Risk Management for Prediction Market Traders
How to size positions, protect your bankroll, survive drawdowns, and build a trading approach that doesn't blow up when you're wrong. The unsexy fundamentals that actually determine long-term outcomes.
Why Risk Management Hits Different on Prediction Markets
Risk management is boring. It's the thing traders read about and then ignore until they've blown up their first account. Then they read it again more carefully.
Prediction markets have some specific risk characteristics that make standard equity or crypto risk frameworks only partially applicable. Understanding those differences is where we start.
First, prediction markets are binary. Almost every position on Polymarket resolves to either YES ($1) or NO ($0). There's no middle ground, no 'almost right,' no partial credit. If you buy YES at 70 cents and the event resolves YES, you made 30 cents per share. If it resolves NO, you lost 70 cents per share. That binary outcome structure means that your entry price carries enormous weight — far more than in markets where 'kind of right' still makes you money.
Second, prediction markets have hard resolution dates. Your position isn't open indefinitely. It will close at a specific time and event, and the market will resolve one way or the other. This creates a fundamentally different time pressure than equity positions where you can theoretically hold forever. A bad prediction market trade can't be 'waited out' past resolution.
Third, liquidity varies enormously across markets. Some Polymarket markets have hundreds of thousands of dollars of depth on both sides. Others are thin enough that a $500 trade meaningfully moves the price. Illiquidity risk in prediction markets is asymmetric — it's often easy to get in but expensive or impossible to get out at a reasonable price if you need to exit early.
Fourth, information has hard edges here. In equity markets, information is continuous — earnings come out, guidance gets revised, the story evolves. In prediction markets, information often arrives in discrete jumps: a court ruling, an election result, an announcement. The market can move 40-60 cents in seconds when key information arrives. Position sizing that ignores this jump risk is sizing that's going to hurt you eventually.
The 1% Rule and Why It's Not Conservative Enough
The classic risk management rule says never risk more than 1-2% of your total trading bankroll on any single trade. This comes from equity trading and it's directionally right but needs adjustment for binary markets.
Here's the math: if you risk 2% per trade and you're wrong 10 times in a row (which absolutely happens in prediction markets — even good traders have losing streaks), you've lost 18% of your bankroll. Uncomfortable but survivable. If you're risking 10% per trade and have that same 10-loss streak, you've lost 65% of your capital. That's much harder to recover from psychologically and mathematically.
The rule I'd actually use: think in terms of maximum dollar loss per position, not position size. Cap your maximum loss — your worst-case scenario if the trade resolves against you — at 2-3% of your total bankroll per trade. Work backwards from there to determine your position size.
Example: $2,000 bankroll. Maximum loss per trade: $50 (2.5%). You want to buy YES at 35 cents, meaning if it resolves NO you lose the full 35 cents per share. Maximum shares you should buy: $50 / $0.35 = 142 shares = $50 maximum downside.
Where the 1% rule is too conservative: good setups don't come along every day. When a genuinely exceptional opportunity appears — high conviction, verifiable information edge, favorable resolution timing — my personal framework allows up to 5% on high-conviction plays while keeping routine positions at 1-2%. But that 5% is reserved, not default.
Bankroll Management: Sizing the Whole Operation
Bankroll management is upstream of position sizing. Before you decide how much to put on any individual trade, you need to decide how much of your total capital is allocated to prediction markets at all.
Decide upfront what your prediction market bankroll is. This is money you've consciously set aside for this activity, separate from emergency funds and long-term investments, that you could theoretically lose entirely without it materially affecting your life.
Inside that bankroll, divide further: how much is available for active trading versus held in reserve? A workable framework is 70/30 — 70% deployed or available for deployment in active positions, 30% held in reserve. That reserve funds averaging into positions that move against you temporarily, lets you capitalize on market dislocations that appear suddenly, and provides psychological cushion.
Review and adjust your bankroll periodically but not too frequently. Monthly is probably right. Looking at it daily during a drawdown period is a recipe for reactive decisions.
Correlation Risk: The Invisible Portfolio Killer
You can be a skilled trader, size your individual positions perfectly, and still blow up your portfolio if you don't manage correlation. This is the most underrated risk in prediction market trading.
Correlation in prediction markets clusters around events and themes. Political cycle correlation is the most obvious — anything that resolves in the same election or same political moment is correlated. But there are less obvious clusters: multiple crypto regulatory markets that all move with SEC sentiment, multiple economic markets that all track the same macro narrative.
A workable rule: no more than 20-25% of your bankroll should be exposed to any single underlying event or narrative. If a series of markets all trace back to the same root cause, treat them as a single bet for exposure purposes.
The simplest version of correlation management: look at your open positions and ask 'what's one single event that would wipe out most of this?' If you can identify such an event easily, you've got correlation risk worth addressing.
Max Exposure Rules and Hard Limits
Hard limits are useful precisely because they're hard. A rule that says 'I shouldn't have more than 60% deployed' that you negotiate with yourself every time isn't a limit. It's a suggestion.
Recommended hard limits:
Max per position: 5% of bankroll. High-conviction exception: 8% if you have genuine informational edge. The 8% exception is not an 'I feel really good about this' permission slip.
Max deployed at any time: 70% of bankroll. The other 30% is dry powder. When you're at 70% deployed, new opportunities get evaluated against existing positions.
Max single-category exposure: 30% of bankroll. Add up all your open positions that share the same underlying correlation cluster.
Max loss per day: 10% of bankroll. If you've hit a 10% daily drawdown, stop trading for the day.
Max loss overall before reassessment: 25% of bankroll. Take a pause and figure out what went wrong before continuing.
Drawdown Recovery: How to Dig Out Without Making It Worse
Drawdowns are inevitable. The question isn't whether you'll have a drawdown. It's whether your behavior during a drawdown makes it worse.
The psychological trap: you're down 20%. Every trade feels like an opportunity to claw it back. You find yourself taking positions you'd have passed on when you were up. This is revenge trading and it turns a 20% drawdown into a 40% one.
The math of recovery is brutal. Down 20% requires a 25% gain to break even. Down 30% requires a 43% gain. Down 50% requires a 100% gain.
Practical steps for drawdown recovery:
First, reduce position sizes — don't increase them. During a drawdown, your bankroll is smaller. Your per-trade limits should shrink proportionally.
Second, go back to basics. Trade only the setups where your edge is most clearly defined.
Third, look honestly at whether the drawdown is bad luck or bad process. Did you follow your system? Were the losses within normal expected variance?
Fourth, extend your timeline. Trying to recover a 3-month drawdown in 2 weeks is itself a risk-taking error.
When to Take Profit vs Let It Ride
This question doesn't have a clean answer. Both 'always take profit at X' and 'always hold to resolution' are wrong. The real answer depends on market conditions, your position structure, and what new information has arrived since you entered.
The case for taking early profit: you entered at 30 cents and the market has moved to 65 cents. You've already captured a substantial portion of the potential move. If you exit at 65, you're locking in a 117% return.
The case for holding to resolution: if your original belief was correct and nothing has changed, the market is still mispriced. Exiting early means you're voluntarily giving up the remaining edge you identified.
A simple heuristic: when a position has reached 50-60% of its maximum possible gain, take half off the table. Let the rest ride. This approach locks in a meaningful gain while preserving upside on the remaining position. You can never be perfectly wrong with this strategy.
The trap to avoid: using 'I'll let it ride' as a rationalization for not making a decision. Holding is a decision just like exiting. It should be made actively based on current conditions.
Hedging Strategies in Prediction Markets
Hedging on prediction markets is more nuanced than on most other markets. Your hedging options are: taking the opposite side of a position, entering correlated markets in the opposite direction, or exiting partially.
Direct hedging — buying the opposite side: if you have YES at 35 cents and the market moves to 65 cents, you could buy NO at roughly 35 cents. This locks in your gain. But this is almost never optimal compared to simply exiting — transaction costs, spread, and opportunity cost all argue against hedging when you can just exit.
Cross-market hedging: entering related markets in opposite directions. This requires genuine understanding of the underlying correlation structure. Get it wrong and you're just adding correlated exposure to your problem.
Partial exits: the most practical approach. When a position has moved strongly in your favor and you have some uncertainty about holding, exit 30-50% of the position. It's clean, simple, and doesn't require predicting the correlation structure of related markets.
The question of whether to hedge at all comes down to whether the cost of the hedge is less than the value of the uncertainty reduction you're buying. Most of the time on liquid markets with good entry timing, just exiting cleanly is more efficient than constructing a hedge.
Frequently Asked Questions
What percentage of my bankroll should I have deployed in prediction markets at any given time?
70% maximum as a hard ceiling. That 30% reserve is not idle — it's your capacity to respond to market dislocations, average into positions that move against you temporarily, and avoid being forced into bad exits because you need liquidity. Staying below 70% deployed feels conservative when things are going well and looks prescient when they're not.
How should I handle a position that's moved 50% against me?
First, separate the question of whether to hold from the emotional question of whether you want to recover the loss. Look at the current probability implied by market price versus your assessment of actual probability. If you entered at 40 cents because you estimated true probability at 60%, and the market has moved to 20 cents — do you still estimate 60% probability? If yes and your reasoning is sound, the trade is now even better value and there's an argument for holding. If your original thesis has been weakened by new information, exit. Never hold a losing position just because you don't want to lock in the loss.
Is it worth hedging on prediction markets or should I just exit?
For most traders, exiting cleanly is better than constructing hedges. Hedges require two transactions, both incurring fees, and require you to correctly model correlation between markets. Direct hedging by buying the opposite side of your own position is almost always worse than just selling your position outright. Cross-market hedging can be useful for large positions where you want to maintain some exposure while reducing risk, but it requires genuine understanding of the underlying market relationships.
How do I know if my drawdown is bad luck or a sign my strategy doesn't work?
Review your last 20-30 closed positions and check two things: first, did you follow your system rules consistently (position sizing, entry criteria, exit criteria)? If no, the drawdown is partially self-inflicted and correctable. Second, on the positions that followed your rules, was the outcome variance within what you'd expect given the probabilities? Five losses in a row at 40% win probability is bad luck, not evidence of no edge. Ten losses in a row at 70% win probability starts to look like the edge isn't there. The sample size for any conclusion is larger than most people think.
Should I use the same risk parameters for copy trading as for my own independent trades?
Yes on the aggregate limits (max total deployed, max category exposure, max daily loss) and slightly more conservative on per-position sizing for copy trades. The reasoning: for independent trades, you understand the specific rationale and can make nuanced hold/exit decisions. For copy trades, you're relying on someone else's thesis and have less ability to assess mid-trade whether conditions have changed. The uncertainty deserves a slightly smaller position.