How much of your account should you risk per trade in crypto?

DarkTrade Research

TL;DR — Risk per trade is the share of your account a stop-out costs, set before you enter. The rules of thumb say 1% or 2%; our product allows 1% to 5% and defaults to 2%. The number only holds if size comes from the stop and you count open positions: on 2026-10-02, taking every signal in a package at its default could put 9% to 39% at risk at once.

How much to risk per trade in crypto is a number you choose before you enter, and it measures one thing: what a stop-out costs, as a share of your whole account. The common rules of thumb are 1% or 2% of the account on each trade. Our product lets you set anything from 1% to 5% and starts at 2%, which on its own example account of $1,000 means a stop-out costs $20.

The percentage is the easy part. A risk number only means something when two other things are true: the position is sized from the stop, so a stop-out costs exactly that share rather than roughly it, and you have counted how many positions will be open at the same moment. This post works through both, with the arithmetic of a losing streak and what we measured across 199 live whales on our board on 2026-10-02.

What does "risk per trade" actually mean?

Risk per trade is the loss you take if your stop is hit, written as a percentage of the whole account. It is not the position size, and it is not the margin you post. The position size is the dial you turn so that the loss at the stop comes out at the number you chose.

On the product's example account of $1,000, a 2% setting means a stop-out costs $20. A 1% setting makes the same stop-out cost $10, and a 5% setting makes it $50. The position behind that $20 can be large or small, because a stop close to the entry allows a bigger position and a distant stop forces a smaller one, but the cost of being wrong stays fixed.

The gap between position size and risk is where most confusion starts. A trader who says "I put 10% of my account into this" has described a position, not a risk. Without a stop, the loss on that position is not a number anyone chose; it is whatever the market does before the trader acts.

How much have large accounts lost on a single position?

Large accounts do not agree on one number. On 2026-10-02 our board had 199 live whales, Hyperliquid accounts held to a $100,000 minimum that is re-checked every night, and the two biggest groups sat at opposite ends: 60 had a worst single-position drawdown under 5% of the whole account, and 60 had one of 20% or more.

Every wallet on the board is grouped the same way. A whale's colour comes from one number: the deepest drawdown any single position put on its whole account, spot and perps together, measured from its own on-chain history. A wallet gets a colour only after at least 50 measured positions, because below that a lucky streak and a real edge are indistinguishable.

The colour table records what happened, not what each whale intended. A colour does not measure the loss a whale planned to accept on a trade; it measures how far the whale's worst position actually went against the account. A colour is a measured drawdown band, not a grade, and it says nothing about how any coming week will go. The full method is in how we measure whales.

The colour split has two consequences for your own number. The 1% and 2% rules are a choice, not a law that large accounts obey: wallets held to the same $100,000 bar have lived with worst single-position drawdowns from under 5% to 20% or more. A whale's colour also does not have to become yours, because a followed position is sized from a stop placed outside that whale's own worst case; a wider worst case means a smaller position, and a stop-out still costs the share you chose.

Why do the 1% and 2% rules exist?

The 1% and 2% rules exist because losses compound against a shrinking account, and the gain needed to recover grows faster than the loss that caused it. A small setting keeps an ordinary losing run survivable; a large one turns the same run into a long climb back.

The table below is plain arithmetic, not a measurement. It shows ten stop-outs in a row, each one taking its percentage from whatever is left.

Each loss is taken from a smaller account than the one before, so ten losses at 2% leave 81.7% rather than exactly four-fifths. The recovery column is the harsher half of the table. After ten stop-outs at 5%, the next trades have to make +67.0% just to return to the starting balance; after the same streak at 1%, they need +10.6%.

Ten is not a forecast. A streak of ten is a round number that makes the compounding visible, and losses are routine even among the large accounts on our board: across our live whales on 2026-10-02, the median ninety-day win rate was 60.5%, with a median of 82 positions behind each rate.

The cost of a small setting is speed. A 1% setting shrinks the winners along with the losers, so the account moves slowly in both directions. The rules of thumb accept that on purpose: they give up some upside so that ten straight stop-outs cost about a tenth of the account at 1%, rather than about two-fifths at 5%.

Why is risk per trade meaningless without a stop?

A risk percentage is only real when it is tied to a price at which you are out. The stop turns "2%" from an intention into a position size, and the size has to be derived from the stop, not chosen alongside it.

Our sizing derives the size from the stop. The position, written as a share of the account (its notional), is computed from the stop distance so that notional% × stop distance% ÷ 100 equals your risk%. A stop-out therefore costs exactly the percentage you chose: not about, exactly. The exactness covers the price move to the stop; exchange fees and any slippage come on top of it.

Where the stop sits matters as much as the sizing. For each signal the stop sits outside the worst loss the whale has genuinely sat through, plus one point, so the stop is not sitting inside the ordinary noise of how that trader already trades. A stop inside that noise gets hit on moves the trader itself holds through, and our guide to stop-loss strategy using on-chain whale data makes the same argument from the chart side.

The trader's win rate does not change the size; the trader's worst case does. In the sample behind our methodology note (published 2026-09-02; ninety-day window; each whale's own account), wallet A had a 52.4% win rate and wallet B 78.1%, yet A's worst case (its deepest drawdown on one position or its worst realized loss on one trade, whichever is larger, plus that one point) was 8.7x tighter than B's. At the same risk setting, an entry of the same size buys A 8.7x the position it buys B. Win rate does not enter the calculation at all.

What is concurrent risk?

Concurrent risk is the share of your account at risk across every position open at the same moment. A 2% setting risks 2% of the account on each position; with ten positions open, 20% of the account sits behind stops at once, and the per-trade number no longer describes your exposure.

Our package cards state concurrent risk outright. Each card multiplies the package's member count by its default risk, which assumes one open position per member at the same time. The figures below were measured on 2026-10-02.

Each package card also states the figure in words. The red package reads: "You choose which signals to take. Taking every one at 3% could put up to 36% of your account at risk at once." The figure is what the defaults allow, not a forecast of what will be open on a given day.

Open positions are not independent coin flips. A broad market move can push several of them toward their stops in the same session, which is why the concurrent figure is the one to measure your tolerance against. Three longs on related coins can behave like one large position, a point our guide to using whale alerts without overtrading makes.

How do you decide how much to risk per trade?

Choose the loss you could take ten times in a row without changing how you trade, then check it against the positions you will hold at the same time. Set the number before the next signal arrives.

Start from the streak, not the win

The losing-streak table is the honest starting point. At a 2% setting, ten straight stop-outs leave 81.7% of the account; at 5%, they leave 59.9% and the climb back needs +67.0%. If the 5% row would make you abandon your plan halfway through, a 5% setting is too high for you, however good the upside looks.

Multiply by what will be open at once

Your setting times the positions you expect to hold is the number that decides a bad week. At the package defaults, that figure ran from 9% to 39% of the account on 2026-10-02. If the figure for your own setting is more than you could watch go to its stops, take fewer signals or lower the setting; either one reduces it.

Keep it independent of how good the trader looks

A high win rate or a strong month is not a reason to raise your setting. The sizing already responds to the trader through the stop: a wider worst case gets a smaller position at the same setting. A trader who looks strong still has losing trades: even the highest win rates on our board leave losses in the record. A higher setting makes every stop-out on that trader cost more.

Know what the number does not do

Risk per trade bounds the cost of being wrong once. It does not make the trader right, and it does not decide when you take a winner off. We never place an order and never hold funds, so the exit is yours; why copy traders lose money even when the trader wins is our measured case for treating the exit as a decision of its own.

In our product the setting runs from 1% to 5% and starts at 2%. Each package card starts from its own default, 1% for white, 2% for green and yellow, 3% for orange and red, and the trade-off at every level is the one the streak table shows: a higher setting moves the account faster in both directions.

---

Every figure above is a whale's own realized result on its own account, over closed trades only. Peak is the most a trade was up before it closed — nobody captures it, and it is never added to anything. A colour is a measured drawdown band, not a promise. How we measure.