Capital at Risk | Trading Glass

Capital at risk is the dollar amount of account equity you stand to lose on a single trade if your stop is hit. It's the single number that determines whether your edge has time to play out — or whether one losing streak ends your career.

Forget the one-size-fits-all “2% rule.” Real risk sizing is personal, adaptive, and math-based.

Introduction

You’ve probably heard this advice before:

“Only risk 1–2% of your capital per trade.”

It’s common. It’s simple. But for serious traders?

It’s dangerously oversimplified.

Risking capital isn’t about following a rule — it’s about:

Your edge’s volatility
Your drawdown tolerance
Your win rate and R:R
And most importantly, your psychological bandwidth

Let’s break down how to size trades intelligently — using math and mindset, not memes.

What “Capital at Risk” Actually Means

The amount of your account equity you’re willing to lose on a single trade if your stop is hit.

For example:

Account = $10,000
Risk per trade = 1% → you lose $100 if stopped
Risk per trade = 3% → you lose $300

This number defines your survivability.

Risk too much → you blow up. Risk too little → your edge takes years to play out.

Start With Your System’s Stats

To size correctly, you must know:

Your expected value (EV — average R-multiple per trade; +0.4R means each trade earns 0.4× your stop distance on average)
Your win rate
Your max drawdown (historical and simulated)
Your streak volatility (max losers in a row)

These are point estimates. With 100 trades, a 42% win rate has a ±10pp 95% CI. Size as if your true edge is the lower bound of that CI, not the point estimate.

Example:

EV = +0.4R per trade
Win rate = 42%
Max drawdown = 12%
Worst losing streak = 6 trades

You wouldn’t want to risk 5% per trade… Because 6 straight losers would wipe 30% of your account.

That’s statistical suicide.

Plug the stats in: target max DD = 10%, expected worst streak ≈ 6 → cap per-trade risk ≈ 10% / 6 ≈ 1.6%. Cross-check with half-Kelly: f* ≈ (0.4·0.42 − 0.58·1.0) / 1.0 ≈ negative for R=1, so the system needs avg-win ≥ 1.7R to be Kelly-positive at all. Result: 1% is the honest ceiling here.

How to Pick Your Trade Risk %

Fixed-% sizing assumes your edge and stop distance scale linearly with equity. They don't. Vol-targeting (size = k·equity / σ_asset) and CVaR-budgeting (size so 95% tail-loss ≤ X% of equity) are the math-honest alternatives. The bands below are training wheels — graduate to one of those once you have 200+ trades.

Conservative = 0.25%–0.5%

For volatile strategies, low confidence, or live forward testing
Designed for survivability, not speed

Standard = 0.75%–1.5%

Works for stable strategies with 100+ trades of data
Good EV + moderate drawdowns → 1% is solid

Aggressive = 2%–3%

Only for high Sharpe/Sortino systems with deep journaling
Not recommended unless your system can handle it with data to prove it

Danger Zone = 3%+

Multiplicative truth: equity after N consecutive losses at risk r is E·(1−r)^N. At r=5%, N=5 → 77.4% remaining (−22.6%, not −25%). At r=10%, N=10 → 34.9% remaining. Linear approximations under-state ruin; always compound.
Compounded: 5 losses at 5% = −22.6%, 10 losses = −40.1%. At 10% risk a 10-loss streak (which a 42%-WR strategy hits ~0.6% of the time over 100 trades — non-trivial) leaves 34.9%. Recovery from −65% requires +186%. That's why fat-tail variance, not slogans, sets the upper bound.
Scaling this high requires strict Max Drawdown Rules and capital buffers.

Equity remaining after N consecutive losses at fixed per-trade risk r. Compounded, not linear.

r=1%r=2%r=5%r=10%

Prop traders: firm max-DD limits cap your sizing independent of edge. If the firm's daily DD is 3%, your per-trade risk × expected daily losing-streak length must stay well below 3% — typically <0.5%.

Framework	Inputs needed	Adapts to vol?	Adapts to edge?	When to use
Fixed-%	risk %, stop	No	No	Beginner / unstable edge
Half-Kelly	win rate, avg R	No	Yes	Stable edge, 200+ trades
Vol-target	σ of returns	Yes	No	Multi-asset, regime-shifting
CVaR-budget	tail dist.	Yes	Partial	Pro / max-DD constrained

Position Sizing Formula (in $)

Risk $ = E × r

Risk $ = dollars at risk on the tradeE = current account equityr = fractional risk per trade (e.g., 0.01 = 1%)

Then convert it into lot size / contracts / position size based on your stop-loss distance.

Per-trade risk is not portfolio risk. If you hold 5 BTC-correlated longs at 1% each, a single liquidity flush takes ~5% — not 1%. Cap total open risk (sum of distance-to-stop × size, weighted by pairwise correlation) at 2–3× your single-trade budget.

Example (BTC Futures):

Account: $25,000
Risk per trade: 1% → $250
Stop distance: $500
Position size: $250 / $500 = 0.5 BTC

Smart Add-ons

Combine With Win Rate & R:R

Use a risk % that allows:

Realistic recovery after expected drawdowns
Confidence in execution over 50+ trades
Scaling up gradually based on performance, not emotion

Optional: Use Kelly Criterion as a ceiling

Kelly fraction f* = (b·p − q) / b, where p = win rate, q = 1 − p, b = avg win / avg loss. With EV +0.4R and 42% win rate, full Kelly ≈ 8% — far above any sane per-trade risk. Half- or quarter-Kelly survives finite-sample edge error and non-stationary vol. Kelly assumes you know p and b; you don't, so haircut hard.

Calculate Kelly %
Risk ½ Kelly or ⅓ Kelly to stay emotionally stable

See the dedicated Kelly Criterion lesson in this curriculum for the derivation and a fractional-Kelly worked example. Further reading on the source material: Ralph Vince, The Mathematics of Money Management (1992) for optimal-f; Aaron Brown, Red-Blooded Risk (2011) for institutional sizing; Van Tharp, Trade Your Way to Financial Freedom for R-multiples.

BONUS: Use a “Psychological Pain Scale”

Rate each risk level on a 1–10 scale:

“How would I feel if I took 5 losers in a row at this risk %?”

If answer = “numb, focused” → probably fine
If answer = “I’d tilt or revenge trade” → it’s too high

Procedure: simulate 1,000 paths of your strategy at risk = X% (Monte Carlo on historical R-multiples). For each X, record the 5th-percentile drawdown. Pick the largest X whose 5th-percentile DD is still below your stated tilt threshold. The 'pain scale' is just a heuristic anchor for that threshold — see Behavioral Risk Management for the journaling protocol that makes this honest.

The right % isn’t what works on paper — it’s what keeps you consistent in real execution.

Final thought

Your per-trade risk % is one input into a system. This is lesson 1 of 9 in Risk Management. The next lessons make it adaptive: Max Drawdown Rules caps the cumulative damage, Daily & Weekly Risk Limits cap the velocity, Building a Tiered Risk Model lets confidence shrink and grow your size honestly, and Recovery Factor plus Ulcer Index measure the cost of getting it wrong. Pick your starting number from the bands above, then let the rest of this module make it dynamic.

What is capital at risk?

Capital at risk is the dollar amount of account equity you stand to lose on a single trade if your stop is hit. It is calculated as account balance × risk % per trade.

Is the 2% rule still a good guideline for traders?

As a beginner ceiling, yes — it caps damage while you build a track record. As a final answer, no: it ignores edge volatility, drawdown tolerance, win rate, payoff ratio, and correlation across simultaneous positions.

How do you calculate position size from a stop loss?

Position size = (account equity × risk %) / stop distance. Example: $25,000 account at 1% risk with a $500 stop → $250 / $500 = 0.5 BTC.

What risk percentage is too high per trade?

Anything above 3% is the danger zone. Compounded, 5 consecutive losses at 5% risk leave 77.4% of equity (−22.6%); 10 consecutive losses at 10% risk leave 34.9% (−65.1%), and recovery from there requires a +186% gain.

Should you use the Kelly Criterion for position sizing?

Use it as a ceiling, not a target. Full Kelly assumes you know your true edge exactly, which you don't. Half- or quarter-Kelly approximates the same long-run growth with a fraction of the drawdown and absorbs the inevitable estimation error in p and b.