Fill the block: a crypto order routing game for one large trade
One block of bitcoin, four venues and a cost report at the end. Split the order, send it, and see how close the fill came to the best mix. Below the game: how each venue is modeled, how the score works and what the five scenarios teach.
The order
Quiet Tuesday. Buy 400 BTC on a calm weekday. The lit book is deep, the dark pool has interest resting, the desk quotes wide.
Order: buy 400 BTC at an illustrative mid of $100,000
Half-spread 1 bp · about 20 BTC per 1 bp level
Public bids and offers. Fast, visible, and every extra level costs more.
About 150 BTC resting at the midpoint · unmatched rest pays 6 bp
Crosses at the midpoint, but only against the size that happens to be resting there.
Quote: mid + 12 bp · minimum ticket 100 BTC
One firm price for the whole slice, with a premium and a minimum ticket.
Pool fee 5 bp · effective depth 200,000 BTC
A constant-product pool: a fee on every swap plus impact that grows with size.
The sliders set weights; the shares always add up to 100%.
Execution report
Ready
The order ticket as it stands. Send it to see what each slice costs.
- Lit order bookthe whole block
- 400 BTC100%
- Dark poolnot used yet
- 0 BTC0%
- OTC desknot used yet
- 0 BTC0%
- DEX poolnot used yet
- 0 BTC0%
Before sending, check three numbers
- Resting size in the dark pool. Anything above it misses the midpoint and pays more on the book.
- The desk's minimum ticket. A smaller slice is declined and lands on the lit book.
- Depth per level. The thinner the book, the faster each extra bitcoin costs more.
0 bp
Cost against the mid $0 · score 0/100
Illustrative venues and numbers, not live markets and not a recommendation. Fees beyond those shown, settlement and credit risk are left out.
In 30 seconds
Fill the block is a short order routing game. The player has one large bitcoin order and four places to send it: a lit order book, a dark pool, an OTC desk and a DEX pool. Each venue charges in a different way, so the cheapest fill almost always comes from a mix. After the order is sent, the report shows the cost of each slice in basis points and dollars, a score out of 100 and the best split for that scenario.
How to play the order routing game
Pick one of the five scenarios at the top of the ticket. Each one gives an order size, an illustrative mid price of $100,000 per bitcoin and a short description of the market on that day. Under it sit four venues, each with a slider and a line of facts: the half-spread and depth of the lit book, the size resting in the dark pool, the premium and minimum ticket of the OTC desk, and the fee and effective depth of the DEX pool.
The sliders set weights, not fixed percentages, so the shares always add up to 100%. With the other two at zero, moving the dark pool slider to the same position as the lit book slider sends half of the order to each. The bar on the right shows the split as it changes. Nothing is priced until the order is sent, the same way a real trader only learns the true cost of a fill after it has happened.
Press Send the order and the execution report appears: the average cost against the mid in basis points, the same cost in dollars, a score and a line for each venue explaining what happened to its slice. The best split for the scenario is shown under the report. Change the mix and send again as often as needed; the browser keeps the best score for each scenario and shows it next to its name.
The four venues and how each one charges
The lit order book charges for depth. The first bitcoin is bought at the mid plus the half-spread, and every further level costs one more basis point. With 20 BTC resting at each level, a 400 BTC order walks 20 levels and pays, on average, the half-spread plus about 10 basis points. The deeper the book, the slower that cost grows. The guide to liquidity explains why depth matters more than the headline price.
The dark pool charges nothing on the part that matches. A slice that meets resting interest crosses at the midpoint, which is the whole appeal of dark pools. The catch is size: only the amount that happens to be resting can match. Whatever is left has to go back to the lit book after the market has started to move, so it pays the walk through the book plus a leakage charge.
The OTC desk charges a flat premium. Every bitcoin in the slice costs the same, whatever the size, which makes the desk expensive for small amounts and cheap for very large ones. Below the minimum ticket the desk does not quote at all and the slice is sent to the lit book instead. OTC trading covers how real desks price a block.
The DEX pool charges a fee plus price impact. The fee is a fixed share of every swap, and the impact follows the constant-product curve: buying q bitcoin from a pool with an effective depth of y bitcoin costs about q/(y-q) above the starting price. The slippage calculator shows the same curve in dollars.
| Scenario | Order | Lit book | Dark pool | OTC desk | DEX pool |
|---|---|---|---|---|---|
| Quiet Tuesday | 400 BTC | 1 bp half-spread, 20 BTC per level | 150 BTC resting | +12 bp, minimum 100 BTC | 5 bp fee, depth 200,000 BTC |
| Thin book | 400 BTC | 4 bp half-spread, 4 BTC per level | 80 BTC resting | +15 bp, minimum 100 BTC | 5 bp fee, depth 150,000 BTC |
| Weekend | 300 BTC | 3 bp half-spread, 3 BTC per level | 20 BTC resting | +40 bp, minimum 100 BTC | 5 bp fee, depth 400,000 BTC |
| Headline shock | 600 BTC | 8 bp half-spread, 3 BTC per level | 30 BTC resting | +28 bp, minimum 150 BTC | 30 bp fee, depth 80,000 BTC |
| Whale block | 2,000 BTC | 1 bp half-spread, 25 BTC per level | 500 BTC resting | +14 bp, minimum 250 BTC | 5 bp fee, depth 300,000 BTC |
How the score works
The score compares the fill with two reference points. Zero is the naive route: the whole block sent to the lit book as one market order. One hundred is the cheapest split the game can find, searched in steps of one percentage point across all four venues, which comes to 176,851 possible mixes per scenario. A fill halfway between the two scores 50. A fill that costs more than the naive route scores zero, which is possible: a large slice dumped into a nearly empty dark pool, for example, pays the leakage on top of the book.
The grade line under the cost turns the score into plain words. From 97 up the split is as good as the grid allows. From 85 a few basis points were left behind, usually because one venue carried slightly too much. Lower scores point to the classic mistakes: one venue doing most of the work, a slice under the desk's minimum, or a pool asked to absorb more than its depth allows.
Costs are shown in basis points of the order value because that is how execution desks compare fills of different sizes. One basis point is 0.01%. On the quiet-day order of 400 BTC at $100,000, worth $40 million, each basis point is $4,000, so the gap between a naive fill of 11 bp and a careful one of about 4 bp is worth roughly $28,000.
What the five scenarios teach about crypto order routing
On a quiet Tuesday three venues share the best split. The dark pool takes about the size that is resting there, the lit book takes roughly half while its levels are still cheap, and the deep pool absorbs the rest. The desk stays unused: its 12 bp premium is more than the deep book charges for the same bitcoin. No single venue is cheap for the whole order, which is the core idea behind smart order routing.
When the book is thin, the lit venue becomes expensive fast and the desk's flat premium starts to look cheap. The dark pool still helps, but only up to the 80 BTC resting there; sending more than that is paid for twice. On the weekend the picture flips: desks quote wide, the book is shallow and the deep on-chain pool carries most of the order. In a headline shock almost everything is expensive, the pool's fee has jumped and the dark pool is nearly empty, so the desk's firm price takes the bulk of the block.
The whale block shows why size changes the answer. At 2,000 BTC even a deep book gets expensive, the dark pool can absorb a quarter of the order at most, and the desk's premium on the remainder becomes the anchor of the trade. The lesson repeats across all five: the right route depends on the size of the order relative to what each venue can take, not on which venue looks cheapest at the top of the screen.
How a routing desk thinks about a block
- Size the orderCompare the block with the depth each venue shows or is likely to hold.
- Take free size firstCross what can be matched at the midpoint before touching the book.
- Skim the bookTake only the levels near the top, where each unit is still cheap.
- Price the remainderAsk a desk for a firm quote, or split it into a pool if it is deep.
- Measure the fillCompare the result with the mid at arrival and record the cost.
How real trading desks split a block
The game compresses into one click what real desks do over minutes or hours. A smart order router checks the venues it is connected to, sends child orders to the best prices and keeps re-checking as the book changes. Execution algorithms spread an order over time, aiming at a benchmark such as the volume-weighted average price, so that no single moment absorbs the whole block.
In crypto the venues are more scattered than in equities. The same coin trades on dozens of centralized exchanges, on on-chain pools and through OTC desks, and there is no consolidated best price that every broker must respect, unlike the national best bid and offer under Regulation NMS in US stocks. That makes routing both more valuable and harder to check. A DEX aggregator does a version of the same job on-chain, splitting one swap across several pools to cut price impact.
Brokers that handle customer orders in regulated markets also carry a best execution duty. In the United States, FINRA Rule 5310 asks a firm to use reasonable diligence to find the best market for a security, taking into account factors such as the size of the order, the number of markets checked and the accessibility of quotations. The game turns those factors into four sliders.
What the game leaves out
The venues and numbers are invented for teaching and do not describe any real exchange, pool or desk. Real fills also depend on trading fees, settlement time, the credit of the counterparty, information leakage before the order is sent and other traders reacting to it. The game is not trading advice and recommends no venue, coin or service.
Where the model is deliberately simple
Each venue is reduced to one or two numbers so that the trade-offs stay visible. The lit book adds the same size at every level, while real books are lumpy and can be refilled within seconds by market makers. The dark pool's resting size is known exactly in the game; in practice it is unknown until an order tries to match, which is why traders send small probing orders first. The leakage charge stands for the market moving against an order that failed to match, a cost real traders estimate from their own records.
The OTC desk quotes the same premium for any amount above its minimum, while a real desk would widen its price for very large size or tighten it for a client it trusts. The DEX pool uses an effective depth to stand for concentrated liquidity, and it ignores gas fees and the risk of maximal extractable value such as sandwich attacks, both covered in the guide to decentralized exchanges.
All five scenarios are buys. A sell works the same way with the book and the pool mirrored, so it was left out to keep the ticket short. For more practice with the same ideas, the daily puzzle tests the vocabulary of market structure one term at a time.
Short answers
Fill the block FAQ
What is crypto order routing?
Crypto order routing is the choice of where to send each part of a trade. A large order can be split between centralized exchanges, dark pools, OTC desks and on-chain pools so that no single venue has to absorb all of it. A smart order router automates that choice by comparing prices and depth across the venues it is connected to.
Why not send the whole block to one exchange?
A large order sent to one order book walks through many price levels, and each level is more expensive than the last. The average price can end up far from the mid. Splitting the order lets each venue take only the part it can fill cheaply, which is why the naive one-venue route scores zero in the game.
Why does the dark pool sometimes make the fill worse?
Only the size resting in the pool can match at the midpoint. Any slice above that size finds no match, and in the game it goes back to the lit book after the market has moved, paying the walk through the book plus a leakage charge. Sending more to a dark pool than it can hold is one of the classic routing mistakes.
What does the OTC minimum ticket do?
The desk in the game only quotes slices at or above its minimum size. A smaller slice is declined and goes to the lit book instead, so its cost is the book's cost, not the desk's premium. Real desks also set minimum sizes, because quoting a firm price takes work that only pays off on large trades.
Is the best mix the same in every scenario?
No. The best split depends on the order size and on what each venue offers that day. On a quiet day the book, the dark pool and the on-chain pool share the order, on the weekend the deep pool carries most of it and in a headline shock the OTC desk does. The report shows the best mix after every send.
Are the prices in the game real?
No. The mid price of $100,000, the spreads, the depths and the premiums are illustrative numbers chosen to show the trade-offs clearly. The game runs in the browser, sends nothing anywhere and is not a quote or a recommendation for any venue or coin.
Sources
- FINRA: Rule 5310, Best Execution and Interpositioning
- SEC: Regulation NMS final rule, Release 34-51808 (2005)
- Investor.gov: Alternative trading systems (ATSs)
- Nasdaq: Midpoint Peg Post-Only order factsheet
- Uniswap docs: How Uniswap works (v2 protocol overview)
- ethereum.org: Maximal extractable value (MEV)
Try it
Tools and games, no sign-up
All of them run in the browser and send nothing anywhere. The numbers are illustrative, never live prices.