A loses money, slowly. Game B loses money, slowly. Play two rounds of A, then two of B, and repeat, and you win. Or toss a coin each round to decide which game to play, and you win. Both of those are exact results, not luck.

The rules, all of them. Every round you win or lose one counter. Game A is a coin that wins 49.5% of the time. Game B has two coins: if your counters are a multiple of three you must toss a bad one (wins 9.5%), otherwise a good one (wins 74.5%). Both games lose in the long run. A plan is the order you play them in.

Game B loses because playing it keeps bringing your counters back to multiples of three: the bad coin comes up 38% of the time, not a third. Mixing in rounds of A scrambles your position so that when you return to B you are more often on the good coin. The dashed lines are exact long-run trends (the pattern of counters-mod-three is followed until it settles) and the experiments check each against twenty million simulated rounds. One correction to a common telling: strictly alternating A, B, A, B loses with these coins. Two of each wins, and so does choosing at random. Juan Parrondo devised the games in 1996. It is not a way to beat a casino: no real pair of games is linked through your winnings like this.