Solana Copy Trading in 2026: What 1,710 Real Trades Reveal

Everyone has an opinion about copy trading on Solana. Almost nobody has data. We run a copy-trading platform, so we pulled the anonymized numbers from our own production database — **1,710 real copied positions, 329 active copiers, 1,282 tracked wallets** — and this is what the on-chain reality actually looks like in 2026. Every number below comes from live production data on [uwuu.ai](https://uwuu.ai), aggregated on September 7, 2026. No survey, no estimates, no vibes. Feel free to cite any of it with a link back to this page.

- **77% of all copied trades execute on the pump.fun bonding curve** — the graduation venues (PumpSwap, Raydium) account for less than a quarter of copy volume - **The median holding time of a copied position is 24 seconds.** Not minutes — seconds. The average is 42 minutes, dragged up by a long tail of bag-holders - **53% of closed copy positions are profitable** (118 of 222 closed positions exited above entry) - **63% of tracked wallets have exactly one copier.** The most-followed wallet has **1,523 copiers** — copy trading has a brutal power law - Automated take-profits fire **2.6× more often** than stop-losses (13 TP exits vs 5 SL exits)

The venue distribution surprised us, and we operate the pipes: | Venue | Share of copied positions | Win rate (closed) | |---|---|---| | pump.fun bonding curve | 77% (1,309) | 84.9% | | Raydium | 14% (239) | 22.7% | | Raydium LaunchLab | 7% (115) | 33.3% | | PumpSwap | 1.6% (28) | 33.3% | | Jupiter (routed) | 1.1% (19) | 0% | Two things stand out. First, **the bonding curve is where copy trading lives**: the traders worth copying are snipers who enter before graduation, and their copiers ride the same curve. Second, win rates collapse post-graduation — by the time a token reaches Raydium or gets routed through [Jupiter](/blog/jupiter-aggregator), the asymmetric entry is gone. A honest methodology note: win rates are computed on *closed* positions only. Losing positions get abandoned as open dust more often than winners, so closed-position win rates overstate the true hit rate. That bias affects every copy-trading stat you will ever read — at least we are telling you.

The single most misunderstood thing about Solana copy trading is time horizon. The **median copied position is closed 24 seconds after it opens**. Copy trading on Solana is not investing — it is latency-sensitive scalping, delegated. This is why execution speed matters more than any other feature of a [copy trading bot](/blog/copy-trading-bot-automate-crypto): a mirror that lands 30 seconds late is not copying the trade, it is exit liquidity for it. It is also why gRPC streaming (we use Yellowstone) beats polling for trade detection, and why we route each copy through the same venue as the original trade instead of an aggregator hop.

Of 1,282 distinct wallets that our users track: - **811 wallets (63%)** have exactly one copier - **271 wallets (21%)** have 2–9 copiers - **200 wallets (16%)** have 10 or more - The most-followed wallet has **1,523 copiers** Most people don't copy influencers — they copy a wallet they found themselves, on [Birdeye](/blog/birdeye-review), [KolScan](/blog/kolscan-review) or a [wallet tracker](/blog/solana-wallet-tracker). The "1,523 copiers on one wallet" number shows what happens when a wallet gets famous; the 811 solo-copied wallets show what copy trading mostly is: private conviction, automated.

Among positions with automated exits configured, **take-profits triggered 13 times and stop-losses 5 times**. Copiers set optimistic exits far more than defensive ones — consistent with the bonding-curve meta where the downside case is a slow bleed (that never hits a clean SL level) and the upside case is a spike worth catching automatically.

1. **Copy wallets that trade the bonding curve.** That's where the measured edge is; post-graduation copying underperforms across every venue we route 2. **Judge a wallet on closed AND open positions.** Anyone can show a 85% win rate by never closing losers 3. **Speed is the product.** With a 24-second median hold, detection latency is the difference between copying a trade and buying someone's exit 4. **Set the stop-loss too.** The data says almost nobody does

On our platform alone, 329 distinct wallets actively copy trade, following 1,282 distinct target wallets. Extrapolated across Telegram bots and other platforms, the realistic total is in the tens of thousands — far smaller than the influencer noise suggests.

On closed positions in our dataset, 53% exit in profit, with bonding-curve trades performing far better (84.9% of closed positions profitable) than post-graduation trades. Beware: closed-position stats overstate performance because losing bags stay open longer.

Measured by both volume and win rate: the pump.fun bonding curve, with 77% of all copied positions and the highest closed-position win rate. See our guide to the [best pump.fun bots](/blog/best-pump-fun-bot).

The median copied position is held 24 seconds. Practically, your copy needs to land within a few seconds of the original trade — which requires streaming trade detection (gRPC), not polling.

This study aggregates anonymized production data from uwuu.ai (1,710 copied positions, September 2026 snapshot). You are welcome to reuse any statistic with attribution and a link to this page. For the raw aggregates or an updated snapshot, reach out on [X @uwuuai](https://x.com/uwuuai).