Futures bots

Funding eats first - the drag chart every leveraged bot ignores

Bot	Type	Price	Trade fee	Futures	Backtest	Paper	Exchanges	US
Freqtrade	Self-hosted	free	-	yes	yes	yes	11	US
Hummingbot	Self-hosted	free	-	yes	yes	yes	50	US
Gunbot	Self-hosted	$29/mo	-	yes	yes	yes	20	US
Bitsgap	Cloud platform	$23-149/mo	-	yes	yes	yes	17	US
Jesse	Self-hosted	$0-1599/mo	-	yes	yes	yes	12	US
PI Pionex	Exchange-native	0.05% per trade	0.05% per trade	yes	-	-	-	US
WunderTrading	Cloud platform	$0-100/mo	-	yes	-	yes	18	US
Altrady	Cloud platform	$30-97/mo	-	yes	yes	yes	17	US
Passivbot	Self-hosted	free	-	yes	yes	yes	8	US
3Commas	Cloud platform	$15-140/mo	-	yes	yes	yes	9	US
HaasOnline	Cloud platform	$19.99-249.99/mo	-	yes	yes	yes	24	US
MI Mizar	Cloud platform	0.1% of volume	-	yes	yes	-	10	-
Binance Trading Bots	Exchange-native	free	-	yes	-	-	-	-
BI Bitget Copy Trading	Copy trading	unknown	-	yes	-	-	-	-
BY Bybit Copy Trading	Copy trading	unknown	-	yes	-	-	-	-
OKX Trading Bots	Exchange-native	free	-	yes	-	-	-	-
GA Gate.io Trading Bots	Exchange-native	free	-	yes	-	-	-	-
KuCoin Trading Bots	Exchange-native	free	-	yes	-	-	-	-
Bitget Trading Bots	Exchange-native	free	-	yes	-	-	-	-
BY Bybit Trading Bots	Exchange-native	free	-	yes	-	-	-	-
TR TradeSanta	Cloud platform	$14-70/mo	-	yes	-	-	8	US
BullX	DEX / Telegram sniper	1% per trade	1% per trade	yes	-	-	-	-

Live values from our database - details and caveats in each review. "-" = not documented.

The chart's message is structural, not predictive: funding is a market-mechanics cost, charged on notional per the venue schedule[1], independent of whether your bot wins. At the balanced-market baseline, leverage alone puts a monthly hurdle on the strategy that spot bots never see. Funding spikes in one-sided markets - exactly when leveraged grids are fully loaded - and the short leg of a neutral grid pays it twice.

That is the honest frame for this lane: futures bots are spot bots plus a leverage multiplier on every input - fees, funding, drawdown, liquidation, and the cost of mistakes. The platform comparison matters, but the risk configuration decides survival.

Which platform for which case

Learning leverage safely: Freqtrade - futures support, funding-aware backtesting and dry-run in a free, open-source framework[3]. The only lane where a blown-up hypothesis costs nothing but a weekend.

Futures-native parameter research: Passivbot - its optimizer treats funding and liquidation as first-class constraints; that focus is its value.

Cloud futures automation: 3Commas ($15-105/mo annual) and peers expose long/short and futures bots on major venues[5]. Convenience and UI, at subscription cost, with API-key custody - the no-API-key lane is the counterweight.

Futures grids with local keys: Gunbot ($29/mo, futures strategies, 20 exchanges) and exchange-native futures grids (Binance, Bybit, OKX, KuCoin, Bitget, Gate) where execution is server-side and free of subscription.

Market making on perpetuals: Hummingbot's perp connectors across 318 total connectors[4] - the maker-focused specialist, with the same funding physics as everything else in this section.

Tips that keep the account alive

• Compute the liquidation price before the entry, not during the drawdown. Leverage presets on native bots are marketing numbers; your strategy's worst excursion plus a multiple should fit between entry and liquidation. Isolate the margin so the answer is bounded.
• Track funding as its own line item. Import the venue's funding ledger (OKX exposes fundingFee separately[2]); if your bot's PnL report merges funding into trading PnL, you are flying on instrument colors.
• Paper trade at leverage for a full funding cycle. Dry-run exists in Freqtrade[3] and paper mode in most cloud platforms - a week of simulated funding payments recalibrates leverage appetite better than any article.
• Prefer wide grids on low leverage over dense grids on high leverage. Our grid fee sweep plus the funding chart point the same way: density multiplies costs, and so does leverage. The profitable corner of the matrix is the boring one.
• Size so the liquidation price survives the -76.6% 2021-22 style cycle. BTC drawdown history is the honest stress test; the bot-vs-hodl backtests carry the regime table.

Warnings

• Leverage does not increase edge - it increases ruin probability. The same strategy at 10x faces a 9%/month funding hurdle at baseline rates alone, plus a liquidation price inside any normal volatility band.
• Funding spikes when you are most exposed. The 0.01%/8h baseline is a balanced-market number; crowded longs in a rally (or shorts in a crash) pay multiples of it[1]. Budget the spike, not the baseline.
• Neutral grids are long-volatility trades wearing a hedge costume. Both legs pay funding; a trend liquidates one leg. Collecting the carry is compensation for that tail, not an exemption from it.
• "Fixed" funding income or delta-neutral returns in ads is scam language - the FTC's marker applies verbatim[8]. Real carry strategies disclose the tail risk.
• Liquidations are executions too. In cascades, fills happen at worse than displayed prices; margin ratios computed on last price lie during the only moments that matter.

How the venues differ where it counts

Perpetuals plumbing varies more than marketing admits. Funding mechanics are consistent in principle - longs pay shorts when funding is positive, charged on notional[1] - but the accounting differs: OKX exposes fundingFee as a distinct PnL component you can import[2], while some platforms fold it into realized PnL unless you dig. Liquidation engines differ in maintenance-margin tiers and in whether they use mark price or last price - the difference is exactly which candle liquidates you in a wick. Cross-collateral rules decide whether your altcoin holdings can back a BTC position, and at what haircut.

None of this appears in a feature bullet. It appears in the venue's own docs and fee pages - and in the exchange hubs we maintain per venue. Before automating on a venue, three concrete checks: the funding ledger export (can you audit it?), the liquidation price formula (mark or last?), and the maintenance margin schedule at your position size. The exchange-native futures grids inherit the venue's choices wholesale; cloud and self-hosted bots inherit the same venue physics underneath[5].

Statistics worth the risk model

• 22 platforms in this lane support futures; 45% offer backtesting and 45% offer paper trading - use both before any leverage is real.
• Median entry price across the subscription lanes: ~$21.49/mo/mo - trivial against liquidation risk if you size correctly, decisive if you do not.
• Baseline funding math: 0.01%/8h = 0.03%/day on notional = 0.9%/month at 1x - multiplied by leverage on margin (our figure above)[1][2].
• BTC's documented drawdown history: 22 episodes of -20% or worse since 2017, worst -76.6% (2021-22), running -53% episode open through 2025-26[7]. Every leverage choice should survive the table, not the screenshot.

Reading a futures product page like a risk officer

Strip the screenshots and four numbers should remain: the funding rate and its history (not just the current value[1]), the maintenance margin schedule at your size, the liquidation formula (mark or last price), and the fee tier your volume actually earns. Any platform page - native, cloud or self-hosted docs[3][5] - that hides one of the four is asking you to discover it in a drawdown. The honest venues publish all four; the honest bots surface them in their PnL reporting[2].

One more lens: compare the futures lane against its own spot alternative. Our regime backtests show spot grids losing less than hold in bears without any liquidation exposure[7] - the futures version must beat that baseline by more than its funding hurdle to justify the risk. Run that comparison before the leverage conversation, not after.

The 111-day community test remains the cleanest illustration: a spot BTC/ETH grid over the 2022 bear returned +1.1% while BTC itself fell -16.6%[6]. The same grid with leverage would have amplified both the protection and the exposure - and added a funding bill the spot run never paid. Leverage is a loan against future volatility, repaid whether or not your thesis arrives.

Margin modes, one decision that outlives every trade

Isolated versus cross margin is the single configuration with the largest tail consequence. Isolated caps the blast radius at the position's allocation - a bot with a bad parameter loses the allocation, the rest of the account survives to fix the bug. Cross margin puts the whole shared balance behind every position - deeper survival in chop, total loss in cascades. Automated strategies compound errors and repeat them; a system that cannot bound its own mistakes should not have access to the wallet.

Practical defaults that survive review: isolated margin on every automated position; leverage set so the liquidation price sits outside the strategy's worst historical excursion plus a safety multiple; a manual max-position cap the bot cannot raise; and a funding-aware PnL display you actually read[2]. None of these are platform features to shop for - they are decisions to make before any platform is chosen, and every serious framework in this table can implement all four[3][5].