Mechanism
Spreads widen when a market maker expects one of these:
- Adverse selection: you might be trading on fresher information.
- Inventory risk: holding the other side could be painful if price gaps.
- Hedging friction: they can’t offload risk cheaply in the underlying venues.
Widening is not “evil.” It’s a live safety valve.
Microstructure note: stops fail most often at the same time liquidity disappears. That is not bad luck; it is structural. Your job is to avoid competing for fills in the worst queue.
- Prefer “don’t trade” windows over cleverness: rollover, open/close, data prints.
- Reduce size before you reduce stop distance. Size is the only lever that always works.
- Measure slippage by regime, not by average.
How it kills accounts
Spread blowout → market order anyway → slippage + worse entry → tighter stop “to compensate” → noise stop-out → re-entry → cost bleed.
How it kills accounts:
- Edge looks fine in backtest.
- Live spreads widen at the exact wrong moments.
- Stops trigger inside noise, so you widen stops.
- Same size + wider stop = silent leverage increase.
- A normal spike becomes structural damage.
Rule that survives
- Create a spread gate: if spread > normal by X%, reduce size or wait.
- Prefer limits in thin conditions; don’t donate optionality.
- If you must trade: smaller size + wider stop + fewer attempts.
Rule that survives:
- Spread is a gate, not a footnote. If it’s abnormal, you don’t trade or you trade smaller.
- Assume worst-case fills in fast markets.
- Size is the adapter: reduce size before changing the stop model.
Example archetype
You hit London open with normal size. Spread triples for 20 seconds. You market in, get slipped, tighten the stop, get clipped, then chase. The loss is mostly execution, not “wrong idea.”
Tell: if the trade only works when the spread is tight and price is smooth, it’s not an edge, it’s a regime bet.
Deep dive
Deep dive
Spreads are not “the market.” They’re the price of getting filled right now when someone else would rather wait.
In quiet conditions, liquidity providers can quote tightly because adverse selection is low. When uncertainty spikes (news, breaks, thin sessions), they protect themselves by widening or pulling quotes.
Practical implication: a spread blowout is often the earliest, cleanest “regime change” indicator you’ll get. It tells you execution risk just jumped, even if the chart looks normal.
Glossary pointers: last look, markout, liquidity vacuum.
Field checklist
- Measure spread before entering. If it’s abnormal, you’re trading the wrong product at the wrong time.
- If volatility expands, reduce size first. Don’t “solve” it by widening stops with the same size.
- Avoid the predictable liquidity holes: rollover, session open/close, first minutes after data.
- Assume your stop may fill worse than your entry. Price the worst-case, not the brochure.
- If you cannot explain where liquidity comes from, trade smaller.