The most frequently asked question on prediction-market subreddits is some version of "who should I follow on Polymarket?" The question is usually framed as a request for specific wallet addresses or trader names, but the productive answer is a methodology rather than a list. Wallets are dynamic; a trader who was sharp last quarter may be cold this quarter, and a trader who is unknown today may produce the next leaderboard entry tomorrow. A list of names goes stale within months. A methodology for evaluating which traders to follow remains valid indefinitely. This article sets out the methodology Rivo applies, with reference to the structural features of the trade tape that make it work.
The four axes of wallet skill evaluation
A defensible methodology for ranking Polymarket and Kalshi traders rests on four axes. Each is necessary; missing any one of them produces a ranking that fails to predict forward returns.
Axis one: resolved trade count
The first axis is the number of resolved positions associated with the wallet. A wallet with fewer than twenty resolved trades does not provide enough statistical surface area to distinguish skill from variance. A trader who is correct on the first three trades they place looks identical, at that sample size, to a trader running a coin-flip strategy. Twenty resolved trades is a rough floor at which the standard error on the observed win rate becomes narrow enough to support modest inferences about underlying skill.
The corollary is that high-profile wallets often have less signal than less-discussed wallets. A wallet that placed one large, correct trade during the 2024 election cycle has approximately the same statistical evidence of skill as any other single correct prediction; the size of the trade does not change the underlying probability that the next trade will also be correct. A wallet with three hundred resolved trades and a positive PnL is meaningfully more reliable as a copy-trading target than a wallet with one famous trade and no broader history.
Axis two: category-specific PnL
Aggregate PnL across categories can mask a wallet that is sharp in one category and consistently wrong in another. A trader with $100,000 of positive PnL in politics and $80,000 of negative PnL in sports has a net positive PnL of $20,000, but the meaningful information is the category specialization. Copy-trading this wallet on sports trades is a recipe for losses; copy-trading the same wallet on politics is a meaningfully positive expected value.
Rivo computes PnL per wallet per category on a rolling twelve- month window. The category breakdown is one of the most important displays in the wallet detail view because it allows copy-traders to apply each wallet's signal selectively. A wallet should be followed only in categories where the resolved PnL is positive across at least twenty resolved trades.
Axis three: entry-price discipline
The price at which a trader enters a market is the most information-dense data point in the trade. A wallet that consistently enters at contested prices below 0.30 or above 0.70 and produces positive resolved PnL is operating with a meaningfully different kind of edge than a wallet that produces positive PnL primarily from positions near coinflip prices.
The first kind of edge is more replicable for copy-trading purposes because the asymmetric payouts on contested-price contracts give the copy-trader more room to absorb the slippage cost of a delayed entry. A trader who buys at 0.10 and resolves correctly is paid nine to one; a copy-trader who fills at 0.12 due to slippage is paid eight to one, which is still attractive. A trader who buys at 0.55 and resolves correctly is paid less than one to one; a copy-trader who fills at 0.57 has surrendered most of the edge before the position is open.
Wallets are ranked partly by the share of their resolved trades that were entered at contested prices. The share is a meaningful predictor of forward replicability, independent of aggregate PnL.
Axis four: holding pattern
Wallets that hold positions to resolution behave differently from wallets that scalp early price moves. Both can be profitable, but they require different copy-trading behaviors. A copy-trader who follows a holder by scalping the position will exit before the underlying thesis plays out, capturing only the modest move that occurred before the exit. A copy- trader who follows a scalper by holding to resolution will give back the edge the scalper captured by exiting early.
Rivo classifies each wallet's typical holding pattern based on the time between opening trades and the wallet's subsequent activity. The classification is approximate but informative; it lets a copy-trader match their own holding behavior to the wallet they are copying.
The wallets that recur on the leaderboards
The most reliable indicator that a wallet is worth following is repeated appearance on the resolved-trade leaderboards. A wallet that appears once in the top ten wins is a single data point. A wallet that appears across multiple top-ten wins over a twelve-month window, with corresponding category specialization, is operating with edge that has compounded.
- 1UFC 328: Sean Strickland vs. Khamzat Chimaev (Middleweight, Main Card)Polymarket·sports·stake $103.6K·@ 19¢+$441.8K
- 2Spurs vs. ThunderPolymarket·sports·stake $198K·@ 33¢+$402K
- 3Will VfB Stuttgart win on 2026-05-09?Polymarket·culture·stake $248.4K·@ 48¢+$270.4K
- 4PGA Championship: Will Xander Schauffele finish top 5 in Round 3?Kalshi·culture·stake $2.6K·@ 1¢+$254.8K
- 5Roland Garros ATP: Ethan Quinn vs Francisco ComesanaPolymarket·sports·stake $380.9K·@ 61¢+$243.5K
- 6Pistons vs. CavaliersPolymarket·sports·stake $151.5K·@ 39¢+$237.0K
- 7Who will win Coach of the Year?Kalshi·sports·stake $2.4K·@ 1¢+$233.9K
- 8Spread: Knicks (-7.5)Polymarket·sports·stake $238.5K·@ 51¢+$229.1K
- 9Spread: Thunder (-15.5)Polymarket·sports·stake $205.8K·@ 49¢+$214.2K
- 10Spread: Pistons (-3.5)Polymarket·sports·stake $213.9K·@ 51¢+$205.6K
The leaderboard above ranks the largest paper-PnL gains across the Rivo dataset. Wallets that recur across the list, particularly across cycles and across categories within their specialization, are the primary candidates for inclusion in a copy-trading universe. The full ranked list is available on the all-time wins page.
How many wallets to follow
A practical copy-trading universe consists of between five and twenty wallets. Below five, the alert volume is too low to produce meaningful trade flow; above twenty, the copy-trader loses the ability to evaluate each individual signal and the portfolio drifts toward the average of the followed wallets.
A defensible starting universe consists of three wallets per primary category. Three politics wallets, three sports wallets, three crypto wallets, three macro wallets produces a twelve-wallet universe that covers the four major categories with enough redundancy to absorb the occasional cold streak from any individual wallet. The universe should be reviewed quarterly and rebalanced based on updated resolved PnL.
What to avoid when picking wallets to follow
Several heuristics for picking wallets are intuitive but produce poor results in the data.
The first is picking wallets based on recent activity. A wallet that placed a large trade in the last week, regardless of resolved history, is not a wallet worth following. The recent activity is the alert, not the qualification. The resolved history is the qualification.
The second is picking wallets based on social-media notoriety. The wallets discussed on Twitter, in subreddits, and on prediction-market commentary blogs are often discussed because they made one famous trade, which is a single data point. The analytical evidence on the rest of their activity is usually thin. Following the famous wallet without examining the broader resolved history is a recipe for disappointing returns.
The third is picking wallets based on absolute PnL without controlling for size. A wallet with $500,000 of positive PnL across twenty trades may have an unusually positive average per-trade PnL of $25,000, which is reassuring, but it may also have produced that PnL from two oversized winning trades and eighteen smaller losing trades. The PnL distribution within the wallet matters; aggregate PnL alone is insufficient.
How Rivo surfaces the best wallets to follow
The wallet detail views within the Rivo platform display all four axes for each tracked whale, alongside resolved-trade history with per-trade outcomes. The platform does not publish a single ranked "best wallets" list because the right list depends on the user's category preferences, holding pattern, and risk tolerance. The methodology is exposed; the application is the copy-trader's responsibility.
For the practical mechanics of acting on the wallets selected through this methodology, see our guide to copy-trading. For the underlying analysis of whether whale signal works in aggregate, see our whale profitability analysis.
Frequently asked questions
Who is the most profitable Polymarket trader?
The single most profitable Polymarket trader on record, by single-position paper PnL, is the wallet behind the 2024 election cluster of trades that exceeded $80 million in aggregate. By rolling twelve-month resolved PnL across a broader sample, the leaderboard is more dispersed and changes as new markets resolve. The methodology above is the right way to identify current top performers; the snapshot at any moment is on the wallet detail views in the platform.
Are anonymous Polymarket wallets worth following?
Yes, with the appropriate caveats. Polymarket wallets are pseudonymous by default, identifiable only by their on-chain address. The lack of an off-chain identity does not affect the methodology described in this article; the resolved history is what matters, and the resolved history is visible regardless of whether the wallet has a public identity.
How often should I review my list of followed traders?
Quarterly is a reasonable cadence. Resolved history accumulates slowly and wallet skill is generally persistent, so monthly rebalancing produces unnecessary churn. Quarterly review is fast enough to catch meaningful shifts in category-specific PnL while slow enough to avoid over-fitting to recent variance.
Can I follow Kalshi traders the same way I follow Polymarket traders?
Yes. The methodology applies identically across both platforms. Kalshi traders are identified by platform-internal trader identifiers rather than by on-chain wallet addresses, but the resolved-history evaluation works the same way. Rivo tracks both venues in a unified data model.
What if a wallet I follow goes cold?
Cold streaks are a normal part of any directional trading strategy. The defensible policy is to continue following the wallet through the cold streak unless the underlying category-specific PnL turns persistently negative across at least ten new resolved trades. Short-term variance is not a reason to drop a wallet with strong long-term performance.