Every time you buy Yes, someone sells it to you. The price you click is another person’s opposite opinion, wearing a number. So the question that decides whether you make money isn’t “am I right about the event?” It’s “why is the one on the other side of this trade wrong?”
Most beginners never ask it. They should, because the answer has changed. Increasingly, the one on the other side isn’t a person at all.
There is no house to beat
A casino profits when you lose, and prices its games so you do. A prediction market works differently, as the comparison with betting covers: the platform takes a fee on activity and doesn’t care which side wins. Every dollar you make comes out of another trader’s position, and every dollar you lose lands in one.
That sounds fairer than a casino, and it is. It also means something beginners skip past: the game is you against whoever took the other side. Beating the house is impossible by design. Beating a counterparty is possible, but you have to be better than that specific counterparty, and it helps to know who they are.
Who’s actually on the other side
Three kinds of traders, roughly.
People like you, with opinions and news feeds. On small markets, mostly them.
Market makers, the ones from the liquidity note, quoting both sides and earning the spread. Trade against them at a fair price and nobody’s the sucker; you paid a small fee for the ability to trade at all.
Bots. Software that watches feeds, scores, and price moves, and reacts in milliseconds. Their share of volume keeps growing, and on some kinds of markets they are, in practice, the whole other side.
A bot doesn’t sleep, doesn’t tilt after a loss, doesn’t fall in love with a position, and reads a goal notification faster than your phone can vibrate. One widely shared post this year claimed $570K in profits from a bot doing latency arbitrage on live sports, essentially buying the outcome a fraction of a second after reality settled it but before slower traders noticed. The account posting it looked like bait, so treat the number as advertising. The mechanic it described is real, and it has a blunt consequence: on any market where the answer arrives on a clock, the fastest reader wins, and the fastest reader is not you.
The trap dressed as an opportunity
Here’s how this bites a beginner in practice.
A match is live. Something just happened, and the price on your screen looks stale, like the market hasn’t caught up with what you just saw on TV. It feels like free money. Slow down: your TV feed runs seconds behind the venue, and the bots don’t watch TV. If a price looks slow after news, the likeliest explanation is that it already moved and settled, and what you’re seeing is the fresh, correct price that only looks stale to you. The trade that feels like picking up a dropped wallet is usually you becoming the other side of a bot’s exit.
A useful reflex: whenever a trade looks obviously, urgently good, ask what you know that the machines reading the same feed don’t. On a live market, the honest answer is almost always nothing.
The horizon is the whole game
A quant I traded replies with on X put it in a way that stuck: on a five-minute market, there is nothing to forecast.
That’s the frame that sorts all of this. Markets differ by how much time sits between the trade and the answer, and that horizon decides which skill gets paid.
Fast markets (live sports, five-minute crypto price bets) leave no room for analysis, because there’s nothing to analyze. Whoever reacts first wins, reaction speed is the entire edge, and machines own it. A human in that lane isn’t competing, he’s donating.
Slow markets (elections, Fed decisions, geopolitics, anything resolving in weeks or months) flip the logic. Speed is worthless: there’s nothing to be fast about for the next ninety days. What gets paid is the quality of the estimate, meaning who understands polling error, or central bank behavior, or how coalition talks actually end. Bots can trade these markets, but they hold no advantage in judging them, and someone who knows the field deeply can still be the smartest one at the table.
So the practical rule writes itself, and it’s the one takeaway I’d defend hardest: match your edge to the horizon. If your advantage is knowing a subject, trade markets where that knowledge has time to matter. If your advantage is reaction speed, I have bad news about your competition.
What to do with this
Before any trade, three questions, in this order:
- Who plausibly took the other side? A market maker earning a spread, a person with a different read, or a machine that’s faster than me?
- What decides this market: speed or judgment? If the answer is speed, leave. Not because you’ll always lose, but because you’re playing an away game against an opponent built for it.
- Why is the other side wrong? If you can’t answer in one sentence, you don’t have an opinion yet, you have a mood. The pre-trade checklist picks up from there.
None of this says bots ruined prediction markets. On slow questions the crowd, human judgment included, is still what sets the price, and that’s where these markets do what they promise. The machines just closed the lanes where thinking was never the point. Arguably they did beginners a favor: the sign on the fast lane used to be invisible, and now it reads plainly. Nothing to forecast here.
Nothing here is financial advice. Prediction markets involve real risk, so never trade money you can’t afford to lose.