Grid trading bots

The fee-per-cycle equation that decides everything

Bot	Type	Price	Trade fee	Grid	DCA	Backtest	Paper	Exchanges	US
Gunbot	Self-hosted	$29/mo	-	yes	yes	yes	yes	20	US
Bitsgap	Cloud platform	$23-149/mo	-	yes	yes	yes	yes	17	US
OctoBot	Self-hosted	$0-29.99/mo	-	yes	yes	yes	yes	12	US
PI Pionex	Exchange-native	0.05% per trade	0.05% per trade	yes	yes	-	-	-	US
WunderTrading	Cloud platform	$0-100/mo	-	yes	yes	-	yes	18	US
Altrady	Cloud platform	$30-97/mo	-	yes	-	yes	yes	17	US
Passivbot	Self-hosted	free	-	yes	yes	yes	yes	8	US
3Commas	Cloud platform	$15-140/mo	-	yes	yes	yes	yes	9	US
HaasOnline	Cloud platform	$19.99-249.99/mo	-	yes	yes	yes	yes	24	US
MI Mizar	Cloud platform	0.1% of volume	-	yes	yes	yes	-	10	-
Binance Trading Bots	Exchange-native	free	-	yes	yes	-	-	-	-
OKX Trading Bots	Exchange-native	free	-	yes	yes	-	-	-	-
GA Gate.io Trading Bots	Exchange-native	free	-	yes	yes	-	-	-	-
KuCoin Trading Bots	Exchange-native	free	-	yes	yes	-	-	-	-
Bitget Trading Bots	Exchange-native	free	-	yes	yes	-	-	-	-
BY Bybit Trading Bots	Exchange-native	free	-	yes	yes	-	-	-	-
TR TradeSanta	Cloud platform	$14-70/mo	-	yes	yes	-	-	8	US

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

Every provider on this page automates the same primitive: laddered buys below and sells above a range you define, harvesting one step per round trip. What separates a paying grid from a donation is a single ratio - your grid step versus your round-trip cost. At Binance's regular 0.100% per side, a 0.3% spacing nets 0.4% gross per cycle; at Pionex's 0.050%, the same spacing nets 0.9%. Same strategy, same range, more than double the take[1][2].

Our own density sweep on real BTCUSDT daily data ($10,000, 2024 chop window, fees per side on notional) makes the density side of the ratio concrete: at 0.1% per side, 10 well-spaced levels lost 6.3% of realized grid profit to fees, 40 levels lost 20.6%, 160 levels lost 75.9%, and 320 levels lost more than the strategy earned - fees at 149% of realized profit[3]. The 78-vs-1000-grid community test on Pionex found the same shape: fewer, wider levels out-earned dense ones once fees were counted[4].

Across the 17 grid-capable platforms on this page, the trading fee you pay underneath ranges from 0.05% to 0.05% - a spread that, compounded over hundreds of round trips, decides the outcome before any strategy skill enters.

Which platform for which case

Fee-light ranging, no subscription: exchange-native bots win the arithmetic outright. Pionex bills 0.05% per side and runs the grid server-side[1]; Binance, OKX, KuCoin, Gate, Bybit and Bitget include grids with the account at their standard trading fees. The trade-off is scope: only that exchange's pairs, no cross-exchange logic, and - on Binance/Bybit/OKX/KuCoin/Bitget/Gate - no US access (Pionex runs a separate US entity).

Cross-exchange grids with a test loop: Bitsgap carries a 7-day trial plus paper trading across 17 venues and prices from $23/mo. 3Commas ($15-105/mo) and TradeSanta ($14-70/mo, 14-day trial) cover the same cloud pattern; Altrady ($30-97/mo) adds a 30-day refund. You pay for orchestration, not execution quality - the exchange fee schedule still applies underneath[8].

Control and futures grids: self-hosted Gunbot ($29/mo, 20 exchanges, futures, local keys), OctoBot (open-source core, $0-29.99 cloud), and Passivbot (free, donation-funded, futures-native parameter sets) put the fee equation fully in your hands - including the ability to route grids through lower-fee venues or BNB discounts.

Config knobs that matter more than the platform

• Level count vs step size is the whole game. Our sweep above and the community's 78-vs-1000 test agree[4]: density buys smooth fills and pays fees for the privilege. Start wide (1%+ steps at 0.1% fees), tighten only if round-trip math still clears.
• Arithmetic vs geometric spacing changes which range the grid actually covers. Equal price steps overweight the bottom of a wide range; equal percentage steps keep fee-to-step ratio constant. The Grid Trading 201 community write-up treats this as the first parameter to fix, before level count[5].
• Initial inventory decides whether a grid is market-neutral in fact. Without it, filled buys sit as pure inventory and your PnL is a leveraged bet on the asset. Our backtests ran both; the numbers above assume inventory.
• Stop-loss / range-exit rules are the difference between a capped loss and a full bag. Platforms differ sharply here - cloud tools expose stop lines per bot, exchange-native grids mostly do not, self-hosted frameworks let you code the exit. Decide it before deployment, not after the first red week.
• Funding awareness on futures grids. Funding accrues hourly and lands as a separate PnL component on OKX and peers[6]; a grid that looks profitable on realized steps can bleed it back on funding. If you run futures grids, track funding as its own line item from day one.

Statistics worth knowing before you size

• Fee share of realized profit by density (our sweep, 0.1%/side): 10 levels 6.3% - 40 levels 20.6% - 160 levels 75.9% - 320 levels 149%[3].
• Regime spread (40 levels, 0.1%/side): chop 2024 +1.2% while hodl fell -9.9%; post-crash range 2021 +48.5% vs hodl +80.0%; bull 2023Q4 +99.4% vs hodl +164.4%; bear 2022 -52.0% vs hodl -64.2%; the 2025-26 bear -33.0% vs hodl -42.1%[3].
• Median entry price across this page's grid lane: ~$21.49/mo/mo for subscription lanes - against a $0 exchange-native baseline that has to be beaten by realized grid profit after fees.
• Oldest operator still listed in this lane: 2013 - grid tooling predates most of the marketing around it.

Tips that actually move the needle

• Compute fee-per-cycle before level count. The grid fee calculator does the round-trip math; if your step clears less than 3x the round-trip cost, widen the grid.
• Size the range on the regime you expect, survive the one you get. Our bot-vs-hodl backtests show range exits dominate every other cost over a full cycle.
• Let the fee tier pick the venue. BNB discount moves Binance from 0.100% to 0.075%[2] - on a dense grid that is worth more than any platform feature.
• Paper trade one full oscillation before committing (Bitsgap, Gunbot, OctoBot and HaasOnline all have it). A grid that never completed a round trip in backtest will not complete one live.
• Keep inventory math honest: with initial inventory, half your capital is the asset. A "flat" grid PnL of +1.2% on our chop window hides that BTC itself fell 9.9% over the same weeks[3].
• Split capital across regimes, not platforms. A blend (70% hold, 30% grid) on our windows captured most of the trend upside while keeping the grid's chop income - the boring allocation beats the optimized single grid.

Warnings

• Grid bots are the favorite costume of fake platforms. The FTC's marker stands: only scammers promise fixed payouts - and "fixed daily grid payouts" is exactly that promise[9]. The scam pattern guide lists the FCA's bot-branded warning entries.
• Futures grids add funding on top of fees. Funding accrues hourly and is tracked as separate position PnL on the major venues[6]; a leveraged neutral grid pays funding on both legs - the drag scales with leverage even when PnL looks flat.
• Dense grids can be negative before you start. Our 320-level run paid $227 in fees to earn $72[3]. If your platform shows projected profit but not projected fees, that number is marketing.
• The bag is a capital lock-in immediately. Downside breakdowns tie up your full inventory below the range - size so a full-fill is survivable, not optimal.
• Backtests are daily-kline models. Our sweep uses daily lows/highs as conservative fills; intraday path, slippage and partial fills will differ. Treat every number here as a model with stated assumptions and daily-kline fills, never as a promise - real fills land between the model and your hopes.
• Subscription pricing changes. The cloud lanes on this page repriced several times in 2025-26 (plan reshuffles, venue changes). Verify the current schedule on the provider's own pricing page before committing - the reviews carry dated plan tables.