Where you are. Two venues so far: a pool that prices from its own balances, and an auction among vetted firms. Both share a property worth naming. In both, the act of trading tells somebody what you are doing.
You want to buy thirty thousand of something. The book in front of you has two thousand at the best price, three thousand behind it, five thousand behind that.
Take the lot and you will pay progressively worse prices as you go, which is bad enough. Worse, everybody watching now knows a large buyer is present, and the price in front of the rest of your order moves away before you get to it.
The idea in one paragraph
For a large order the dominant cost is not the fee or the spread but market impact: the price moves against you because your own trading revealed your intention. Venues exist along a spectrum defined by how much is revealed before execution. A lit order book shows everything and is cheapest for small orders. A quote auction shows the trade to a few vetted firms. A dark pool shows nothing at all, matching orders internally and reporting only after the fact, which removes the impact and replaces it with a different problem: there may be nobody on the other side, and waiting has its own cost.
Wider than the screen; scroll it sideways.
Measuring the difference
The exercise walks the same thirty thousand order through a lit book and crosses it in the dark, and measures both against the price on the screen when the decision to trade was made.
Walking the book costs an average of 100.4967 against an arrival price of 100, so the shortfall is 14,900. Crossing in the dark at the mid, with a small charge for the delay in finding a counterparty, costs 2,400.
Then the control that makes the lesson: the same book, an order of two thousand, and a shortfall of exactly zero. It fills at the top of the book and the arrival price is what it pays.
| order | venue | average price | shortfall |
|---|---|---|---|
| 2,000 | lit book | 100.0000 | 0 |
| 30,000 | lit book | 100.4967 | 14,900 |
| 30,000 | dark cross | 100.0800 | 2,400 |
n = 3 trials · the same book, against an arrival price of 100.00
Why this belongs in a course about tokens
Everything in this lesson is decades old and none of it is about ledgers, which is the point.
When large orders start trading on shared ledgers, the same forces produce the same structures, and they have already: quote-based venues for size, private order flow arrangements, and matching that reports after the fact rather than before. A transparent public ledger sounds like the opposite of a dark pool, and institutions trading size on one will build the equivalent anyway, because the cost that drives it has not gone anywhere.
The interesting question for a shared ledger is not whether these venues appear. It is that on a public ledger everything is visible after the fact, in a way it is not in the legacy market, which changes what “dark” can even mean.
Check yourself
1. What is the dominant cost of a large order, and why?
Market impact: the price moves against the order because the act of trading revealed a large buyer or seller. It dominates fees and spread at institutional size, because the rest of the order is executed against a market that has already updated on the information.
2. The same book gives a 2,000 order zero shortfall and a 30,000 order 14,900. Why?
Because the small order fills entirely at the best price, so it pays exactly the arrival price. The large one exhausts each level and walks up the book, and the levels it reaches later have also had time to move away from it.
3. What does a dark venue cost, if not price?
Certainty of execution. There may be no counterparty, so the order waits while the market moves, and the trade may not happen at all. Measuring shortfall only on filled orders systematically flatters dark venues, because the worst outcomes never enter the sample.
4. Why would these venues appear on a shared ledger, which is public by design?
Because the cost that produces them, market impact on large orders, is unchanged by the ledger. The structures reappear in whatever form the venue allows. What does change is that a public ledger reveals everything after the fact, which narrows what darkness can mean to the period before execution rather than after it.
Do this
Measure the same order both ways.
python3 code/dark_pool.pyThe starter provides the lit book, the walk that consumes it, and the dark cross at the mid, and leaves you shortfall: what the order paid above the price on the screen when somebody decided to trade, times the size. The assertions check that a small order costs nothing extra, that the large order costs more on the lit book than in the dark, and that both are positive.
The completed version is in solutions/dark_pool.py.
What you can now do. You can measure implementation shortfall, place a venue on the lit-to-dark spectrum, and say what a dark venue costs when its cost is not in the price. The next lesson takes the trading machinery built so far and folds it into a payment.