Five Strategy Archetypes. Pick One. Know Why It Breaks.
A strategy is two things: an assumption about how the world works, and a rule for betting on that assumption. Every strategy that has ever worked can be traced to one of five assumption families.
Knowing the family tells you two things that any individual strategy doesn't: what the strategy is actually betting on, and what kind of world will break it.
Here are the five families, with a Kalshi example for each.
I · Mean Reversion
The assumption: prices bounce around a fair value. When they get too far from it, they come back.
Mean reversion is the oldest family in trading. It works because most short-term price moves in liquid markets are noise, and noise by definition cancels out.
Kalshi example: "Buy YES on any S&P-up-today market trading below 42 cents between 2:30 PM and 3:30 PM Eastern if the index is within 0.3% of its morning open."
When it wins: liquid, range-bound markets. Quiet news days. Markets with well-anchored fair values.
When it breaks: regime shifts. A jobs report that moves the real economy isn't noise. A tariff announcement that reprices risk isn't noise. If you are mean-reverting into a regime shift, you are shorting a trend, and trends compound against you.
The usual fix is a stop-loss by time or by magnitude. If the reversion hasn't happened in X minutes or if the price has moved Y cents further away, exit.
II · Momentum
The assumption: prices that have moved will keep moving.
Momentum is the opposite bet from mean reversion, and it wins in exactly the markets where mean reversion loses. A jobs report that moves the real economy will keep moving markets as the implications propagate. A polling surprise will cascade through related contracts.
Kalshi example: "If the 12-hour Fed-rate-cut market has moved more than 8 cents in the last two hours in either direction, buy in the direction of the move and hold for four hours."
When it wins: trending regimes. News-driven days. Markets that are reacting to information that has second- and third-order effects.
When it breaks: range-bound markets. News that is already priced in. The moment the crowd realizes "we already moved too far" and the reversion kicks in.
The trick with momentum is that it works more often than it doesn't, but when it fails it fails big. Position sizing matters more than entry timing.
III · Market Making
The assumption: traders will pay a small premium to transact immediately. If you quote both sides, you collect that premium over thousands of trades.
Market making is the family that made the modern prop trading industry. It is unglamorous, capital-intensive, and reliable.
Kalshi example: "On any weather contract with more than $50k daily volume, post a 3-cent-wide spread centered on the current midprice. Re-quote every 90 seconds. Cap inventory at $2,000 per contract per side."
When it wins: liquid markets with steady two-way flow. Markets where your spread is tighter than the next-best quoter.
When it breaks: adverse selection. If informed traders are hitting your bids because they know something you don't (a pending resolution, a news event), you lose every trade. Markets near resolution are the classic trap.
Market making on Kalshi is quieter than on crypto exchanges. That's good for the simple reason that there are fewer competing market makers, but bad because volumes are lower. Patience is part of the strategy.
IV · Arbitrage
The assumption: two related contracts should be priced in a specific relationship to each other. When they aren't, trade the spread.
Arb is the most mechanical family. You don't need to predict what prices will do. You only need to notice when two related prices are wrong relative to each other, and trade until they agree.
Kalshi example: "Monitor the BTC-above-$95k and BTC-above-$100k contracts on the same expiry. If the $95k contract ever trades below the $100k contract, buy the $95k and short the $100k. Close when the spread inverts."
When it wins: whenever mispricings exist. On Kalshi, they exist because liquidity is fragmented across contracts and different traders are active on different ones.
When it breaks: execution risk. If you fill the first leg but not the second, you are now naked-long one side, and the price can move against you before you close. Partial fills on arbitrage are how arbitrage desks lose money.
Arb bots live or die by how smart they are about legging in. Aggressive orders on both sides at once. Wait-for-both-fills before considering the position open.
V · News Reaction
The assumption: when news drops, the market takes time to fully incorporate it. If you are faster than the median trader, you capture that gap.
This is the family most people imagine when they think "trading bot." In practice it's also the family with the shortest half-life.
Kalshi example: "When a scheduled economic data release is published, parse the actual number against the consensus estimate from our data provider. If the actual beats by more than 0.2 standard deviations, buy YES on any rate-cut contract within 30 seconds."
When it wins: scheduled releases, because you know exactly when news will drop and you can preload the execution path. Also unscheduled news, if you have a signal that's faster than the Kalshi orderbook.
When it breaks: you are competing with other fast traders, and every quarter the field gets faster. A bot that was profitable a year ago on news reaction is probably not profitable today unless it's been retuned.
News reaction is also the family most prone to false positives. The release you were watching came in as expected, the algorithm thought it was a beat because of a rounding issue, and you bought into a contract that was about to fall.
How to Pick
If this is your first strategy, start with mean reversion or market making. Both are forgiving. Both tolerate small mistakes. Both are mostly a game of letting many small wins accumulate.
If you have an edge on a specific news cycle or data release, start with news reaction. Narrow focus beats broad intuition.
If you have capital and patience, market making is the most durable family.
Momentum and arbitrage are advanced families. They work, but they require stricter discipline: position sizing for momentum, execution logic for arbitrage. Don't start there.
The common mistake across all of them is picking the family based on what shows the highest backtest ROI. A mean-reversion strategy that backtests at +50% is not better than a market-making strategy that backtests at +12%, because the first is fragile and the second is robust. Pick based on which family your conviction actually matches.
This post is for informational purposes only and does not constitute financial advice. All strategies described carry risk of loss. Never trade with money you can't afford to lose.