Choosing Your Marks And Price Sources

A stock is quoted $49.95 bid, $50.05 offer, last trade $50.00. You are long. Which price is your position worth? All three are defensible, all three are "the market price" in some sense, and all three give a different daily P&L number if used consistently over months — the choice is not a rounding error, it is a policy decision that compounds.

The honest answer depends on which side you would trade

If you had to sell this position right now to raise cash, you would hit the bid — $49.95. If you had to buy more right now, you would pay the offer — $50.05. The midpoint, $50.00, is what a market maker quotes as their fair estimate of value, but it is not a price either side of the market actually transacts at without paying the spread. Marking a long position at the offer, or a short position at the bid, systematically overstates what you could actually realize by closing — it marks you at a price only a counterparty on the other side of the trade would get.

The standard, conservative convention most books use is: mark longs at the bid, shorts at the offer, or mark everything at the mid and separately track the spread cost as an explicit "cost to liquidate" line rather than pretending it does not exist. What is not defensible is marking longs at the offer and shorts at the bid — the combination that always makes the book look better than it could actually close out at.

Worked example

A book holds $5m long and $5m short in two different, moderately liquid names, each quoted with a 20bp bid-offer spread.

Position	Mid mark	Bid-side mark (conservative for the short)	Offer-side mark (conservative for the long)
$5m long	$5,000,000	—	$5,005,000 (favorable, wrong direction)
$5m short	-$5,000,000	-$4,995,000 (favorable, wrong direction)	—

If the desk marks the long at the offer and the short at the bid — the flattering combination — the book shows an extra $10,000 of value that does not exist: neither position could actually be closed at those prices without crossing the spread. Marked correctly (long at bid, short at offer, or both at mid with the spread cost disclosed separately), that $10,000 disappears from the reported number, because it was never really there.

Over a book with hundreds of positions and a policy that consistently leans toward the flattering side, this effect adds up to a P&L number that looks smoother and better than the book that would actually result from liquidating everything — precisely the gap that shows up as a nasty, unexplained markdown the day the positions are actually closed.

Where this bites hardest

The distortion is largest exactly where it is easiest to hide: wide-spread, less liquid names, where the gap between bid, mid and offer is real money rather than a rounding difference, and where fewer eyes are watching the mark day to day. The fix is a written, consistently-applied marking policy — not a per-position judgment call — reviewed by someone other than the position holder, for the same reason model marks need independent review: the incentive to lean flattering is always present, and it is invisible until the position actually trades.