A 7.94% Sunday hour on below-average volume forces a market maker to re-price its quote size
Should the desk shrink its quoted size and widen its spreads for the whole week, or only during the late UTC hours when the week's largest hourly move occurred?
The market facts are real; the decision maker is illustrative.
The week and the position
In the week of 18 June 2018, ETH fell 8.60% from 497.15 to 454.41, with a 30.67% range between 548.20 and 419.53 and 100.00% annualised realised volatility of price movement. The worst day fell 12.20% on Friday, and the largest one-hour move, 7.94%, landed at 16:00 UTC on Sunday, late in the UTC day, one hour after the weekly low. Dollar volume of 0.49bn was 0.9 times the prior 12-week average, three of the four prior weeks were down, and ETH trailed BTC by 3.83 percentage points despite a 0.92 hourly correlation. An illustrative market-making desk quotes both sides and must keep inventory small.
Judgment
The week's loss is less important to this desk than the shape of its risk: a 7.94% single hour arriving late in the UTC day, when volume is thin and inventory from the day's two-sided quotes cannot be unwound at the quoted price. Cut quoted size and widen spreads during the late UTC hours, and price the spread from the 7.94% hour rather than from the 30.67% weekly range, which understates what happens inside one hour. Accept lower turnover in exchange for inventory that survives a late-Sunday gap.
What would change it
If several consecutive weeks show largest one-hour moves well below 7.94% and dollar volume returns above the prior 12-week average, the late-hour size cut can be lifted. If the week's 18 weeks of history grow enough to place the 100.00% realised volatility in a stable percentile, spreads can be set from history rather than from this week.
What sets this case apart
The decision is about the intraday timing of one hourly move against thin volume, not about the direction of the week's 8.60% decline.