The last few lessons were about luck — slots, packs, random reveals. This one is different. You actually know some of this stuff. You can read a game. You'd beat your friends in a pick'em.
So here's a fair question: if you're genuinely good, you should be able to win money picking winners… right? Let's start with one pick.
Twelve matchups. Each one, pick who you think wins. Every pick is a real bet at standard odds: risk 11 credits to win 10. Use what you know — the records are right there. Watch your bankroll.
You risk 11 to win 10. So a win nets you 10, but a loss costs you 11. Win exactly half your bets and you don't break even — you're down, because every loss costs more than every win pays. To actually break even you have to win 52.4% of the time. That gap is the vig (the cut, the juice) — the operator's margin, baked into the odds. It's about 4.5% of every dollar you risk, win or lose.
A great night isn't a season. Let's run 500 bets at three honest skill levels and watch the bankroll. Same starting stack of 600 credits each.
When scientists tested experienced sports bettors, the experts did pick winners more accurately than random guessing — the skill was real. But the money they walked away with was no better than picking at random, because the margin ate the edge (Cantinotti et al., 2004). The wins they remembered, plus the near-misses, kept the feeling of being an expert alive — and the ads lean on exactly that feeling, selling "do your research, you've got an edge" (Lopez-Gonzalez et al., 2017).
An activity can be genuinely skill-influenced and still be a money-loser, because the operator sets a margin — the vig — that sits above the edge your skill can produce. Skill might lift you from a coin flip to, say, 53% right. Real! But the house only needs you under about 52.4% to keep printing money, and almost nobody sustains more than that. Variance — the hot streak, the great night — is what makes a losing game feel like a skill you're mastering: you remember the wins, the ads tell you it's your read, and the margin quietly collects over the long run.
The question is never "are you good?" It's "are you good enough to beat the margin?" — and the margin is set so that, for almost everyone, the answer is no.
Think of something you're sure you're good enough at to come out ahead — a game, a bet, a pick, a card flip with friends. Be honest with yourself: are you beating the other players, or beating a house that takes a cut every round? And how would you even know the difference?
Cantinotti, M., Ladouceur, R., & Jacques, C. (2004). Sports betting: Can gamblers beat randomness? Psychology of Addictive Behaviors, 18(2), 143–147. pubmed.ncbi.nlm.nih.gov
Lopez-Gonzalez, H., Estévez, A., & Griffiths, M. D. (2017). Marketing and advertising online sports betting: A problem gambling perspective. Journal of Sport and Social Issues, 41(3), 256–272. journals.sagepub.com
Cantinotti, M., Ladouceur, R., & Jacques, C. (2004). Sports betting: Can gamblers beat randomness? Psychology of Addictive Behaviors, 18(2), 143–147. (Experienced hockey bettors selected winners more accurately than random selection, but their monetary outcomes were not significantly better than random; the information used and near-misses reinforced a perception of expertise.)
Lopez-Gonzalez, H., Estévez, A., & Griffiths, M. D. (2017). Marketing and advertising online sports betting: A problem gambling perspective. Journal of Sport and Social Issues, 41(3), 256–272. (Identifies "skill-enhancing narratives" in betting ads that over-emphasize the bettor's knowledge and sense of control.)
The break-even math (−110 / "standard juice"). A bet that risks 110 to win 100 breaks even at win probability p where p·100 = (1−p)·110, i.e. p = 110/210 = 52.38%. Expected value per bet at win rate w is w·100 − (1−w)·110: at 50% that is −\$5 per \$110 risked (≈ −4.5%), at 52.38% it is \$0, at 54% it is +\$3.40. The lesson uses an 11-to-win-10 version of the same odds. The hold on a two-sided −110 market is about 4.5% of money wagered.
Honest scope. A sustained 54–55% win rate genuinely profits, so the lesson does NOT claim "skill always loses money." It claims (a) the break-even bar is above a coin flip, (b) real edges are small and rarely sustained, and (c) a winning night (variance) is routinely mistaken for a winning skill — precisely Cantinotti's "more accurate, not more profitable" result. Poker among friends with no rake is included as a genuine-edge control so "skill" is not taught as always a trap.
Note for the build team: the load-bearing claims — that the margin lifts break-even above 50%, and that skilled bettors picked more accurately yet won no more money — are well supported. Keep the framing at "skill must beat the margin, and almost no one sustains it," never "skill always loses." Mechanism, never a personal verdict about the student.