The Last 14 Minutes: Why end-of-day option mids can lie — and what traders should do instead

# The problem in plain terms Options end‑of‑day prices in most data files are the midpoint between the last bid and ask. Around 3:46pm and 4:00pm — the last 14 minutes — those mids stop behaving like reliable prices. Market makers widen quotes as they go flat overnight or brace for macro news, so mids can swing wildly even when the trade itself didn't change.

# Two real examples that explain how bad it can get

• Example B (ADBE, Aug 10): The stock drifted half a point in the final minutes, so the closing snapshot's ATM strike moved up one strike. The closing quotes on that new ATM were wider than $4 and pushed the calendar's measured value down to $0.10, while the pre‑close market had priced it at $0.40. The RV printed 0.10% at close versus 0.50% pre‑close.

# Why it happens Market makers must carry positions overnight. As the bell approaches, they reduce daytime risk by widening quotes. Attention shifts to the underlying's closing auction, so option quotes become sleepier: spreads widen, some quotes go stale, a few cross or collapse. Because the midpoint depends directly on those bid/ask quotes, your derived measures (debit, spread cost, relative volatility, fills in backtests) inherit the noise.

A second mechanism magnifies the effect: ATM is defined by the underlying price at the snapshot moment. A tiny late move can change which strike is measured. That new anchor may naturally be thinner or fatter structurally, and if it occurs exactly when quotes are widest, the combined effect can be extreme.

# When it's worst

# Practical consequences for traders and researchers

• Backtests that use the 4:00pm midpoint can overstate costs or understate position values on about 1 in 18 entry days (5.6%), per the re‑verification cited.
• Screeners that flag "expensive" options by comparing to close mids can produce false signals in either direction.
• Relative volatility or spread‑cost ratios that rely on close mids can be biased just when you most care about clarity (before macro prints or earnings).

# What to do instead

• Check quote widths before trusting a midpoint. If the front leg spread is several dollars, treat the midpoint as suspect.
• When backtesting, simulate fills using pre‑close market conditions rather than the 4:00pm mid, or flag days adjacent to major macro events for special handling.

Note: the SteadyOptions app removed its "4pm compare" feature because of these dynamics.

# Bottom line A closing midpoint is a legitimate snapshot of a market that is often half‑asleep. If you act on it without verifying bid/ask widths or considering ATM drift, you can be misled in either direction. Use pre‑close sampling or explicit width checks to make end‑of‑day option measures more reliable.