Insurance Orders in Crypto: DCA Safety Nets Explained
Insurance orders — also called safety or DCA orders — are pre-placed limit orders below your entry that add to the position if price moves against you, improving the average entry. They are safe only when ladder depth, step size and multiplier are fixed in advance and total exposure at full depth is computed before entry.
An averaging ladder converts a single-entry trade into a distributed one. Done with fixed depth and a computed worst case, it meaningfully improves the hit rate of range-bound setups. Done without those, it is the mechanism behind most account-ending losses in automated crypto trading. The difference is entirely in four parameters.
- The three parameters that define a ladder: depth, price step, and size multiplier.
- Compute total committed exposure and the liquidation price at full depth before the first entry.
- A ladder without a stop below the final rung is an unbounded loss with extra steps.
- Insurance orders and stop-losses solve opposite problems and you need both.
- Exchange notional floors silently break small ladders — some rungs place, some do not.
What are insurance orders?
Insurance orders are limit orders placed below your entry (for a long) at the moment the position opens. If price falls to a rung, that order fills and adds to the position, lowering the average entry price. The trade then needs a smaller bounce to reach break-even.
The names vary — safety orders, DCA orders, averaging orders, grid entries — and refer to the same mechanism.
Concretely: enter 1 BTC at $50,000 with rungs at $49,000 and $48,000, each 1 BTC. If both fill, you hold 3 BTC at an average of $49,000. Price needs to recover to $49,000 rather than $50,000 for break-even — a smaller move.
What you actually traded
The trade-off is exact and worth stating plainly: you increased your probability of a profitable exit and increased the size of the loss when it does not happen. Expectancy has not improved by itself. What changed is the shape of the outcome distribution — more small wins, fewer but larger losses.
That is a legitimate trade in the right context. It is not a free improvement, and treating it as one is where the trouble starts.
The three parameters
Depth — how many rungs
Directly sets your maximum exposure. Three to five is a normal range. Depth must be a fixed number decided before entry; a ladder that deepens because price kept falling is not a strategy, it is capitulation with extra steps.
Step — spacing between rungs
Fixed percentage steps are the simplest. Widening steps (each further apart than the last) put more capital at deeper prices and survive larger moves. Volatility-scaled steps — spacing as a multiple of ATR — adapt across regimes and are usually the better default, because a fixed 1% step is tight in a volatile week and wide in a quiet one.
Multiplier — size progression
Equal sizing (1x) keeps exposure linear and predictable. Mild progression (1.2x–1.5x) improves the average entry more per rung while keeping growth manageable. A 2x multiplier is martingale, with the exponential growth that implies — see why that breaks accounts.
| Multiplier | Rung sizes | Total at 4 rungs | Character |
|---|---|---|---|
| 1.0x | 1, 1, 1, 1 | 4.0 | Linear, predictable |
| 1.3x | 1, 1.3, 1.7, 2.2 | 6.2 | Balanced |
| 1.5x | 1, 1.5, 2.3, 3.4 | 8.2 | Aggressive |
| 2.0x | 1, 2, 4, 8 | 15.0 | Martingale |
The calculation to do before entry
Every ladder needs these four numbers computed before the first order, not discovered later:
- Total committed capital at full depth. Sum every rung. This is your real position size — the initial entry is just the first instalment.
- Average entry at full depth. Size-weighted across all rungs. This is your actual break-even.
- Liquidation price at full depth. The number that matters most, and the one most often skipped. Each fill moves liquidation closer to spot. Compute where it lands when every rung has filled.
- Loss if the stop below the final rung is hit. This is your real maximum loss, and it must be a survivable number.
If any of these is uncomfortable, the base size is too large. Reduce it — do not reduce the depth of the ladder you already decided you needed.
Insurance orders versus stop-losses
These are opposites and the confusion between them is expensive:
| Insurance order | Stop-loss | |
|---|---|---|
| Direction | Increases position | Closes position |
| Assumes | Price will revert | The thesis is wrong |
| Effect on risk | Increases | Bounds |
| Effect on liquidation | Moves it closer | No effect |
You need both. Insurance orders handle the case where price dips and recovers. The stop handles the case where it does not. A ladder without a stop is an unbounded loss with intermediate steps, and it is precisely the configuration that produces the account-ending trade.
Placement: the stop goes below the final rung, far enough that ordinary noise around that level does not trigger it, and sized so that reaching it is survivable at full ladder exposure.
The notional floor problem
An implementation detail with real consequences. Exchanges enforce minimum order sizes — Hyperliquid rejects reduce-only limit orders below roughly $10 of notional.
For a laddered position this bites in a specific way. A small position split across several rungs can produce individual rungs below the floor. Some place successfully, some are rejected, and the resulting exposure does not match the configuration. The same applies on the exit side: a multi-leg take-profit ladder against a small position can have its later legs silently fail.
Two defences: validate every rung's notional at configuration time and refuse a configuration whose legs fall below the venue floor, and verify after placement that the number of live orders matches the number configured. See setting up on Hyperliquid for the venue specifics.
When ladders do not work
Strong trends. Averaging assumes reversion. In a sustained one-way move every rung fills and none recover. This is the dominant failure case and it is not rare.
Illiquid instruments. Wide spreads make each fill worse than modelled, and the exit at recovery is also worse.
Structural breaks. When a level fails for a reason — a hack, a delisting, an enforcement action — averaging into it is adding size to a thesis that has been invalidated by information, not by noise.
High leverage. Ladder fills move liquidation toward spot. At high leverage the ladder reaches liquidation before it reaches its own stop, which means the stop can never fire.
Correlated ladders. Three laddered positions in correlated assets during a broad move fill together, and the combined drawdown is the sum.
A conservative default
- Depth: 3 rungs.
- Step: volatility-scaled, roughly 1 ATR apart.
- Multiplier: 1.0x–1.3x.
- Stop: below rung 3, sized so full-depth loss is 1–2% of equity.
- Leverage: low enough that liquidation at full depth sits far beyond the stop.
- Total exposure: computed and written down before entering.
TradeFloor exposes depth, step, step multiplier and size multiplier explicitly, and computes total committed exposure and the liquidation price at full ladder depth before a configuration can be saved — because the failure here is rarely a bad trade, it is a configuration whose worst case was never calculated.
Frequently asked questions
What are insurance orders in crypto trading?
Limit orders placed below your entry when the position opens, which add to the position if price moves against you and improve the average entry. Also called safety orders, DCA orders or averaging orders. They raise the probability of a profitable exit and raise the size of the loss when there isn't one.
How many insurance orders should I use?
Three to five rungs is a normal range, and the number must be fixed before entry. What matters more than the count is that total committed capital, average entry, liquidation price at full depth and the loss at the stop are all computed beforehand. If any is uncomfortable, reduce base size rather than depth.
Do I still need a stop-loss if I use insurance orders?
Yes, and this is the most important point. They solve opposite problems: insurance orders handle a dip that recovers, a stop handles a thesis that was wrong. A ladder without a stop below the final rung is an unbounded loss with intermediate steps, which is exactly the configuration behind most account-ending trades.
What size multiplier is safe for a DCA ladder?
1.0x to 1.3x keeps exposure predictable while still improving the average entry. 1.5x is aggressive. 2.0x is martingale, where four rungs already commit fifteen times the base size and growth is exponential. The multiplier is the parameter that decides whether the worst case is a number you can state.
Why did some of my ladder orders not place?
Almost certainly the exchange minimum notional. Hyperliquid rejects reduce-only limit orders below roughly $10, so splitting a small position across several rungs can put individual legs under the floor — some place, some fail, and actual exposure no longer matches the configuration. Validate every leg's notional before submitting.
When should I not use an averaging ladder?
In strong trends, where every rung fills and none recover — the dominant failure case. Also in illiquid instruments, after a structural break where the level failed for a reason rather than from noise, at leverage high enough that full ladder depth reaches liquidation before the stop, and across several correlated positions that will fill together.
Martingale in Crypto: Why It Works Until It Destroys You
The arithmetic of exponential sizing, and why leverage terminates the sequence before it can recover.
// questions or corrections · [email protected] · more essays · /blog