FAQ

RealPnL is the public trade record of Swarm, a crypto futures trading bot. Every trade, open position and loss is published as it happens, rebuilt from the bot's own records every 10 minutes, together with the research behind each of its rules. Trades are never removed or edited after the fact.

Swarm is a systematic trend-following strategy on crypto perpetual futures, run on Binance and Hyperliquid. It holds a few of the largest coins at a time, long or short, sizes each position from its volatility, and keeps a stop-loss order on the exchange for every position. Investment approach.

Not yet. The Binance account runs on Binance Demo Trading, where orders go through the exchange's own order handling but the capital is simulated. Its fill prices and slippage come from the demo order book. The Hyperliquid account is a paper account: fills are simulated at live Hyperliquid prices and no orders are sent; every paper fill is charged as a taker (Hyperliquid fee plus 0.02% slippage), and maker fills are never simulated. Both have run since 27 September 2026. A real-money account, once opened, will be labelled LIVE and linked to its public page on the exchange.

The Binance account DEMO stands at -11.27% since 27 September 2026 (47 days; equity at 13 Nov 2026, 12:10 UTC against start equity), with a maximum drawdown of -11.5% over 67 fills. A sample this short says very little about a trend strategy, which can go several months without a new equity high. Full trade list in the track record.

No. No return is targeted or promised. The live record since 27 September 2026 is the only evidence from trading, and it is short. The backtest is hypothetical: net of costs and funding, its Sharpe ratio, from the tear sheet's monthly returns, was about 2.1 in 2021–2023 and about 0.8 in 2024–2026, and it returned about 6% in 2025. Recent years have been weaker. Journal, 31 October 2026.

From 2021 to 2025 the strategy's backtest returned 43.5% a year after costs and funding. Over the same years the S&P 500 returned 14.3% a year (total return), the HFRI Fund Weighted Composite of hedge funds 7.1% and Bitcoin 24.8%. Bitcoin's maximum drawdown was −77.1%, against −18.8% for the strategy. The strategy figure is a hypothetical backtest, most of it earned in 2021–2023; the others are realised index returns. Year-by-year comparison.

45.7% a year from January 2021 to September 2026 after 0.07% cost per unit traded, with a Sharpe ratio of 1.67 and a maximum drawdown of −18.5%. Deducting Binance's historical funding payments brings that to 41.8% a year. The result is hypothetical and flattered by 2021. Net of costs and funding, the Sharpe ratio from the tear sheet's monthly returns was about 2.1 in 2021–2023 and about 0.8 in 2024–2026, and 2025 returned about 6%. Year-by-year results on the strategy page.

Ten USDT perpetual futures: BTC, ETH, BNB, XRP, SOL, DOGE, ADA, LINK, AVAX and DOT. It holds a few of them at a time and can be long or short each one.

The model runs every four hours and builds up new positions gradually rather than all at once. Most runs change nothing; an active day has a few trades. Stop-loss orders rest on the exchange, so every position is covered between runs.

The account is set to 5× cross margin, but positions are sized far below that. In the backtest, gross exposure averaged about 0.3× equity and peaked at about 1×. Hard caps: no coin above 25% of equity, and total exposure never above 1.5×.

Every position has a stop-loss order on the exchange 25% from entry. If the account falls 25% below its peak, all positions are halved until a person reviews it. After an 8% loss in one day, no new positions are opened until midnight UTC, and a coin that hits its stop cannot be bought back for 24 hours. All risk limits.

A lead trader's orders are reproduced automatically in your own account, scaled to your allocation, in return for a share of your profit (10% by default on Binance). On Hyperliquid the equivalent is a vault, in which depositors' USDC is pooled and traded as one on-chain account. You can stop copying at any time. Guide 08 covers both.

For the copier, only if the lead has an edge after costs, the copy reproduces it (enough capital, fixed-ratio mode) and the copier stays allocated through drawdowns. In our backtest, a strategy compounding at 41.8% a year still lost money in 52% of weeks. A copier who exits during drawdowns can therefore lock in losses on a strategy that is profitable over the full period. Worked figures in the guide.

Rank by maximum drawdown rather than ROI. Then look for an unbroken record of at least three months, measured exposure rather than the leverage setting, visible losing trades, and returns stated net of fees and funding. Guide 06 lists the warning signs.

No. Copy trading is closed. It will not open before the live record covers at least 90 days and passes its review. The copy trading page describes how a Binance copy would work and what it would cost.

No. Copying on Hyperliquid works through a vault: depositors' USDC is pooled, the vault trades as a single account, and every position is visible on-chain. No vault exists for this strategy. Copy trading is closed. It will not open before the live record covers at least 90 days and passes its review.

Standard exchange trading fees on every copied order, plus a share of profit to the lead trader or vault leader, charged only on profit. In the backtest, trading costs came to about 3% of equity a year and funding to about 2.7%. No referral link is published on this site; a link marked as one would pay us a commission out of the exchange's fees, with the account holder's fees unchanged. Full disclosure.

Copy trading of Swarm is closed, so no amount is suggested. As a fact about order sizes: Binance rejects futures orders below 5–50 USDT depending on the coin, and many of the bot's orders are small resizes. On a $100 copy, 68% of its orders would be too small to place; at $1,000, about 7%; from $2,500, almost none. For any strategy, an allocation that could not stay through a 20% fall is too large. Note 06: minimum order sizes.

The backtest had twelve drawdowns deeper than 8% in under six years, about two a year. The deepest was −18.5%, and the longest time to a new high was 241 days (March to November 2024). Four months in ten were negative. Each drawdown, with dates.

Anyone who needs the money within a year, would stop after a 20% drawdown, wants frequent trading, cannot legally use crypto futures where they live, or is basing the decision on the backtest. More on the copy trading page.

Yes. The performance page lists every open, add, trim and close with time, price and P&L as a share of equity, and the same record is published as JSON. The site is rebuilt from the bot's records every 10 minutes, and no trade is ever removed. Once real-money accounts open, each will link to its public page on the exchange.