Position Sizing
How much to trade is a separate decision from what to trade, and it is the one that determines whether a losing run is survivable.
MadStockAlerts Research · Updated August 28, 2026
What to take away
- Size follows from a chosen loss amount and the distance to the exit.
- Position size and risk are not the same thing: a wide stop makes a small position risky.
- The one percent convention is a widely cited rule of thumb, not a rule.
- Volatility-adjusted sizing equalises risk across securities that move differently.
- Correlated positions are closer to one position than to several.
MAD Academy Training Video · 0:46
Size Is the Decision, Not the Entry
Position size is what turns a stop distance into a fixed, known risk — and it is the only variable you fully control.
This lesson is part of a Stock Alerts + Tools plan.
The arithmetic
shares = amount at risk / (entry price - exit price)
- amount at risk is the loss accepted if the exit is reached
- the denominator is the distance to that exit, per share
On a $50,000 account, accepting $500 of loss on a trade entered at $40 with an exit at $37 gives $500 divided by $3, or 166 shares. That is a $6,640 position, about thirteen percent of the account, risking one percent of it.
The position size and the risk are different numbers, and confusing them is the most common error here. A $20,000 position with a stop half a percent away risks $100. A $2,000 position with a stop twenty percent away risks $400. The larger position is the safer one.
The order of operations follows from that. The exit is chosen from the chart, the accepted loss is chosen from the account, and the size is whatever those two produce. Choosing the size first and then looking for somewhere to put a stop reverses the logic and puts the exit wherever the size requires.
- 1Where the idea is wrongRead off the chart, before any money is involved
- 2What a loss there costsChosen from the account, not from the trade
- 3SharesWhatever the first two produce, and nothing else
The one percent convention
Limiting the loss on any single position to about one percent of an account is a widely repeated convention. Its appeal is arithmetic rather than magical.
| Risk per trade | Account remaining after 10 consecutive losses |
|---|---|
| 1% | About 90% |
| 2% | About 82% |
| 5% | About 60% |
| 10% | About 35% |
| 20% | About 11% |
Ten consecutive losses is not an unusual event for a method that wins half the time; it has a probability of about one in a thousand per specific sequence, and across hundreds of trades such runs occur. Any sizing approach has to survive a run that will occur.
The bottom row is the one that ends accounts. At twenty percent per trade the arithmetic of recovery has already become impossible before the run is over, which is the drawdown problem arriving through the sizing decision.
Volatility adjustment
A fixed dollar stop is much tighter on a volatile security than on a quiet one, so identical position sizes carry very different risk. Sizing against Average True Range puts them on the same footing.
shares = amount at risk / (ATR multiple x ATR)
- a wider ATR produces a smaller position for the same accepted loss
The effect is that a quiet large-cap and a volatile small-cap can both be held with the same amount of money at risk, which is what makes a portfolio of very different securities comparable at all.
Correlation and total exposure
Five positions each risking one percent are not risking five percent if they are five semiconductor companies. In a sector-wide decline they move together and behave much more like a single position of five times the size.
This is why total portfolio exposure and the correlation between holdings belong in the sizing decision rather than being considered afterwards. A per-trade rule with no portfolio-level rule permits an account to be fully exposed to one idea expressed ten ways.
Sizing when the exit is not a price
The arithmetic assumes an exit at a known price, which suits a position whose invalidation is a level. Many positions do not have one: the reason for holding might be a thesis about a business, and the thing that would disprove it is a disclosure rather than a price.
Two conventions cover that case. The first is to size on an assumed adverse move rather than on a chosen exit, using a volatility measure to set the assumption: if the security routinely moves three percent in a day, a position sized so that a three-standard-deviation stretch is survivable is sized on the security's own behaviour rather than on a level. The second is to treat the whole position as the risk, which is the honest treatment for anything that can gap to a fraction of its value on a single disclosure.
shares = amount at risk / (k x ATR)
- ATR is the average true range, a measure of typical daily movement
- k is a multiple chosen in advance, commonly between 1.5 and 3
- the result equalises risk across securities that move at different speeds
Both conventions produce smaller positions in more volatile securities, which is the property that makes them comparable. A fixed dollar position in a security that moves one percent a day and one that moves eight is not one risk setting, it is two.
The gap risk the arithmetic ignores
Every sizing calculation assumes the exit can be reached at the exit price. Overnight, that assumption does not hold. A security that closed at $40 with an exit planned at $37 can open at $28, and the loss taken is what the market offers rather than what was planned.
- Scheduled events, principally earnings, are the largest source and the only one that is known in advance.
- Unscheduled disclosures, including trial results, regulatory decisions and management departures, arrive without warning by construction.
- Sector or index shocks move every constituent at once, which is when correlation and gap risk arrive together.
- Thin securities gap on far less news, because there is nothing resting in the book to absorb an imbalance.
The practical consequence is that the planned loss is a floor on how good the outcome can be rather than a cap on how bad it can be. Holding through a scheduled event is a decision to accept an unbounded version of the planned risk, and it is a separate decision from the one that opened the position.
Sizing that would be uncomfortable at three times the planned loss is sizing that has not accounted for this, because three times the planned loss is an ordinary earnings gap rather than an extreme one.