WhitmanTrading

How to Use Margin Without Blowing Up

To use margin without blowing up, find your broker's maintenance requirement first and calculate what price would trigger a call. Size the position so that price is beyond any move you would accept. Leverage magnifies drawdown faster than it magnifies return.

Buying power is the number brokers display and the maintenance requirement is the number that ends accounts. This procedure works backwards from the second one and ignores the first.

Before you start

Your broker’s maintenance requirement in writing. The percentage of the position’s value your equity must stay above. It differs by instrument and the broker can raise it without much notice.

The current margin interest rate. It accrues daily on the borrowed portion whether the position moves or not, and it is quoted as an annual figure that people mentally discount.

A position size calculated from the maintenance level, not the buying power. Worked out before the order exists, because after it exists the arithmetic no longer has any influence.

The steps

1. Write down the maintenance requirement, not the buying power

Price bars with planned levels drawn as horizontal lines.
The number that matters is the maintenance level. Illustrative chart - not real market data.

Buying power tells you what the broker will lend. The maintenance requirement tells you when it gets taken back. Only the second one has consequences.

2. Calculate the price that triggers a call

A candlestick chart of the site's shared price history.
Borrowed money magnifies the loss first. Illustrative chart - not real market data.

Work out the exact price at which your equity falls below the requirement. Write it on the chart as a line. That price is the real stop, whatever your intended stop says.

3. Size so that price is beyond any move you would accept

The first half of the long price series.
While the drawdown went from 3.76% to 11.08%. Illustrative chart - not real market data.

If an ordinary adverse move reaches the call price, the position is too large. Reduce until the call price sits outside the range of normal movement.

4. Price the interest before you price the trade

A section of the price series.
You are borrowing at a rate that changes without notice. Illustrative chart - not real market data.

Multiply the borrowed amount by the daily rate and by your expected holding period. A position held for months pays this every single day.

5. Account for the decay on anything held

A long-horizon candlestick view of the same price series.
And leverage decays: 3x returned 8.93% where naive said 10.82%. Illustrative chart - not real market data.

Leveraged returns compound on a path, not on a total. The longer the hold and the choppier the path, the further the result falls behind the simple multiple.

6. Decide what you will sell before the broker decides for you

The second half of the long price series.
The broker closes you, and chooses what and when. Illustrative chart - not real market data.

Write the order you would place if the call price is reached. A broker liquidating on your behalf picks whatever is easiest to sell, not whatever you would have chosen.

7. Check the liquidity of what you are holding

A candlestick chart with a volume histogram beneath it.
In a thin market a forced sale moves price against you. Illustrative chart - not real market data.

A forced sale into a thin market moves the price against the seller. Leverage on an illiquid instrument compounds the problem it was supposed to solve.

How to tell it worked

Run three checks before the order, and one after 30 days.

Check the distance to the call price in bar ranges. On the shared series a typical bar spans 0.493 and the ninetieth percentile is 1.101. If your call price is under 5 typical bars away, the position is sized for a market that does not move, which is not the market you are in.

A candlestick chart annotated with the round-trip cost.
Interest runs daily, on top of 2% of a bar per trip. Illustrative chart - not real market data.

Check the interest against the expected gain. Borrowing costs accrue every day and a round trip already costs 2% of a median bar’s range, so a position needs to clear both before it clears zero.

Then, after 30 days, count the times you were within one bar of the call price. The acceptable answer is 0. Anything above that means the sizing at step three was optimistic, and the position survived on luck rather than on arithmetic.

What leverage actually does to the numbers

A sideways, range-bound candlestick series.
And a flat market still charges interest every day. Illustrative chart - not real market data.

The measured figures on this site’s shared series are the clearest version of the argument. Unleveraged, the series finished +3.61% with a worst drawdown of 3.76%. At 2x, the actual result was 6.61% where naive doubling predicted 7.22%, and the drawdown went to 7.45%. At 3x, the actual result was 8.93% against a predicted 10.82%, with a drawdown of 11.08%.

Read those as pairs. Trebling the exposure did not treble the return — it fell short by nearly two percentage points — but it did very nearly treble the worst loss. The gap between 10.82% and 8.93% is the decay, and it comes from compounding on a path rather than on a straight line.

The figures are in research/series-measurements.json. They describe one series and are not a forecast for any other, but the direction of the effect is a property of the arithmetic and holds generally: the drawdown scales more faithfully than the return does.

The original data

Of the 24,971 unique videos in research/search-study-corpus.jsonl, 9 have an instruction-shaped title mentioning margin, at a median of 38,013 views across 8 channels, with a maximum of 341,625. Leverage appears in 17 instructional titles at a median of 16,703 across 17 channels. The counts come from site/rank_howto.py.

A rising stretch of the price series cut short at the decision bar.
The position is down and the call is due. Add? Illustrative chart - not real market data.

Nine videos across eight channels is almost no coverage at all, for a mechanism that is available by default in most brokerage accounts and is a common way for accounts to end. The subject is unglamorous, which is exactly why the supply is thin and the median is high.

The answer to the question on that chart is that adding money to meet a call is a new decision and should be judged as one. The question is not whether you want to keep the position — it is whether you would open it at this price and this size. If the answer is no, meeting the call is buying something you would not buy, funded by the fact that you already own it.

When it fails

A candlestick series containing several opening gaps, with the largest marked.
And a gap can take you past the level before you can act. Illustrative chart - not real market data.

A gap through the call price is the failure no procedure prevents. Price opens beyond the level, there was no opportunity to act between the two prints, and the position is already liquidatable before the session starts. Sizing is the only defence, because a stop cannot fill at a price that never traded — which is why step three is about size rather than about stops.

The second failure is a flat market. Interest accrues daily regardless, so a position that goes nowhere for months has still cost real money.

A third is treating buying power as a target. It is a limit on what the broker will lend, not a recommendation of how much to borrow.

A fourth is holding leveraged exposure for long periods. The decay above compounds against you the longer and choppier the path.

A fifth is leverage on an illiquid instrument. The forced sale moves the price against you at the exact moment you cannot choose the timing.

And a sixth is meeting a call by adding funds without re-examining the position. That converts a sizing error into a larger sizing error.

Margin account covers what the account is and how initial and maintenance requirements differ. Leverage is the general mechanism and where the decay comes from. And drawdown is the measurement that leverage magnifies fastest, which is the whole argument on this page.

What I actually do

The number that changed how I think about this is not the return figure, it is the drawdown one. Trebling the exposure on the shared series took the worst fall from 3.76% to 11.08% while the return went from 3.61% to 8.93% — so the pain nearly trebled and the reward did not quite double. Once I saw those two numbers side by side, leverage stopped looking like a multiplier and started looking like a trade.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.