Spread-Widening in Uncertainty and Volatility
Spreads are not constant. They widen when uncertainty rises, and understanding why is one of the clearest windows into how market makers think.
Two different risks, both rising
Inventory risk rises with volatility. Once filled, you hold a position until you can offset it. The potential loss over that holding period scales with volatility, roughly as using the square-root-of-time rule. Double the volatility and you need roughly double the compensation for the same holding period.
Adverse selection rises with uncertainty. When news is imminent or arriving, the chance that whoever trades with you knows something increases sharply. Your quotes go stale faster, and the traders quickest to act on new information are precisely the ones hitting you.
These are distinct. The first says a fill is more dangerous; the second says a fill is more likely to be against you. Both push the same direction.
Volatility raises the cost of holding a position; uncertainty raises the chance the person trading with you knows more. Spreads widen for both reasons, and around scheduled news the second dominates.
The awkward consequence
Market makers widen exactly when everyone most wants to trade. During a shock, spreads blow out, depth thins, and the cost of exiting rises sharply, all at the moment participants are most desperate to reduce risk.
This looks like liquidity providers abandoning the market when needed. From the market maker's side it is the opposite: continuing to quote at normal spreads during a crisis would be quoting at a loss, and a firm that did so would not survive to provide liquidity afterwards.
The practical lesson for anyone holding a position: liquidity is not a stable property. A position that can be exited cheaply in calm conditions may cost many times more to exit during stress, and any risk assessment that ignores this understates the true cost.
Scheduled versus unscheduled
The two cases are handled differently.
Scheduled events (earnings, central bank decisions, economic data at a known time) can be prepared for. Market makers widen in advance, reduce size, or pull quotes entirely for the seconds around the release. Nobody wants to be the resting quote when a number prints.
Unscheduled events (a surprise headline, a large unexplained trade) must be reacted to, which is where latency earns its cost. The faster you can cancel, the less you lose on quotes that are already wrong.
The tradeoff being made
Widening reduces fill rate and raises profit per fill. Whether that improves expected P&L depends on how the two move relative to each other:
In stable conditions, tight quotes maximise this because volume dominates. In volatile conditions the profit per fill must rise faster than the fill rate falls, or you are simply trading more at worse terms. Getting this balance right, continuously and under pressure, is the core tradeoff of the job.
"Volatility just doubled, what do you do with your quotes?" is a standard interview question. Widen, reduce size, and check your inventory first, since an existing position is now roughly twice as risky as it was a moment ago.