Polymarket Position Sizing: How Much to Bet When You Copy a Whale
Pick the right trader to copy and size the trades wrong, and you'll still lose money. Pick a merely decent trader and size correctly, and you'll probably survive long enough to find a better one. Position sizing is the least glamorous decision in copy trading and the one that actually determines whether you're still in the game in six months. Here's the complete sizing playbook for Polymarket — proportions, fixed units, Kelly math, and the hard caps that keep one bad week from becoming your last one.
Position sizing on Polymarket, in short: (1) never copy a whale's dollar amounts — copy the fraction of bankroll the trade represents; (2) if you can't estimate their bankroll, use a fixed unit of 1–3% of your account per position, scaled by conviction; (3) if you want the math, use the binary Kelly formula — stake = (p − price) / (1 − price) — at quarter strength; (4) cap exposure per market, per trader, and in total, so no single outcome can take you out.
Why sizing beats selection
Copy traders obsess over who to copy and improvise how much. That's backwards, and the reason is arithmetic: losses compound against you asymmetrically. Lose 10% of your bankroll and you need 11% to get back to even. Lose half and you need to double your money. Deep drawdowns aren't just painful — they're mathematically expensive to climb out of.
| Drawdown | Gain needed to recover |
|---|---|
| −10% | +11% |
| −25% | +33% |
| −50% | +100% |
| −75% | +300% |
Every sizing rule in this guide exists to keep you off the bottom rows of that table. A skilled trader with a real edge still loses 30–40% of their trades — if any one of those losses is big enough to knock you into deep drawdown, the edge you're copying never gets the sample size it needs to pay you.
Rule 1 — Copy proportions, not dollars
The cardinal rule. A whale with a $500,000 bankroll putting $10,000 on a market is making a 2% bet — routine, survivable, boring. If you mirror the $10,000 on a $2,000 account, you've bet five times your entire bankroll on the same outcome. Identical trade, catastrophic difference in risk.
The proportional version: 2% of your $2,000 is $40. That's the equivalent trade. If $40 feels too small to bother with, that's not a sizing problem — it's your bankroll telling you the whale's strategy operates at a scale you should scale down, not stretch toward.
This rule is step three of the full how to copy trade on Polymarket loop, but it's important enough to deserve its own guide — because in practice, "copy the proportion" immediately raises the question the next section answers.
How to estimate a whale's bankroll
Polymarket wallets don't wear their bankroll on their sleeve, but on-chain data gets you a workable estimate:
- Total open exposure. Add up the wallet's current open positions. A trader with $80,000 across open markets is operating on at least that; their true bankroll is usually meaningfully larger, since disciplined traders keep reserves.
- Typical position size vs. their own history. What you really need is not their bankroll in dollars but what a normal-sized bet looks like for them. If a wallet's positions cluster around $5,000 and today's trade is $5,000, that's a routine 1-unit bet — copy it at your own 1-unit size. If today's trade is $25,000, that's a 5-unit conviction swing, and you can scale accordingly.
- Realized P&L as a floor. A wallet that has banked $200,000 in profits and keeps trading similar sizes clearly isn't running a $10,000 account.
Shortcut: a wallet analysis that shows a trader's open positions, position-size history, and true P&L (including redeems) gives you all three estimates in one place — that's exactly what a Polyvision wallet scan returns, alongside the 0–10 copy score.
The fixed-unit system (the sane default)
If estimating bankrolls sounds like work, use the system most disciplined copy traders end up at anyway: a fixed unit of 1–3% of your account per position, scaled by conviction.
- Base unit: 1%. Your default for any trade you copy from a vetted wallet.
- Scale by score, not vibes. Copying a trader scoring 8+ on a rigorous skill evaluation? You can justify 2–3%. Score 6–7.9? Stay at 1%. Below 6, the question isn't sizing — it's why you're copying them at all.
- Never exceed your cap because a trade "feels" certain. Prediction markets resolve binary: the 85¢ favorite that loses takes 100% of your stake with it, not 15%.
On a $2,000 bankroll this means $20–$60 per position. On $10,000, $100–$300. Small enough that a losing streak is an annoyance; large enough that a genuine edge compounds.
The Kelly criterion, for people who want the math
Prediction markets are one of the cleanest real-world fits for the Kelly criterion, because a share is a pure binary bet: it costs c (the price, between 0 and 1) and pays $1 if the outcome happens. If you believe the true probability is p, the Kelly-optimal fraction of bankroll to stake is:
f* = (p − c) / (1 − c) — your edge over the price, divided by what a winning share still has to run.
Worked example: a whale you trust bought "Yes" at 50¢, and their track record in this category convinces you the true probability is around 60%. Then f* = (0.60 − 0.50) / (1 − 0.50) = 20% of bankroll.
Do not bet 20% of your bankroll. Full Kelly assumes your probability estimate is exactly right — and when you're copying, your "estimate" is secondhand conviction in someone else's judgment. Overestimating your edge with full Kelly doesn't just cost returns; it actively increases your risk of ruin. The standard fix:
- Use quarter-Kelly. 20% full Kelly becomes a 5% stake — aggressive but defensible for a top-scored trader. Half that again if the wallet is new to you.
- Respect the sign. If the market has drifted to 65¢ by the time you can enter and your honest probability is still 60%, then f* = (0.60 − 0.65) / 0.35 < 0. Kelly isn't telling you to bet smaller — it's telling you the trade no longer exists at your price. This is the mathematical version of "don't chase entries."
- Kelly assumes independent bets. Three simultaneous positions on the same election resolve as one bet, not three. Treat correlated markets as a single position for sizing purposes.
If this feels like overkill: it mostly is. The fixed-unit system above is quarter-Kelly-ish in practice for typical edges, without the estimation burden. Kelly's real gift to a copy trader is the negative-edge check — a built-in alarm for overpaying.
Portfolio caps: sizing above the single trade
Per-trade sizing protects you from one bad market. These caps protect you from one bad week:
- Per-market cap. All positions on one market (or on tightly correlated markets) count as a single bet — cap them together at your per-trade maximum.
- Per-trader cap. Even a 9+ wallet can go cold or change style. Cap the total bankroll allocated across any one copied trader — around 20% is a common ceiling — so no single stranger's tilt becomes your drawdown. Better yet, copy a small basket: our copy-trading strategy guide covers building one across categories.
- Total-exposure cap. Keep a meaningful fraction of your bankroll — 40–60% for most people — out of open positions. Polymarket capital is locked until resolution; dry powder is what lets you take the next great entry instead of watching it.
- A drawdown stop. Decide now, in writing: "if I'm down 25% from my peak, I stop opening positions and re-vet every wallet I copy." The moment you need this rule is the moment you're least capable of inventing it.
The sizing mistakes that end copy-trading careers
- Mirroring dollar amounts. The classic. Five-times-bankroll bets don't get a second chance.
- Martingaling losses. Doubling size to "win it back" is the exact red flag you'd disqualify a trader for — don't become it. (Loss-chasing is one of the patterns that caps a wallet's copy score for a reason.)
- Sizing up on a hot streak. Five wins in a row is well within normal variance, not a promotion. Change your unit when your bankroll changes materially, not when your mood does.
- Ignoring correlation. Copying three traders who are all long the same outcome is one large bet wearing three disguises. Check what you actually hold across wallets.
- Betting rent money. No sizing system fixes a bankroll you can't afford to lose. Prediction markets go to zero routinely — fund your account accordingly.
Putting it together: a worked example
Say you have a $2,000 bankroll and you're copying two vetted wallets: Wallet A (score 8.6, sports specialist) and Wallet B (score 7.2, politics). Your rules might look like:
- Unit size: 1% ($20) baseline; 2.5% ($50) max for Wallet A's routine-sized trades; 1% flat for Wallet B.
- Per-market cap: 3% ($60) across both wallets combined.
- Per-trader cap: $400 (20%) allocated to either wallet's open positions at once.
- Total exposure: never more than $1,000 (50%) locked in open markets.
- Drawdown stop: at $1,500 (−25%), freeze and re-vet.
Wallet A opens a position that's twice their usual size, in a market priced near their entry — you take your $50 max unit. Wallet B piles into a market that's already run 12¢ past their fill — Kelly's sign check says the edge is gone at your price; you skip it. Boring, repeatable, survivable. That's what winning sizing looks like.
Where Polyvision fits
Polyvision doesn't place trades or size them for you — you always execute on Polymarket yourself. What it gives you are the inputs the sizing rules above need: true P&L with redeems, position-size history, open exposure, red flags like martingale patterns, and the 0–10 copy score you can key your conviction scaling to. The daily feed shows each top position's entry price against the current price, so the "has it run past the edge?" check takes seconds instead of spreadsheet time. And if you want to automate your own sizing rules, every metric is available through the REST API and MCP server.
Frequently asked questions
How much should I bet per trade on Polymarket?
A sane default is a fixed unit of 1–3% of your bankroll per position, scaled by conviction: closer to 1% for a moderate copy (score 6–7.9), up to 3% for a strong one (8+). The exact number matters less than applying it every time — no single market, and no single trader, should be able to take you out.
Should I copy a whale's exact bet size?
No — never copy dollar amounts. Copy the proportion of bankroll the trade represents. A whale betting $10,000 from a $500,000 bankroll is risking 2%; the equivalent on your $2,000 account is $40. Mirroring dollars instead of proportions is the fastest way to blow up while copying a profitable trader.
Does the Kelly criterion work on Polymarket?
Yes — binary prediction-market shares fit Kelly cleanly: stake fraction = (p − price) / (1 − price). But full Kelly assumes your probability estimate is exactly right, which it never is when you're borrowing someone else's judgment. Use quarter-Kelly or less, and if the formula goes negative — the price has run past the edge — skip the trade.
How much of my bankroll should be deployed at once?
Cap total open exposure at roughly 40–60% of bankroll, with per-trader caps (~20% to any one wallet) and per-market caps on top. Polymarket locks capital until resolution, so dry powder is both risk control and the ability to take the next good entry.
What if I can't estimate a trader's bankroll?
Use the fixed-unit fallback: 1–3% of your own account per copied position, scaled by the trader's copy score. Proportional copying is the ideal, but a consistent fixed unit captures most of the benefit.
Get “The Copy List” — free
This morning’s 10 highest-scoring Polymarket wallets, ranked by our 10-point copy score, in your inbox. No wallet connection, 1-click unsubscribe.
Size with real numbers, not guesses
Paste any Polymarket wallet into Polyvision for true P&L, position history, open exposure, red flags, and a 0–10 copy score — free.
Start Free Analysis
Polyvision