Risk is the only thing you control.

1 · Think in R, not in money ↑ top

R is your risk on a single trade — the distance from entry to stop, in money. Every outcome is then measured in multiples of it. This one change of unit makes results comparable across instruments, position sizes and account sizes, and it strips the emotion out of a number like "−£340".

2 · Expectancy — the number that decides everything ↑ top

# Expectancy per trade, in R
Expectancy = (WinRate × AvgWin_R) − (LossRate × AvgLoss_R)

# Example: 45% win rate, winners average +2.1R, losers average −1.0R
= (0.45 × 2.1) − (0.55 × 1.0) = 0.945 − 0.55 = +0.395R per trade

# Over 20 trades/month → +7.9R/month. At 1% risk per trade → ~+7.9% monthly.

That last row is the most common losing profile in retail trading: cutting winners early and letting losers run past the stop. High win rate, negative expectancy. It's why win rate alone is a vanity metric.

3 · Position sizing ↑ top

# The only sizing formula you need
Size = (Account × Risk%) ÷ (StopDistance_points × PointValue)

# Worked: £25,000 account, 1% risk, GER40 long, stop 22 points away,
#         point value £1 per point per contract
RiskBudget  = 25000 × 0.01 = £250
Size        = 250 ÷ (22 × 1) = 11.4 → round DOWN to 11 contracts
ActualRisk  = 11 × 22 × 1 = £242  (0.97% — always round down, never up)

4 · Drawdown — the asymmetry nobody feels until it's late ↑ top

This asymmetry is the entire argument for hard loss limits. A −10% month is a bad month you trade out of; a −50% month is a different career. The defence is mechanical:

• Daily stop: −3R (or −3% at 1% risk). Hit it and you're done for the day — platform closed, not "one more to get it back".
• Weekly stop: −6R. Hit it and you're done for the week; spend the remaining days reviewing the journal instead.
• Drawdown de-risking: at −10% from equity high, halve position size until you've recovered half of it. This mathematically slows the bleed exactly when your judgment is worst.
• Recovery is a process problem, not a size problem. The instinct to increase size to "win it back faster" is precisely what converts a −10% into a −50%.

# Monte Carlo: what does MY edge actually feel like over a year?
import numpy as np

def simulate(win_rate=0.45, avg_win=2.1, avg_loss=1.0,
             risk_pct=0.01, trades=250, runs=10_000, seed=42):
    rng = np.random.default_rng(seed)
    finals, max_dds, ruined = [], [], 0
    for _ in range(runs):
        equity, peak, max_dd = 1.0, 1.0, 0.0
        for _ in range(trades):
            r = avg_win if rng.random() < win_rate else -avg_loss
            equity *= (1 + r * risk_pct)          # compounding, fixed-fractional
            peak = max(peak, equity)
            max_dd = max(max_dd, 1 - equity / peak)
        finals.append(equity); max_dds.append(max_dd)
        if equity < 0.5: ruined += 1          # "ruin" = lost half the account
    return {
        'median_return': np.median(finals) - 1,
        'p10_return'   : np.percentile(finals, 10) - 1,   # the bad-luck year
        'median_maxDD' : np.median(max_dds),
        'worst_maxDD'  : np.percentile(max_dds, 95),
        'prob_halving' : ruined / runs,
    }

print(simulate())                       # baseline
print(simulate(risk_pct=0.03))         # same edge, 3% risk — watch the tail explode
print(simulate(win_rate=0.40))         # edge decays 5 points — still viable?

5 · Correlation — when three trades are really one ↑ top

Risking 1% each on US500, US100 and GER40 longs is not 3 × 1% of diversified risk — it's roughly one 2.5% bet on "equities go up". Correlation is the hidden way disciplined sizing turns into oversizing.

• Cluster rule: cap total risk per bloc at ~1.5× your single-trade risk. Two index longs? Take 0.75% each rather than 1% each.
• Silver and Gold are one trade in the same direction — pick the better chart rather than doubling up. Silver's extra volatility means the pair is effectively a leveraged gold position.
• Watch the hidden common factor: a hot US CPI print hits equities, metals and oil simultaneously through the dollar/yields channel. On big data days, correlations converge toward 1 — exactly when you're most exposed.
• Total heat limit: never have more than ~3R of open risk across the whole account, regardless of how many good setups appear at once. Good setups cluster precisely because one macro driver is moving everything.

6 · Trade management — partials, trails and the exit problem ↑ top