The premise of copy-trading rests on a testable claim: that large directional bettors on prediction markets produce better resolved outcomes, on average, than the broader population of participants. The claim is testable because every prediction-market contract resolves to a known outcome, and every trade has a recorded entry price. This article examines the available evidence on whale profitability, sets out the structural reasons that prediction markets reward size in a way that other venues do not, and explains the practical methodology Rivo uses to evaluate wallet performance.
The receipts: top whale wins on file
Before the analysis, the data. The list below ranks the largest paper-PnL gains across the population of whales Rivo has tracked on Polymarket and Kalshi. The figures are calculated as if each position was held to resolution at the entry price; the methodology is described in detail in our resolution mechanics article.
- 1Will United States win on 2026-06-12?Polymarket·culture·stake $920K·@ 46¢+$1.1M
- 2Will IR Iran win on 2026-06-15?Polymarket·politics·stake $790.8K·@ 49¢+$823.1K
- 3Will Spain win on 2026-06-15?Polymarket·culture·stake $78.6K·@ 9¢+$794.4K
- 4Will Spain win on 2026-06-15?Polymarket·culture·stake $47.4K·@ 9¢+$479.4K
- 5UFC 328: Sean Strickland vs. Khamzat Chimaev (Middleweight, Main Card)Polymarket·sports·stake $103.6K·@ 19¢+$441.8K
- 6Will Canada win on 2026-06-18?Polymarket·culture·stake $1.3M·@ 76¢+$420.1K
- 7What will the announcers say during Jordan vs Argentina?Kalshi·culture·stake $4.1K·@ 1¢+$407.0K
- 8Spread: Spain (-2.5)Polymarket·culture·stake $345K·@ 46¢+$405K
- 9Spurs vs. ThunderPolymarket·sports·stake $198K·@ 33¢+$402K
- 10Roland Garros ATP: Jakub Mensik vs Andrey RublevPolymarket·sports·stake $370.0K·@ 48¢+$400.8K
What the leaderboard reveals about whale behavior
The structural features of the leaderboard are more informative than any individual trade. Three patterns repeat across the largest paying positions.
The first is that almost none of the largest wins were entered near a coinflip price. The entry prices cluster in two ranges: below 0.20 on contracts that eventually resolved YES, and above 0.80 on contracts that eventually resolved NO. In both cases the whale was taking the side of the book that the broader market did not want, and was paid handsomely when the call resolved correctly. The trades that produced six-figure gains are almost uniformly asymmetric positions where the market priced the outcome at a probability the whale judged to be materially wrong.
The second pattern is concentration by category. The leaderboard is dominated by political and crypto markets, with a smaller but meaningful contribution from macro contracts on Kalshi. Sports markets are notably absent from the largest wins, despite producing high trade volume in aggregate. The reason is that sports prices are tightly quoted by professional traders working from well-developed models, so the price rarely sits at the extreme levels that produce large multipliers on a winning position.
The third pattern is wallet repetition. A relatively small number of wallets account for a disproportionate share of the largest wins. The same identifiers appear across the top of the leaderboard for cycle after cycle, which is the strongest single piece of evidence that whale outperformance is a function of skill rather than luck. Pure variance would distribute the largest wins more uniformly across the population of large traders.
Why prediction markets reward size differently
Most financial markets do not reward size in the way that prediction markets do. A large position in an equity index, for example, often reflects passive flow rather than directional conviction, and the size of the position carries little information about the underlying view. Prediction markets are different in several specific ways that combine to make size a more interpretable signal.
The first difference is the absence of passive flow. No institutional investor holds a portfolio of binary event contracts as an index allocation. Every position on a Polymarket or Kalshi contract was entered by a trader who made an active decision to deploy capital against a specific binary outcome. The directional intent is unambiguous in a way that it is not on a broad-market exchange.
The second difference is contract duration. Prediction- market contracts have defined resolution dates, typically within months, and cannot be held indefinitely as a long-term allocation. The short duration means that any position taken at size must be predicated on a view of the underlying event resolving within the contract's life. Traders who take positions for non-directional reasons, such as market-making or arbitrage, generate distinctive trade patterns that can be filtered out of the directional signal, as discussed in our definition of a whale.
The third difference is market size. Even the largest prediction-market contracts have open interest measured in tens of millions of dollars, which is small relative to traditional financial markets. The implication is that a whale-sized position is a meaningful fraction of the open interest on a given contract, and the position cannot be entered without showing conviction. A trader who places a six-figure position is doing so visibly, with full knowledge that the trade will be observable and will move the price.
The qualifier: not every whale is profitable
The argument above establishes that the population of whale trades, in aggregate, carries directional information. It does not establish that every individual whale is profitable or that every whale trade is worth copying. A meaningful share of the wallets that meet the whale-size threshold on any given week do not produce positive PnL over their full resolved history. Size is a necessary filter for whale classification; it is not a sufficient predictor of edge.
The Rivo dataset shows a wide dispersion in resolved PnL across wallets meeting the size threshold. The top decile of wallets produces substantial positive returns across the resolved sample; the bottom decile produces substantial negative returns; the middle of the distribution is roughly break-even before fees and spread. Copy-trading the population indiscriminately captures the average of this distribution, which is not the goal. Copy-trading the top decile, weighted by resolved history within the relevant category, is the framework that the data supports.
Methodology: how Rivo evaluates wallet performance
The basic scoring mechanism is described in the resolution mechanics article. For wallet evaluation, Rivo applies several additional filters to produce a defensible measure of skill.
The first is a minimum resolved-trade count. Wallets with fewer than twenty resolved positions are excluded from skill rankings because the sample size is too small to distinguish performance from variance. Twenty trades is a rough floor at which the standard error of the observed win rate becomes narrow enough to support modest inferences about underlying skill.
The second is category-specific scoring. A wallet's performance in political markets is tracked separately from its performance in sports or crypto markets. The aggregate PnL across categories can mask a wallet that is genuinely sharp in one category and consistently wrong in another. Category-specific scoring is particularly important for copy-traders, who should weight a wallet's signal by the category in which the copy-trade is being placed.
The third is entry-price discipline. A wallet that consistently enters at contested prices, below 0.30 or above 0.70, and produces positive PnL across the resolved sample is operating with a different and more interesting kind of edge than a wallet that produces positive PnL primarily from positions taken near a coinflip price. The first kind of edge is more replicable for copy-trading purposes; the second kind depends on execution quality that retail copy-traders rarely match.
The implication for copy-traders
The defensible reading of the evidence is that whale signal is real but narrow. Copy-trading the full population of whale-sized trades produces returns roughly in line with the market average, which is not worth the effort. Copy-trading the population of whales with at least twenty resolved positions, weighted by category-specific PnL and biased toward wallets with contested-price entry discipline, produces materially better expected returns. The framework is not complicated, but it requires discipline in selecting which whales to follow and which to ignore.
For the practical mechanics of executing copy-trades on the whales the framework selects, see our guide to copy-trading. For the categories where the biggest paying trades concentrate, see where whales win the most.
Frequently asked questions
What percentage of whale trades are profitable?
Across the full Rivo dataset, the population-level win rate on whale-sized buys is modestly above 50 percent, with substantial variation by category. The aggregate win rate is less informative than the category-specific PnL distribution, because the value of a winning trade depends heavily on the entry price.
How long does it take for whale trades to resolve?
Resolution times range from hours to many months depending on the contract. Sports and macro contracts typically resolve within days. Political and crypto contracts can take months. The mean time to resolution across the Rivo dataset is roughly four to six weeks, weighted by category mix.
Are recent whale trades a leading indicator of price movement?
Sometimes. Whale activity can attract follow-on flow, particularly from market participants who treat large trades as a sentiment signal. The effect is most pronounced in thin books and least pronounced in deep markets where the whale's entry can be absorbed without moving the price. As an investment thesis, "follow the whales for short-term price impact" is less reliable than "follow the whales for resolution-time outcome accuracy."
Does Rivo include fees in its PnL calculations?
No. The paper PnL figure assumes the position was held to resolution at the entry price, gross of fees and spread. The realized PnL for any specific trader will be lower after fees, with the magnitude depending on the platform, the contract, and the trader's specific fee schedule. For typical retail-size copy-trades, the difference is small relative to the trade outcome.
Can I see the resolved history of a specific wallet?
Yes, on the per-wallet pages within Rivo. The resolved trade list is displayed alongside aggregate PnL and category breakdown. Wallets with fewer than twenty resolved positions are flagged as having insufficient history for high-confidence skill inference.