Learn / Risk & Position Sizing
Scaling in vs entering all at once
Scaling into a position and averaging down produce the same weighted average. The difference is entirely in when the decision was made — and that difference determines whether the position size is one you chose or one you arrived at.
The test
Could you have written the full plan down before the first entry?
If yes — total size, entry levels, the condition for each addition — it is a scale-in, and it is a legitimate technique.
If it only makes sense given that you are currently losing, it is a rescue, and the arithmetic looking identical is what makes it persuasive. See averaging down vs dollar-cost averaging.
What entering all at once gives you
Certainty of size. You know your exposure immediately, so the position size calculation describes what you actually hold rather than an intention.
One decision. No further choices to make badly under pressure.
Full participation. If the move goes immediately, you have the whole position.
Simpler records. One entry, one average, one R multiple that means what it says.
What scaling in gives you
A better average, sometimes. If price goes against you before going your way, later entries improve it. If it goes immediately, they do not.
Reduced timing risk. No single entry is decisive, which is the same property that makes DCA useful.
Information between entries. You learn something from how price behaves before committing further.
Smaller market impact. For size that is a meaningful fraction of depth, working in reduces impact — see spread, depth and what liquidity actually costs you.
What it costs
More fees. Every entry is a round trip leg. Three entries plus one exit is four fee events rather than two — see the trading fee calculator.
Incomplete positions. If price runs from the first entry, you are holding a third of what you planned. The win is smaller than the strategy assumed, and that asymmetry is systematic: you get the full position in the trades that go against you and a partial one in the trades that work.
That is worth sitting with. Scaling in concentrates your size in the losers by construction, unless the later entries are also the better ones.
A harder stop question. One stop for the whole position at the original invalidation, or a stop that moves with the average? The first is correct — the stop belongs at invalidation — but it means the later entries have a smaller R than the first.
Complicated records. Multiple entries at different prices make R ambiguous unless you fix 1R as the original entry-to-stop distance and hold it there.
Sizing a ladder correctly
The mistake that turns a scale-in into an oversized position:
The total, fully filled, must sit inside your normal risk budget. Not each tranche — the total. If you plan three entries of 1% each, you have planned a 3% position, and calling it “1% per entry” is a labelling exercise.
So:
- Decide total size from your risk budget and the stop at invalidation.
- Divide it into tranches.
- Define the level or condition for each.
- Cap the total. It does not grow because price moved.
Step 4 is the one that fails, and it fails as a series of reasonable-looking additions — see why position sizes drift upward.
When each is right
| Situation | Approach |
|---|---|
| Specific level, clear invalidation | All at once |
| Breakout or time-sensitive entry | All at once — partial fills miss the move |
| Size is meaningful vs book depth | Scale in, for impact |
| Directional view without a precise level | Scale in |
| Position would be uncomfortable at full size | Neither — reduce total size |
That last row matters. Scaling in to make an oversized position feel manageable does not make it smaller; it makes it slower. The discomfort is information about the total, and the fix is the total.
The AI-assisted note
A model asked to “build a position” will happily produce a ladder, and each tranche looks reasonable in isolation.
The control is the same as everywhere else: the total must be checked against a per-account cap in code, before submission. A limit evaluated per order does not see a ladder, and a ladder is exactly how per-order limits get circumvented without anyone intending to — why MCP needs an approval layer.
FAQ
Is scaling into a position better than entering all at once?
Neither is generally better. Scaling in reduces timing risk and market impact and improves the average when price moves against you first; entering all at once gives certainty of size and full participation when the move happens immediately. The decisive factor is whether you have a precise level or a directional view.
What is the difference between scaling in and averaging down?
When the decision was made. A scale-in has its full size and entry levels defined before the first entry; averaging down is a discretionary response to being underwater. The arithmetic is identical, which is what makes the comparison persuasive and misleading.
How should I size a scale-in?
From the total, fully filled, against your normal risk budget — then divide into tranches. Sizing each tranche at your usual risk means a completed ladder is several times your intended position. The total is the number that must be capped.
Where does the stop go when scaling in?
At the invalidation level for the idea, which does not move because you added. That means later entries carry a smaller R than the first, which is a real cost and the correct one to accept. Moving the stop to accommodate a larger average is widening a stop, with everything that implies.