発行済み
The 2%-per-trade rule assumes your trades are independent — that one going wrong tells you nothing about the others. That assumption holds beautifully in a textbook and almost never in crypto. When Bitcoin drops 5%, the majority of altcoins drop with it, and the higher-beta ones drop harder. The "diversification" of holding SOL, an L2, a memecoin and two AI tokens is mostly cosmetic: they're different tickers reading off the same risk-on/risk-off switch. So when the switch flips, they don't fail one at a time — they fail together, on the same candle, and your five neat 2% stops become one 10% hole.
Per-trade risk vs portfolio heat
Here's the distinction the rule hides. Per-trade risk is what you lose if this stop hits. ポートフォリオの熱 is what you lose if every stop hits — the sum across your whole open book. Watch how fast it climbs on a disciplined-looking 2%-per-trade account:
| Open longs @ 2% each | ポートフォリオの熱 | Loss on $5,000 |
|---|---|---|
| 1 position | 2% | 100ドル |
| 3 positions | 6% | $300 |
| 5 positions | 10% | $500 |
| 8 positions | 16% | $800 |
Nothing here breaks the 2% rule. Every single trade is textbook. Yet the eight-position trader is one bad session from a 16% drawdown — and because the positions are correlated longs, "one bad session" isn't a tail event, it's a Tuesday. A common desk guideline is to keep total heat under about 6%. The three-position trader is at the ceiling; the five- and eight-position traders have quietly walked past it while feeling responsible the whole way.
Why correlation makes it worse than the sum
The table above is actually the optimistic reading, because it treats the stops as if they might trigger independently. They won't. If all five positions are long and the market turns, they lose together — the diversification that was supposed to spread your risk does nothing, because there's only one risk factor underneath. In a sharp move, high-beta alts routinely fall 1.5–2× as far as Bitcoin, so a 6% BTC flush can be a 10–12% move in your alt basket and blow through stops you thought had room. The uncomfortable truth: five correlated longs behave like one position sized at the total heat, just with five sets of fees and funding attached.
The recovery tax nobody prices in
A 10% portfolio drawdown doesn't need an 10% gain to recover — it needs 11.1%, because you're now compounding off a smaller base. A 16% hole needs 19%. The deeper the heat lets you fall, the more asymmetric the climb back, which is exactly the drawdown math that grinds accounts down over a year of "small, disciplined" trades that all happened to be open at the wrong time. Heat isn't just today's risk; it's a claim on all your future winners.
What the math says to do
- Add up your heat before every new position. Sum the risk across all open trades. If the total clears ~6%, the next "great setup" is one you skip or size down — run it through the ポートフォリオ熱計算ツール 初め。
- Count correlated longs as one bet. Five altcoin longs share one risk factor. Treat their combined heat as a single position's risk, not five independent ones — and size accordingly.
- Use the correlated worst case, not the raw sum. Assume the same-direction cluster loses together, because in a real move it does. If that number scares you, you're over-exposed regardless of how tidy each individual trade looks.
- Genuine hedges reduce heat; more longs don't. A real offsetting short lowers net exposure. Adding a sixth long to "spread risk" just raises heat under a diversification label.
Put your open positions into the new ポートフォリオ熱計算ツール to see your true aggregate risk and the correlated worst case, size each trade from a sensible stop with the ポジションサイズ計算機, and sanity-check your bet fraction against the ケリー基準. The account that survives isn't the one with the tightest per-trade stop — it's the one that knows what all its stops add up to. That correlation risk gets worse, not better, when the whole market is leaning one way — see 23 of 24 pairs with both retail and top traders net long right now.
Method: portfolio heat = sum of per-trade risk (the loss if each stop is hit) ÷ account equity; recovery gain = 1 ÷ (1 − drawdown) − 1. Correlation and beta figures describe typical crypto risk-off behavior, not a live measurement — actual correlation varies by asset and regime and can spike toward 1.0 in stress. These are exact arithmetic illustrations, not a backtest; funding, fees and slippage make real outcomes worse, not better.
- ポートフォリオ熱計算ツール — your total open risk and the correlated worst case
- ポジションサイズ計算機 — size each trade from account risk and stop distance
- ケリー基準計算機 — the bet fraction your edge actually justifies
- 破滅のリスク計算機 — the odds a string of losses ends the account
- ドローダウン回収計算機 — the gain it takes to climb back