After the Fill: The Gap Nobody Sees

The trading lessons in this catalogue end at the fill. In practice the fill is roughly the midpoint of the process, and the half that follows is where the money and the securities actually move.

When an order executes, nothing has changed hands. What exists is an obligation: one side will deliver securities, the other will deliver cash, at a specified future date. Until that happens, the buyer has a claim rather than an asset, and the seller has a claim rather than cash.

The gap has a name at each end. Trade date is when the agreement is struck. Settlement date is when delivery occurs. The interval between them, currently one business day in the United States and two in most of Europe, is the settlement cycle.

That interval is where an entire industry lives, because during it both parties are exposed to something that has nothing to do with the price: the possibility that the other side does not perform.

"The other side fails" covers three distinct risks with three different sizes, and conflating them makes the problem look worse than it is.

Replacement cost risk is the realistic one. If the counterparty defaults before settlement, you do not lose the trade's full value; you lose the difference between the agreed price and the price at which you can redo the trade now. Agree to buy at 100, the counterparty fails, and the market is at 103: your loss is 3, not 100.

Principal risk is the catastrophic one. It arises only if you deliver your side and the counterparty does not deliver theirs. Then you have lost the entire value. This is what settlement mechanics are designed to eliminate outright.

Liquidity risk is the mundane one. The counterparty performs, but late. Nothing is lost permanently, and you may have to borrow to cover obligations that assumed the delivery arrived on time.

The distinction matters because it sizes the problem. Replacement cost is proportional to price movement over the gap, which is small and manageable. Principal risk is proportional to notional, and must be engineered away rather than managed.

If the gap causes the risk, an obvious question follows: why not settle instantly?

Several things have to happen between agreement and delivery, and each takes real time.

Confirmation and allocation. An asset manager trading for forty funds executes one block and must then allocate it across accounts, each with its own custodian and instructions.

Funding. The buyer must have the cash in the right currency at the right place. A foreign investor may need a currency transaction that itself settles on a cycle.

Securing the securities. The seller must have the securities available. If they are on loan, they must be recalled.

Netting. This is the substantial one. Over a day a participant may trade the same security hundreds of times. Settling each trade individually means thousands of movements; netting them to a single net figure per security is enormously cheaper, and netting requires a window over which to accumulate.

So the cycle is a deliberate trade: a shorter gap means less counterparty risk and less time to arrange everything, and the industry has spent decades moving that dial.

The saving from netting is easy to underestimate, so it is worth computing.

def netting(trades):
    """trades: (security, side, qty, price). Returns gross vs net movements."""
    gross_moves = len(trades)
    net = {}
    for sec, side, qty, px in trades:
        signed = qty if side == "buy" else -qty
        s = net.setdefault(sec, {"qty": 0, "cash": 0.0})
        s["qty"] += signed
        s["cash"] -= signed * px
    net_moves = sum(1 for s in net.values() if s["qty"] != 0)
    return gross_moves, net_moves, net

day = [("ABC", "buy", 500, 10.0), ("ABC", "sell", 300, 10.2),
       ("ABC", "buy", 200, 10.1), ("ABC", "sell", 400, 10.3),
       ("XYZ", "buy", 100, 50.0), ("XYZ", "sell", 100, 50.5)]

netting(day)
# (6, 0, {'ABC': {'qty': 0, 'cash': 160.0}, 'XYZ': {'qty': 0, 'cash': 50.0}})

Six trades, and no securities move at all: both names net flat. Only 210 of cash changes hands, against 24,250 of gross traded value.

At market scale this is the difference between a system that works and one that does not. Netting typically eliminates the large majority of settlement obligations, and it is the main reason the settlement cycle is not zero.

Two distinct functions sit between execution and delivery, and they are routinely confused.

Clearing is the determination of who owes what. It takes the raw trades, confirms them, nets them, calculates obligations, and manages the risk of non-performance during the gap. Nothing moves.

Settlement is the actual exchange of securities for cash. It happens once, at the end.

The risk timeline maps onto that split. Replacement cost risk exists across the whole interval, growing with price volatility and shrinking as the cycle shortens. Principal risk exists only at the instant of exchange, and only if the two legs can come apart.

Separating them explains the architecture of the whole industry. One institution addresses the interval, and a different mechanism addresses the instant. The next lessons take them in that order.

Now the manoeuvre that makes anonymous trading possible.

On an electronic exchange you do not know who took the other side of your trade. You cannot assess their creditworthiness, and you would not trade with a stranger for settlement in two days if their failure were your problem.

Novation removes the question. The original bilateral contract between the two parties is legally extinguished and replaced by two new contracts: one between the buyer and a central counterparty, and one between the CCP and the seller.

The economics are unchanged. The buyer still buys at the agreed price and the seller still sells at it. What changes is who each faces. Both now face the CCP, and the CCP faces both.

The consequences are large and immediate. Counterparty analysis collapses to one question: is the CCP sound? Anonymous trading becomes safe, because identity stopped mattering. And netting becomes multilateral rather than bilateral, since all obligations now run through one entity.

What it does not do is make the risk disappear. It moves it, and where it moved is the subject of the next lesson.

Central clearing is not universal, and the pattern of where it applies is informative.

Exchange-traded instruments are almost always centrally cleared. Trading is anonymous, so it could not work otherwise, and the contracts are standardised, which makes them easy for a CCP to value and to close out.

Standardised derivatives are largely cleared, in many jurisdictions by mandate following post-crisis reform.

Bespoke bilateral contracts are generally not. A CCP must be able to value a position daily and liquidate it in a default, and neither is possible for a genuinely one-off contract with no market.

That last row states the real precondition. Central clearing requires standardisation and liquidity, because the CCP's defence in a default is to close the defaulter's positions in the market. A position that cannot be closed cannot be cleared, however desirable clearing would be.

So the boundary between cleared and bilateral is not primarily regulatory. It is a boundary of liquidity, and it moves as instruments become standardised.

The fourth row is worth a moment. Ownership and entitlements follow the settlement date, not the trade date, which is why buying just before a dividend record date may not entitle you to the dividend.

The pattern across the table is that this half of the business is unglamorous, largely invisible when working, and completely determining when not. It is also where a surprising share of operational cost sits, because every one of these steps is a place where instructions can disagree.

The next lesson turns to the institution that absorbed the counterparty problem, and what it does when a member actually fails.