The math of liquidation
Liquidation is the price at which your losses would consume your collateral. The exchange closes you there to avoid being on the hook for further losses if the market keeps moving against you. Once you understand the math, you can decide how much room to give yourself. If you're new to Bitcoin perpetuals and want the instrument explained from first principles, start with the Bitcoin perpetual futures guide.
Worked example: $1,000 at 10× leverage. You long BTC at $90,000 with $1,000 collateral and 10× leverage. Position notional: $10,000 (about 0.111 BTC). Your liquidation price is roughly 10% below entry — so around $81,000 — minus maintenance margin and fees, so realistically around $81,500.
Same $1,000 at 25× leverage. Notional $25,000 (about 0.278 BTC). Liquidation is roughly 4% below entry — around $86,400.
Same $1,000 at 100× leverage. Notional $100,000 (1.111 BTC). Liquidation is roughly 1% below entry — around $89,100. A normal one-minute wick on a quiet day is enough to liquidate you.
These numbers approximate — exact liquidation prices depend on fees and maintenance margin, which vary by exchange and position size. But the order of magnitude is right and is what matters for sizing decisions. Higher leverage = closer liquidation = smaller acceptable adverse move before you're done.
Isolated vs cross margin
Isolated margin assigns a fixed collateral amount to a specific position. If that position liquidates, only that amount is lost — the rest of your account is untouched. This is the safer default and the one most professionals use for tactical trades.
Cross margin pools your entire account balance as collateral for all open positions. This gives any individual position more buffer against liquidation (because it can pull from the rest of the account), but the downside is that a single runaway position can drain everything you have.
If you're new to leveraged trading, default to isolated margin on every position. The marginal liquidation-distance benefit of cross margin is not worth the catastrophic-loss risk. Switch to cross only when you have a specific reason — usually a multi-leg strategy where positions hedge each other.
Glimpse's bot uses isolated margin for every trade. The exact maintenance-margin policy and per-trade sizing logic lives on the method page. The API key model that ensures your collateral can't be moved by the bot is explained on the security page.
Position sizing — the rule that prevents most blow-ups
The most important number in any leveraged trading account is maximum loss per trade. Pick a percentage of your account that, if lost on a single position, would not change your strategy or your sleep. For beginners, that's around 1% of account balance per trade. For very conservative traders, 0.5%.
From there, you size every position so that stop-loss-to-entry distance × position size = your max loss. If your stop-loss is 2% below entry, and your max-loss-per-trade is $20 (1% of $2,000 account), your position notional should be $20 / 2% = $1,000. At $90,000 BTC, that's 0.0111 BTC notional, which at 10× leverage requires $100 of collateral. Five trades like this can be open simultaneously without your account being meaningfully at risk.
Most beginners do the opposite — they pick a position size first ("I want to make $500 on this trade") and then back into a leverage and stop-loss that gets them there. That's how you end up with stop-losses at the liquidation price, which means you don't have a stop-loss at all.
Always size from max-loss-per-trade down to position. Not the other way around.
Stop-loss discipline — placed on the exchange, not in your head
A stop-loss is a pre-placed order to close your position if the price hits a level you specify. It exists to do one thing: take you out of a trade at a loss you can survive, rather than at a loss you can't.
The stop-loss must be placed on the exchange, not in your head. Mental stops fail under stress — at the moment the price hits your level, you'll have nine reasons to wait "just a few more candles." Then the market gaps through your level and your loss is twice what you'd planned. Pre-placed exchange-side stops execute regardless of how you feel about them.
The stop-loss must be above your liquidation price, not at or near it. If the stop and the liquidation are the same price, the exchange will liquidate you first (with worse fees) before your stop fires. Set the stop at least a percent or two clear of the liquidation level.
The stop-loss must be informed by the trade's structure, not pulled from a hat. "2% below entry" is a default; "1% below the recent support" or "just past the level the thesis would invalidate" is better. The point of a stop is to close the trade when the thesis is wrong, not to close it at an arbitrary price.
Glimpse's bot places exchange-side stops on every position, calibrated to the structure of the trade and to the bot's max-loss-per-trade rule. See the method page for the calibration logic.
Why "too much leverage" is the universal answer
Almost every Bitcoin futures liquidation story collapses to the same root cause: the trader used more leverage than the move they were trying to capture justified. The thesis might have been correct; the sizing wasn't.
A 3% adverse move is normal on any given Bitcoin day. A 5% adverse move happens every few weeks. A 10% adverse move happens several times a year. If your leverage is high enough that any of these normal moves liquidates you, you're not trading — you're betting that the next normal move won't happen during your position. Eventually it will.
The fix is to lower leverage until your liquidation price is far enough from entry that a normal adverse move can't get there. At 5× leverage, liquidation is around 20% below entry — a comfortable margin even in volatile conditions. At 10× it's around 10%, still survivable for most setups. At 25× and above, you're betting against normal market noise.
There's no thesis good enough to justify 50× or 100× leverage on Bitcoin perpetuals. The exchanges offer those leverage ratios because they're popular, not because they're sound. Just because Bybit lets you set leverage to 100× doesn't mean you should.
Automated risk caps — the second line of defense
A well-built automated trading system has hard caps that activate without your input: maximum leverage per position, maximum loss per trade, maximum daily loss (the account stops trading until tomorrow if today's losses exceed a threshold), and maximum total open exposure across all positions.
These caps don't replace good judgment — they catch the trades where good judgment lapses. Every trader has bad days, tilted afternoons, revenge-trade impulses after a loss. Automated caps don't get tired and don't tilt.
Glimpse's risk caps are baked into the bot itself, not bolted on after. Hard maximum leverage, hard per-trade SL, hard daily-loss halt that pauses copy-trading-side execution. We document the full set on the method page and the security page. The bot also runs continuous monitoring — if a position drifts toward conditions that historically precede stop-outs, exit-pressure logic moves before the stop-loss fires.
If you're trading manually, build your own version of these caps on the exchange's risk-management settings — Bybit and Toobit both offer per-account leverage limits and per-position SL automation. If you're running an automated system like Glimpse, the caps should be visible in the dashboard and not user-overridable for the account-protective ones.
What to do when you're already underwater
If you have an open losing position, you have three choices: close it, hold it, or add to it. Do not add to it.
Adding to a losing position ("averaging down") is the single most common path from "down 5%" to "liquidated." The logic feels right — your average entry gets better, so you need less of a bounce to break even — but the math is brutal: you've just doubled your exposure to a thesis that's already proven wrong, with the same collateral. Your new liquidation price is now closer to current price than before, not further away.
Holding it is acceptable if (a) your original stop-loss is still in place, and (b) you genuinely believe the original thesis is still valid. Most of the time it's neither — the original stop got moved "just a little" to give the trade more room, and the thesis got rationalized post-hoc. If you can't honestly answer yes to both, close.
Closing it at a loss is unpleasant. It's also the action that preserves your ability to trade tomorrow. Losses are not failures — they're a cost of running a trading account. The traders who survive are the ones who accept small losses cleanly and move on. The ones who blow up are the ones who refuse to take a 2% loss and end up taking 100%.
If you find yourself wanting to add leverage *after* a losing trade to recover, stop. That's the impulse that ends accounts. Walk away, come back tomorrow, and re-evaluate when you're not emotionally attached to the previous trade.
Behavior — the part nobody puts in the guides
Most liquidations are not caused by exotic market conditions. They're caused by ordinary market conditions interacting with ordinary human behavior — fear during drawdowns, greed during rallies, and the specific cocktail of "I'm down, I need to recover this" that turns a 1% loss into a 50% loss.
Decide your risk rules when you're calm, not when you're in a trade. Write down: maximum leverage, maximum loss per trade, maximum daily loss, the loss level at which you stop trading for the week. Tape it to the wall if you have to. When the trade is going against you, you'll be tempted to revise the rules. Don't.
Don't trade when emotionally charged. Big news, bad day at work, relationship stress, sleep deprivation — all of these correlate with worse decisions. The market will be there tomorrow.
Track your trades honestly. Write down every trade: entry, exit, thesis, what you got right, what you got wrong. After 50 trades, patterns emerge. After 200, the patterns are obvious. The traders who improve are the ones who keep the journal.
Glimpse takes the emotional layer off the trader. The bot doesn't tilt, doesn't revenge-trade, doesn't move stop-losses on hope. The track record shows the cost of that discipline — losses are taken cleanly, gains are taken without flinching. It's not magic; it's just consistency that's hard for humans to sustain at 3am. If you want to put that discipline to work on your account, see pricing for what each tier covers.