a-book, b-book, hybrid: a calm look and a way to check
Posted: Fri Oct 30, 2026 11:45 am
A-book: the broker passes client orders on to liquidity providers and earns from spread or commission. B-book: the broker takes the other side itself. Hybrid: some flow goes out, some stays in-house, based on rules the broker rarely publishes.
None of this is automatically shady. Many brokers run hybrids, and risk-managing client flow is a legitimate business. The conflict of interest is real though, since in a B-book a client's win is the broker's loss.
So how could a scalper test for odd behaviour? Here is a checklist of questions:
- Is slippage symmetrical, or does it lean one way?
- Do winning trades take longer to fill than losing ones?
- Do requotes or rejections appear more on profitable days?
- Does the quote differ from an independent feed at the moment of fill?
- Does behaviour change after a run of wins?
Each question needs data. A few trades prove nothing, and a few hundred logged carefully start to mean something. Even then, a pattern points to a question for the broker and not a verdict.
Fill timing is probably the easiest thing to look at. In a pure pass-through model, the time to fill shouldn't care whether the trade ends up profitable. If winners consistently take 300 ms longer than losers over a few hundred trades, that gap is worth noting, though a busy moment or a bigger lot size can explain it just as well.
Another angle is the account itself. Scalping is a style some brokers treat differently, since short, frequent trades are harder to manage as risk. Policies differ, rarely get published in detail, and what applies to one account type may not apply to another. Reading the order execution policy and asking support direct questions costs nothing. Written answers are worth keeping, since the wording of a reply can matter later if a dispute starts, and vague replies are an answer of sorts too.
Which tests would you add, and what sample size would you trust before drawing any conclusion?
None of this is automatically shady. Many brokers run hybrids, and risk-managing client flow is a legitimate business. The conflict of interest is real though, since in a B-book a client's win is the broker's loss.
So how could a scalper test for odd behaviour? Here is a checklist of questions:
- Is slippage symmetrical, or does it lean one way?
- Do winning trades take longer to fill than losing ones?
- Do requotes or rejections appear more on profitable days?
- Does the quote differ from an independent feed at the moment of fill?
- Does behaviour change after a run of wins?
Each question needs data. A few trades prove nothing, and a few hundred logged carefully start to mean something. Even then, a pattern points to a question for the broker and not a verdict.
Fill timing is probably the easiest thing to look at. In a pure pass-through model, the time to fill shouldn't care whether the trade ends up profitable. If winners consistently take 300 ms longer than losers over a few hundred trades, that gap is worth noting, though a busy moment or a bigger lot size can explain it just as well.
Another angle is the account itself. Scalping is a style some brokers treat differently, since short, frequent trades are harder to manage as risk. Policies differ, rarely get published in detail, and what applies to one account type may not apply to another. Reading the order execution policy and asking support direct questions costs nothing. Written answers are worth keeping, since the wording of a reply can matter later if a dispute starts, and vague replies are an answer of sorts too.
Which tests would you add, and what sample size would you trust before drawing any conclusion?