Futures trading bot guide: leverage, funding and liquidation control

What futures actually let a bot do

A spot bot can only buy an asset it can afford and sell what it holds. A futures bot trades contracts whose value tracks an underlying price, and in crypto the dominant instrument is the perpetual swap — a futures contract with no expiry date. Two capabilities make this attractive for automation. First, the bot can go short as easily as it goes long, so it can try to profit when a market falls, not just when it rises. Second, it can apply leverage, controlling a position larger than the cash it has posted as margin.

For a strategy bot, that flexibility is genuinely useful: a mean-reversion or trend rule that works in both directions can be expressed symmetrically. But every one of those advantages is paired with a hazard that does not exist on spot, and the entire job of building a futures bot well is managing that second column honestly.

Step 1 — Understand that leverage cuts both ways

Leverage is often sold as a way to amplify gains. It is more accurately a multiplier on both directions of price movement. At 10x leverage, a 1% favourable move is roughly a 10% gain on your margin — and a 1% adverse move is roughly a 10% loss on your margin. The exchange does not care which way it goes; the maths is symmetric. What is not symmetric is the consequence at the extremes, because losses can hit a hard floor that gains never do: liquidation.

This is the single biggest difference from spot trading, and it changes how a bot must be designed. On spot you can hold through a drawdown and wait. On leveraged futures, a deep enough drawdown doesn't just hurt — it ends the position entirely, on the exchange's terms, often at the worst possible moment.

Step 2 — How liquidation actually works

When you open a leveraged position, you post margin — a deposit that backs the much larger contract value. As the price moves against you, your unrealised loss eats into that margin. The exchange continuously tracks how much margin remains versus the maintenance margin it requires you to keep. The moment your equity on the position falls to that maintenance threshold, the exchange force-closes (liquidates) the position to protect itself. You don't get to wait for a recovery: the position is gone, and in the typical case you lose the whole margin committed to it (sometimes a sliver is returned, often nothing, and fees make it worse).

Two numbers govern this. Initial margin is what you must post to open the position — at 10x leverage, that's roughly 10% of the contract's value. Maintenance margin is the smaller buffer you must keep to stay open. The gap between them is your room to be wrong. Higher leverage means a smaller initial margin, which means a smaller buffer, which means the price needs to move only a little against you before that buffer is gone.

Step 3 — Isolated vs cross margin

Exchanges offer two margin modes, and the choice materially changes a bot's blast radius. With isolated margin, only the margin you assign to a position can be lost — if that trade liquidates, the damage is capped at that allocation and the rest of your account is untouched. With cross margin, your entire account balance acts as shared collateral; a losing position can pull from your whole balance to avoid liquidation, which delays the force-close but means one bad trade can endanger everything.

For most bots, isolated margin is the safer default. It turns each position into a contained experiment with a known maximum loss, which is exactly the property you want when code — not a human watching the screen — is deciding when to enter. Cross margin has its uses for hedged or portfolio strategies, but it removes the clean per-trade loss cap that makes a bot's risk easy to reason about.

Step 4 — The funding-rate drain on perpetuals

Perpetual swaps never expire, so exchanges use a funding rate to keep the contract price tethered to the underlying spot price. At regular intervals — commonly every eight hours — a payment is exchanged directly between traders: when the rate is positive, longs pay shorts; when it's negative, shorts pay longs. It's usually a fraction of a percent, but it compounds. A position held for days through several funding windows can quietly bleed a meaningful amount, and because the payment is on the leveraged notional, the drag relative to your margin is larger than it looks.

For a bot, funding is not a footnote — it's a line item. A backtest that ignores funding will overstate the returns of any strategy that holds positions across funding times, and a live bot that ignores it will watch a "winning" trade slowly turn flat. Model funding explicitly, the same way you model fees and slippage. We break down the full cost stack in crypto trading bot fees explained.

Step 5 — Size on risk, never on leverage

The most common and most expensive mistake in futures botting is treating leverage as a position-size dial: "the exchange lets me use 20x, so I'll use 20x." Leverage is not a strategy decision — it's a margin-efficiency setting. Position size should be driven by how much you're willing to lose if your stop is hit, divided by the distance to that stop. The leverage you end up needing is just whatever falls out of that calculation, and it should be the output, not the input.

Concretely: decide that any single trade may risk, say, 1% of equity. Measure the distance from your entry to your stop-loss. Size the position so that hitting the stop costs exactly that 1% — no more, regardless of what leverage the exchange offers. Done this way, a tighter stop means a larger position and a wider stop means a smaller one, and your account risk stays constant. This is the same discipline we cover for spot bots in position sizing for trading bots, applied to leveraged contracts where getting it wrong is far less forgiving.

Leverage is a consequence of your stop distance and risk budget, never the starting point. If your sizing logic begins with "max leverage," your bot is sized to be liquidated, not to survive.

Step 6 — A rough liquidation-distance sketch

Before you size any futures trade, it helps to know how much room you have to your liquidation price and to size the position from risk rather than leverage. This sketch shows the intuition — a rough adverse-move estimate from leverage and a risk-based size — not exchange-exact maths, which varies by tier and fees.

python · futures_sizing.pydef approx_liq_move(leverage, maint_margin=0.005):
    # Rough % adverse move that wipes the margin (ignoring fees/funding)
    return (1 / leverage) - maint_margin

def position_size(equity, risk_pct, entry, stop):
    # Size from RISK and stop distance, not from leverage
    risk_cash = equity * risk_pct                 # e.g. 1% of equity
    stop_dist = abs(entry - stop) / entry        # fractional move to stop
    notional  = risk_cash / stop_dist             # contract value to trade
    return notional

eq, entry, stop = 5000, 60000, 58800     # 2% stop on BTC perp
notional = position_size(eq, 0.01, entry, stop)
lev_used = notional / eq                          # leverage falls OUT of the maths
print(f'notional ${notional:,.0f}  ~{lev_used:.1f}x  liq if >{approx_liq_move(lev_used)*100:.1f}% adverse')

Run that and the leverage is whatever the trade requires — you never typed in a leverage number. The bot then checks that its stop sits comfortably inside the liquidation distance, so the stop-loss fires long before the exchange's force-close ever could. If the stop is wider than the liquidation buffer, the trade is too big and must be cut.

Step 7 — The safeguards a futures bot can't skip

Everything above turns into a short, non-negotiable checklist for the bot's risk layer. Build these before you build any profit logic, because on leveraged futures the trade that wipes you out arrives faster than on spot:

• A hard stop-loss on every position, set inside the liquidation distance — never rely on liquidation as your exit.
• Conservative leverage as an output of risk-based sizing, not a dial set to the exchange maximum.
• Isolated margin so a single bad trade is capped and can't drain the whole account.
• A maximum position cap and a max-concurrent-positions limit, so the bot can't pyramid itself into a corner.
• Continuous monitoring of liquidation distance — alert or de-risk automatically when price drifts toward the liquidation price.
• Funding-aware PnL so held positions account for the recurring drain, and a daily-loss kill switch that halts trading entirely.

None of this is optional flair. A spot bot with weak risk controls loses slowly; a futures bot with weak risk controls loses suddenly and completely. The same risk-first philosophy we apply across the site — detailed in trading bot risk management — simply matters more here, where leverage compresses the time between a mistake and its consequence. Use the position-size calculator to pressure-test your numbers before any of this touches real margin.

Frequently asked questions