Thick books, thin books and where DOM reading works

New order flow traders usually start on the most liquid product they can find, reasoning that more liquidity means cleaner information. For DOM reading specifically, the opposite is closer to the truth. This lesson explains why the DOM on a liquid index future is mostly unreadable by eye, why it becomes readable on thinner products, and how to choose where to practise.

What "thick" means in practice

Consider two futures contracts side by side at 10:30 in the main session. Numbers are illustrative of the kind of difference, not exact figures for any product.

On the thick product, a human trader lifting 2 contracts is 0.2% of the displayed ask. Their aggression is invisible. Size at each level is dominated by algorithms that add and pull in milliseconds, so what your eye sees as "480 on the bid" was 520, 310 and 490 within the last second. Reading pulling and adding by eye is not possible at that speed; the DOM becomes a flickering column of numbers whose changes you cannot attribute to anything.

On the thinner product, a 20-lot market order is a visible event: it clears a level, the tape shows one print, and the next level's behaviour (pull, hold, add) plays out over a second or two. The behaviours from lessons 2 and 3 are visible to a human.

Why thick is still useful, just differently

Thick products are not worse for order flow; they are worse for raw DOM reading by eye. Three things still work well on them:

1. Footprint and delta (Module 3). Aggregating prints per price per bar smooths out the millisecond noise and shows where aggression concentrated. This scales to any liquidity.
2. Volume profile (Module 4). Entirely liquidity-independent; more volume means a cleaner profile.
3. Absorption at major levels. Even on a thick product, when 6,000 contracts trade at one price and it does not move, that is visible on the inside-volume column and on the footprint.

What does not work on thick products is watching individual levels a few ticks away and inferring intent from size changes. That size is mostly quoting algorithms, and their changes carry little information a discretionary trader can use.

Key idea: DOM reading needs prints that are large relative to displayed size. On thick books they are not, and the book is dominated by fast algorithmic quoting. Learn the DOM on a product where a single trade visibly changes the ladder.

Too thin is also a problem

Go too far the other way and the DOM becomes unreadable for a different reason.

  Bid | Price  | Ask
  ----+--------+-----
      | 82.60  |  3
      | 82.55  |
      | 82.50  |  8
      | 82.45  |  1
      | 82.40  |
   2  | 82.35  |
      | 82.30  |
   6  | 82.25  |
      | 82.20  |
   1  | 82.15  |

Gaps, one-lots, a wide inside spread. A single 10-lot market order would move price four ticks. Nothing here is "absorption" or "pulling"; it is just emptiness. Reading behaviour requires enough participants that behaviour exists. Products like this also punish execution: your own market orders cause the slippage you are trying to read.

A practical spectrum

The medium band is where most traders find that DOM reading actually teaches them something. The product does not need to be exotic; it needs prints that are large relative to displayed size and a book that updates at human speed.

Session time changes the answer

The same product can be thick at 9:45 and thin at 3:00 in the morning. Overnight sessions on index futures are often more readable on the DOM than the main session, simply because displayed size drops and individual trades matter more. The trade-off is that overnight moves are less meaningful and can reverse entirely at the open. Use quiet sessions to learn the DOM; do not assume what you see there predicts the main session.

Stocks, crypto and fragmentation

For stocks, remember that any one exchange's book is a fragment; a consolidated feed helps but still omits dark pool volume. For crypto, each exchange has its own book and its own participants, and the thickness varies wildly by pair and venue. Perpetual futures books on major exchanges behave like thick index futures. Spot books for smaller pairs behave like thin stocks. Apply the same test: can a single visible trade change the ladder in a way you can attribute?

Try it: Pick three products you have data for. For each, watch the DOM for five minutes and record (a) typical displayed size at the inside, (b) the size of the largest single print you see, and (c) whether you could follow the behaviour of a level as price approached it. Rank the three by readability. That ranking, not popularity, decides where you practise.

Recap

• On very thick books, displayed size is dominated by fast algorithmic quoting and single trades are invisible; DOM reading by eye does not work well.
• Footprint, delta and profile still work on thick products because they aggregate prints.
• Too thin is also unreadable: gaps and one-lots mean there is no behaviour to read and execution is costly.
• The medium band, where a single trade visibly changes the ladder, is where DOM skills develop.
• The same product's readability changes with session time; quiet sessions are good classrooms but poor predictors.