How to scale: growing size without blowing up
The answer
Increase size only when a written milestone has been met, and decide the milestones before the good period arrives. The reason is not caution. It is that a run of profits and a genuine edge look identical from inside, and the only thing that separates them is a number of trades large enough to be measured.
Why this costs you money
The sequence is the same every time and it does not require any mistake in judgement.
A trader has a method. They trade it small for 3 months. It works. The account is up 11%.
They conclude the method is proven and triple the size. This is the reasonable conclusion from the evidence available to them, which is why it happens to almost everybody.
The method then produces its ordinary losing run, which was always going to arrive. At the old size that run would have cost 6% of the account. At the new size it costs 18%.
Two things now happen at once, and the second is worse.
The account is down 18% and needs a 22% gain to return to its previous value.
And the trader is now managing a position size they have never held through a loss. The decisions become different decisions. The stop that was easy to honour at Rs 5,000 is not easy at Rs 15,000. Positions get closed early. Losses get held, because closing one now means recognising a number that matters.
So the size increase did not just multiply the loss. It changed the person executing the method, and it did so at the exact moment the method was in its worst period.
A 3-month winning run is not evidence. With a method that has no edge at all, a positive 3-month period is common. That is the entire problem, and no amount of confidence about your own method changes the arithmetic.
How it works
Scaling is a set of rules with 3 parts: what you measure, what triggers an increase, and what triggers a decrease.
What you measure: expectancy
Expectancy is the average result per trade, measured in R, where 1R is the amount you risked on that trade. Risk Rs 4,000 and make Rs 8,000, and that was a 2R win.
Expectancy = (win rate × average win in R) − (loss rate × average loss in R)
Two examples show why the win rate alone tells you nothing.
- Win rate 40%, average win 2R, average loss 1R: (0.4 × 2) − (0.6 × 1) = +0.2R per trade.
- Win rate 60%, average win 0.5R, average loss 1R: (0.6 × 0.5) − (0.4 × 1) = −0.1R per trade.
The second method is right more often and loses money. Expectancy is the number to scale on, and it is the number nobody calculates.
Expectancy needs a sample. Below about 30 trades it is noise. At 100 trades it is a rough estimate. At 200 or more it starts to be a measurement, and even then it changes when market conditions change.
What triggers an increase
A phased scaling rule has 4 elements, all written before you start.
1. A minimum number of trades in the phase. Not a time period. Fifty trades is a reasonable floor for the first phase, 100 is better. A month is not a unit of evidence; a month can contain 3 trades or 60.
2. A minimum expectancy over those trades. Positive is the floor. Stating the number in advance prevents the common move of deciding afterwards that the result was "good enough".
3. A maximum drawdown during the phase. If the method reached the expectancy target but had a 20% drawdown getting there, it has failed the phase, because the drawdown at the next size up would be unmanageable.
4. A fixed increment. Increase risk per position by a set step, not by a multiple. Moving from 0.5% to 0.75% is a step. Moving from 0.5% to 1.5% is a different method with a different drawdown profile, and you have no data on it.
A workable ladder for a retail account:
| Phase | Risk per position | To advance you need |
|---|---|---|
| 1 | 0.25% | 50 trades, positive expectancy, drawdown under 5% |
| 2 | 0.50% | 100 trades, expectancy at or above phase 1, drawdown under 8% |
| 3 | 0.75% | 100 trades, expectancy holding, drawdown under 10% |
| 4 | 1.00% | This is the ceiling for most retail accounts |
The numbers are a starting point. The structure is the part that matters: a trade count, an expectancy floor, a drawdown ceiling, and a small step.
What triggers a decrease
This half is missing from almost every description of scaling, and it is the half that protects the account.
Reduce size when the account falls a stated percentage from its peak. A common rule: at a 10% drawdown, drop 1 phase. At 15%, drop 2 phases. Return to the previous size only by meeting the phase requirement again, from the beginning.
Written that way, the reduction is automatic and it is not a judgement made while losing. This matters because a discretionary reduction during a drawdown has the failure described in article 3: you are at reduced size when the recovery arrives.
The difference between the 2 is that a written rule also states how you get back up. A discretionary reduction has no route back, so it becomes permanent by default.
Compounding against pyramiding
Compounding means the risk per position is a percentage of the current account, so the position size rises as the account rises. It is automatic, it is gradual, and it is the correct default.
Pyramiding means adding to an open position that is working. It is a different activity and it is governed by the rules in article 4: never add to a loser, add only after 1R of favourable movement, move the stop first, and keep total open risk on the position within the per-position limit.
The confusion between these 2 is expensive. Compounding increases size across trades based on the account. Pyramiding increases size within a trade based on the price. A trader doing both without noticing can double their intended risk inside a week.
What it tells you, and what it does not
A phase system tells you whether your results are stable enough to justify more capital. It does not tell you the method will keep working. Expectancy measured over 200 trades in one market regime is a description of that regime.
It also cannot separate a genuine edge from a lucky sample with certainty. It only makes the sample large enough that luck is a less likely explanation. That is the honest claim, and it is enough to act on.
And it says nothing about capacity. A method that works at Rs 2,00,000 may not work at Rs 20,00,000 if it trades instruments with limited daily volume. Size that exceeds what the market can absorb becomes its own opponent.
The decision rule
Size increases are earned by trade count, not by profit.
Before any increase, all 4 must be true:
- The minimum number of trades in the current phase is complete.
- Expectancy over those trades is positive and at or above the stated floor.
- The largest drawdown in the phase stayed under the stated ceiling.
- The increment is the next step on the ladder, not a multiple.
And the reduction rule has to exist before the increase rule is used. A ladder with no way down is not a ladder.
Try this now
Five minutes with your trade history. This is the number the whole article turns on.
- Download your closed trades for the last 12 months from your broker's report section.
- Count the total number of closed trades. Write it down.
- Count the wins and calculate the win rate as wins ÷ total.
- Add up all the winning amounts and divide by the number of wins. That is the average win. Do the same for losses to get the average loss. Use the figures after costs.
- Calculate:
(win rate × average win) − (loss rate × average loss). That is your expectancy per trade in currency.
What you should see. Three possible results, and each has a specific action.
Fewer than 50 trades. You cannot scale yet. There is no measurement here, whatever the account balance says. Complete the phase.
Expectancy negative. Increasing size multiplies a negative number. The action is to reduce size to the minimum your account permits and work on the method, not on the size.
Expectancy positive with 100 or more trades. Now divide the expectancy by your average risk per trade to express it in R. If it is above about 0.15R, you have something measurable. Write down your ladder before you use it.
One extra step, 1 minute. Find the highest account value in the period and the lowest value that followed it. That percentage is your drawdown for the phase. Compare it against the ceiling you would have set. Most people find their drawdown was larger than any ceiling they would have accepted in advance, which is the argument for setting the ceiling in advance.
Three real cases
1. The Turtle programme, Richard Dennis and William Eckhardt (recruits trained 1983 and 1984) — size defined by rule, not by confidence The written rules given to the recruits defined position size in units linked to each market's recent volatility, and specified how many units could be added as a position moved in their favour, with a maximum. The size decision was removed from the trader entirely. The point is structural. Novices were given a scaling rule before they were given capital, not after a good quarter.
2. Long-Term Capital Management, December 1997 and September 1998 — scaling by shrinking the base The fund had performed strongly and, at the end of 1997, returned approximately $2.7 billion of capital to investors while keeping its positions. The positions did not shrink with the capital, so leverage on the remaining base rose. When Russia defaulted in August 1998 and the mispricings widened, the smaller capital base could not support the same positions. A consortium of banks recapitalised the fund in an arrangement organised by the Federal Reserve Bank of New York on 23 September
- Increasing size does not require buying more. Reducing the capital behind
the same positions achieves it silently.
3. The LF Woodford Equity Income Fund, suspended 3 June 2019 (United Kingdom) — the method outgrew what it traded Neil Woodford had a long and well-known record. The fund attracted a very large amount of capital and held significant positions in small and unquoted companies. When investors asked for their money back, the holdings could not be sold quickly enough at reasonable prices. Dealing was suspended on 3 June 2019 and the fund was later wound up. The method was not disproved by the market. It was disproved by its own size relative to the liquidity of what it held.
The question that resolves it
A novice asks: the method is working, so how much should I increase?
An expert asks: how many trades has it produced, what is the expectancy across all of them, and what is my written rule for reducing size if the next 20 go badly?
The first question has an answer available in 5 seconds and it is usually wrong. The second takes 5 minutes with a spreadsheet and it is the same question every professional risk function asks.
What would make this wrong
If a 3-month profitable period reliably predicted the next 3 months, phased scaling would be unnecessary caution. It does not. Short-run results from methods with small edges are dominated by variation, which is arithmetic rather than opinion.
The honest limits.
Phases slow you down when the edge is real. A trader with a genuine method and 4 phases at 50 to 100 trades each will spend a long time at small size, earning less than they could have. That is a real cost. It is paid to avoid a scenario that removes the account entirely, and the trade is worth making only because losses compound differently from gains.
Trade counts are not the only valid trigger. Some professional systems scale on the stability of the expectancy rather than on a raw count. That is better and it requires more data than a retail trader will have.
The ladder numbers are conventions. Nothing establishes 0.25% as the correct starting risk. What matters is that the steps are small, the triggers are written, and the reduction rule exists.
Capacity limits are real at every size. A method trading small companies runs out of room far earlier than one trading index instruments. Position size above roughly 5% of average daily volume makes your own order the reason the price moved.
In India
Margin sets a hard floor on the first phase. After SEBI's peak margin framework, phased in from December 2020 through September 2021, the full applicable margin must be collected upfront and margin is measured on intraday snapshots. The effect on scaling is direct. A small account cannot start at a very small size in derivatives, because the margin for 1 lot is a fixed amount that may already exceed the intended risk.
Lot sizes make the ladder lumpy. SEBI's October 2024 measures raised the minimum contract value for index derivatives, and lot sizes were revised upward. A ladder moving from 0.25% to 0.5% risk needs a position size that can change by that amount. In cash equity it can, because you can buy 1 share. In index derivatives it often cannot, because the smallest available step is 1 lot.
The practical consequence. For an Indian account below a certain size, the early phases of a scaling ladder can only be run in cash equity. Attempting them in index options means the first phase is already at 4% or 5% risk per position, which is not phase 1 of anything.
Tax adds a step. Trading income is commonly treated as business income, which brings bookkeeping and, above certain turnover thresholds, audit requirements. Scaling raises turnover quickly.
In the United States
Fractional shares make the ladder smooth. Dollar-based and fractional orders mean a risk step from 0.25% to 0.5% is exactly executable in equities at any account size. The ladder can be run precisely from the first phase.
The pattern day trader rule caps the first phases for small accounts. An account under $25,000 in equity cannot make 4 or more day trades in 5 business days in a margin account. For a day trading method, this means the trade count needed to complete a phase accumulates slowly, which is an accidental and quite effective scaling control.
Margin availability increases with size, which works against the ladder. Portfolio margin becomes available above the broker's threshold, commonly $100,000 or more. Exactly when an account reaches the size where the ladder says risk per position may rise, the broker also permits a larger position for the same capital. Two increases can arrive in the same week, and only 1 of them was planned.
Options assignment complicates the count. Short option positions can be assigned before expiry, which changes the position without a decision from you. A phase count should record what happened, not what was intended.
Where they differ, and what that tells you
The American structure limits how fast you can accumulate trades. The Indian structure limits how small you can start.
The pattern day trader rule slows the trade count for a small American account, which stretches phase 1 across months. That is inconvenient and it is close to what a well-designed scaling rule would impose anyway.
Indian margin and lot sizes do the opposite. They set a minimum position size in derivatives, which for a small account is a large percentage of capital. There is no slow first phase available in that segment. The first trade is already at a risk level the ladder would place in phase 4 or beyond.
What that tells you is where each market's beginner should serve their first 100 trades. For an Indian beginner, the answer is cash equity, where 1 share is a valid position size and a genuine phase 1 exists. Starting in index options is not starting small. It is starting at the top of a ladder that has no lower rungs, which is why the loss statistics for that segment look the way they do.
Carry this
- A 3-month winning run is not evidence. A method with no edge produces those regularly.
- Scale on expectancy across 50 to 100 trades per phase, not on profit.
- Every phase needs 4 written numbers: trade count, expectancy floor, drawdown ceiling and increment size.
- The reduction rule must exist before the increase rule is used.
- Compounding raises size across trades. Pyramiding raises it within a trade. Doing both without noticing doubles your risk.