Risk of ruin and position sizing: the pillar guide

A trader can spend a thousand hours studying entries, exits, indicators, market structure, news flow and order types. They can pick the best strategy in the world. And if they have not done thirty minutes of arithmetic on what their per-tra

A trader can spend a thousand hours studying entries, exits, indicators, market structure, news flow and order types. They can pick the best strategy in the world. And if they have not done thirty minutes of arithmetic on what their per-trade risk actually does to the long-run distribution of their account, they will eventually blow up — not because the strategy was wrong, but because the size was wrong. Risk of ruin is the only piece of trading math that is genuinely non-negotiable, because it is the math that decides whether you are still in the game long enough for any strategy to play out.

This is the pillar guide we use inside the Tradoki desk and the curriculum to make this math unavoidable. It is longer than our other pieces on purpose. The arguments in here either get internalised once or get re-learned the expensive way.

The simplest version of the question

The only question risk of ruin asks is this: given my strategy's win rate, my average win-to-loss ratio, my per-trade risk, and my starting capital, what is the probability I get to a level of drawdown I have decided is ruin?

The level of drawdown that counts as ruin is a personal choice and a strategic one. For some traders it is full account loss; for most professionals it is 50% drawdown, because at 50% drawdown the account has to make 100% to recover, and the empirical recovery rate from a 100% climb is poor. For a beginner, ruin should be set even more conservatively — often a 20% or 25% drawdown — because beyond that point the psychological capacity to keep trading the system as designed tends to evaporate.

The output of the question is a probability. Not a yes or no. Risk of ruin is a property of the system, not a guaranteed outcome. A 5% risk of ruin means, in a long-run sense, one in twenty traders running this system blows up; the other nineteen do not. A 50% risk of ruin means a coin flip. Most retail traders are running, without realising it, systems with double-digit risk of ruin.

The classical formula and what it actually tells you

The classical risk-of-ruin formula assumes a binary outcome (win or loss), a fixed win probability p, a fixed payoff ratio R (average win divided by average loss), and a fixed per-trade risk fraction f. Under those assumptions, the probability of ruin from current capital to zero is approximately:

RoR ≈ ((1 - A) / (1 + A)) ^ (capital_units / risk_units)
where A = p × R - (1 - p)        [the per-trade expected value in risk units]

The formula is not exact for real trading — payoffs are continuous, win rates drift, risk is variable — but the qualitative behaviour matches reality, and that qualitative behaviour is what matters.

The behaviour to internalise:

• Doubling per-trade risk does not double risk of ruin. It increases it by a factor that depends on edge and on the ruin distance, but the factor is typically much larger than two. Under common parameters, doubling per-trade risk from 1% to 2% can multiply the risk of ruin by five or ten times.
• Even small negative edge produces certain ruin given enough trades. If A is negative, the formula says risk of ruin approaches 1 as the number of trades grows. There is no per-trade size small enough to make a losing strategy a winning one over a long sample. Risk of ruin is the discipline of not sizing negative-edge strategies into the ground.
• Even strong positive edge produces non-trivial ruin probability if per-trade risk is large enough. A strategy with a 60% win rate and a 1.5R payoff has an excellent expected value per trade — and at 5% per-trade risk it still has enough variance to put double-digit ruin probabilities on the table over a few hundred trades.

The formula is not the point. The qualitative behaviour is the point.

Why retail education gets this wrong

The most common per-trade risk recommended in retail trading education is between 1% and 2%, with frequent recommendations as high as 5%. The math says these are too high for the strategies being recommended.

The retail recommendation comes from a bad heuristic: "never risk more than 2% on a trade so a single loss does not hurt." This is true for a single loss. It is wildly insufficient for a streak of losses, which is the actual risk distribution. Strategies with 50% win rates produce eight-trade losing streaks regularly. Strategies with 40% win rates produce them frequently. At 2% per-trade risk, an eight-trade losing streak is a 16% drawdown. At 5%, it is 40%. Both are inside the realm of plausible bad weeks, not freak events.

The fix is mechanical. Calculate the expected length of your worst losing streak across the number of trades you intend to take in a year. Calculate what that streak does to your account at your chosen per-trade risk. If the resulting drawdown is past your defined ruin point, the per-trade risk is too high. The per-trade risk is too high because the math says so, regardless of what your trading psychology says about being able to "handle it."

For a deeper treatment of why "handling it" is not actually a thing, see trading psychology without the pop science.

The Kelly criterion: useful, scary, and almost always overstated

The Kelly criterion is the bet size, given known win rate and known payoff ratio, that maximises the long-run geometric growth rate of capital. The formula is simple:

f* = (p × R - (1 - p)) / R
where p = win probability, R = win/loss ratio, f* = optimal fraction

For a strategy with a 55% win rate and a 1R payoff, full Kelly says size 10% per trade. For a strategy with a 60% win rate and a 1.5R payoff, full Kelly says size 33%.

These numbers feel preposterous because they are. Full Kelly assumes the win rate and payoff ratio are known with perfect precision. They are not. They are estimated from a sample, with sampling error, and the sampling error compounds: a one-percentage-point overstatement of edge can shift full Kelly by tens of percentage points, and the resulting size can produce drawdowns that even a profitable strategy cannot recover from psychologically or financially.

The standard professional adjustment is fractional Kelly: take a fraction (commonly one-quarter to one-half) of full Kelly, on the explicit acknowledgement that the edge estimate is imprecise. Quarter-Kelly produces sizes that are typically inside the "small enough to survive a bad sample" zone — usually 1–2% per trade for strong retail strategies, much less for weak ones.

The Kelly framework is useful because it grounds per-trade risk in the strategy's properties rather than in an arbitrary number. Used as a ceiling, not a target, it is one of the cleaner tools in the kit. Used at full size, it is a way to lose money efficiently.

What the position size formula actually looks like

The per-trade position size that comes out of this thinking is mechanical:

risk_per_trade ($) = account_equity × risk_fraction
position_size (units) = risk_per_trade / (entry - stop) × point_value

The risk_fraction is chosen up front, from the risk-of-ruin and Kelly analysis. The (entry − stop) distance is set by the strategy's structure, not by the size — you do not move the stop to make the position bigger. The point_value is the dollar value of one point of price movement, which depends on the instrument.

A worked example, illustrative only and not a recommendation:

• Account equity: $10,000.
• Chosen risk_fraction: 0.5%.
• risk_per_trade: $50.
• Entry: 1.0850 on EUR/USD.
• Stop: 1.0830 (20 pips away).
• Standard pip value at $10/pip per standard lot: position_size = $50 / (20 pips × $10/pip per lot) = 0.25 lots.

The same trader, same setup, with risk_fraction at 2% (account equity $10,000, risk_per_trade $200), would position-size to 1.0 lots — a 4× bigger position for a 4× bigger drawdown when stopped out, a 4× bigger gain when target is hit, and a roughly 5–10× higher risk of ruin over a year of trading.

The choice of risk_fraction is the only parameter the trader sets in the size equation. Everything else flows from the strategy and the instrument. Treating risk_fraction as "the trader's most important number" is not metaphor; it is the literal mathematical truth of the system.

Correlation: the silent multiplier

Risk of ruin math, applied to a single position, gives a clean answer. Applied to a portfolio of simultaneous positions, the answer depends critically on correlation — and correlation is where most retail accounts hidden-blow-up.

Consider a trader holding three forex positions sized to 1% each:

• Long EUR/USD.
• Short USD/JPY.
• Long GBP/USD.

The trader believes they are running 3% total risk. The positions are not independent. EUR/USD and GBP/USD are typically 70–85% correlated; both are short USD by construction. Short USD/JPY is also short USD. On a USD-strength day, all three positions move adversely together. The effective risk on a high-correlation day is closer to 2.5–3% (because there is some diversification benefit) but the single-day drawdown is much closer to the sum of the worst-case losses than the trader expected.

The fix is to size total exposure, not per-position exposure, on the high-correlation scenario:

• Identify the dominant correlated factor (here: USD).
• Sum the directional exposure across positions to that factor.
• Cap that sum at the per-trade risk you would tolerate on a single position to that factor.

For multi-asset portfolios — equities and commodities and forex held simultaneously — the analysis gets harder but the principle is the same: risk is per dominant factor, not per position. We cover the asset-class-specific framing in the asset selection framework.

Daily, weekly and monthly loss limits

Beyond per-trade risk, the math implies a structure of cumulative loss limits. A trader who can take five losses in a row at 1% per trade has lost 5% of equity that week. The risk-of-ruin math shows that streaks of this length and worse occur regularly; the question is whether the trader is allowed to keep trading after them.

The discipline we teach inside the curriculum:

• Daily loss limit. Typically 2–3 times per-trade risk. After hitting it, the platform is closed for the day. No exceptions, no "one more setup."
• Weekly loss limit. Typically 5–8 times per-trade risk. After hitting it, the trader stops for the week and reviews the journal before resuming.
• Monthly drawdown threshold. Typically 10–15% of equity. After hitting it, the trader reduces per-trade risk by half until the equity recovers a defined fraction of the drawdown.

The numbers are guidance, not gospel. The principle is non-negotiable: the trader is allowed to take the market's worst day occasionally, but is not allowed to compound their own worst day with continued discretionary trading. The cumulative limits exist because the alternative — discretion in the middle of a drawdown — is the most expensive discretion in trading.

The journal entry that makes the math real

Risk-of-ruin numbers stay abstract until they are journaled. Inside the desk, every trader keeps a per-session entry that includes:

• Per-trade risk fraction used.
• Total exposure to dominant factor at maximum heat.
• Number of trades taken vs. number of triggers seen.
• Cumulative P&L for the day, week, month.
• Distance to daily/weekly/monthly limits.
• A single sentence on whether any of those numbers were almost breached.

The "almost breached" line is the single most useful field. It catches the days when the trader was approaching a limit and stayed inside it because the market saved them, not because the discipline did. Those are the days that produce the next blow-up if not caught.

The journal template lives in the trading journal and post-mortem template. The deliberate-practice arc that builds journal discipline lives in the ninety-day deliberate practice plan.

You will not blow up because you took a bad trade. You will blow up because you took a bad trade after taking three more bad trades that day, all sized as if each were independent.

— The Tradoki desk note

What the math looks like for typical retail strategies

To make this concrete, here are rough risk-of-ruin numbers for some common retail strategy archetypes, computed against a 50% drawdown ruin point over a 500-trade horizon. These are illustrative; your strategy's parameters will differ.

• Trend-following on majors, 40% win rate, 2.5R payoff, 1% per trade. Risk of ruin: low single digits. The high payoff ratio compensates for the low win rate. Sustainable.
• Trend-following on majors, same parameters, 3% per trade. Risk of ruin: high teens to low twenties. The same edge is now sized into the danger zone.
• Mean reversion on indices, 65% win rate, 0.7R payoff, 0.5% per trade. Risk of ruin: very low. The high win rate plus the small per-trade risk produces a robust system. (See: mean reversion on forex and indices.)
• Mean reversion on indices, same parameters, 2% per trade. Risk of ruin: mid teens. The small payoff ratio means individual losers are 1.5R wins lost; oversizing is punishing.
• Breakout on volatile single names, 35% win rate, 3R payoff, 1% per trade. Risk of ruin: low to mid single digits. Acceptable.
• Breakout on volatile single names, same parameters, 5% per trade. Risk of ruin: 30–40% range. Bordering on coin-flip on whether the account survives.
• News scalping with retail execution, 50% win rate, 1R payoff, 1% per trade. Risk of ruin: depends entirely on whether the win rate is real after slippage. Often the live edge is negative; risk of ruin asymptotes to 1. (See: the news fade.)

The pattern: the number that decides survival is the per-trade risk, more than the strategy. Two traders running identical strategies at 1% versus 3% risk are running materially different probability distributions of outcome.

What to do today

If you are reading this and have never done the math on your own strategy, the steps are:

1. Calculate your strategy's win rate and average win-to-loss ratio from at least 100 prior trades, or from a paper-trading sample if you do not have a live history.
2. Plug the numbers into the classical risk-of-ruin formula for your chosen per-trade risk and your chosen ruin point. Spreadsheet, calculator, anything.
3. If the resulting risk of ruin is above 5%, halve your per-trade risk and re-run.
4. If the strategy has negative expected value (A in the formula above is negative), no per-trade size makes it work. Stop trading the strategy.
5. Apply the correlation cap if you trade more than one position at a time. Total exposure to the dominant factor is the number that matters.
6. Set the daily, weekly and monthly loss limits. Write them in the journal. Make them automatic.

This is thirty minutes of work. It is the highest-leverage thirty minutes a trader will ever spend. We cover it explicitly in week one of the eight-week curriculum because every other lesson sits on top of it.