Most beginner accounts in perpetual futures do not fail because every idea was wrong. They fail because a few oversized losses made recovery mathematically hard, often while the trader was using leverage, ignoring funding or treating a signal as an order. This guide starts with the arithmetic that explains why, turns it into a handful of written rules, and lists ten common mistakes with the habit that prevents each.
Nothing here tells you whether a particular trade will work. The aim is to stay in the game long enough to learn whether your process does.
Why losses compound against you
A loss shrinks the base from which you have to recover. The gain needed to return to the starting balance after a loss L is 1 ÷ (1 − L) − 1.
| Account drawdown | Gain needed to return to even |
|---|---|
| −10% | +11.1% |
| −20% | +25.0% |
| −25% | +33.3% |
| −50% | +100% |
| −75% | +300% |
Check one row: after a 20% loss, 1 ÷ (1 − 0.2) − 1 = 0.25, a 25% gain. After a 50% loss, 1 ÷ (1 − 0.5) − 1 = 1, a 100% gain. The CFA Institute’s digest on drawdown research notes that drawdown “reflects only factual returns from a past period and has no direct predictive value.” The table is arithmetic about recovery, not a forecast of whether it will happen. The drawdown and recovery guide develops it.
Losing streaks at different risk levels
Even a sound process sometimes loses many times in a row. Here is what ten consecutive losses do when each loss is a fixed share of the current balance.
| Risk per trade | Balance after 10 losses | Gain needed to recover |
|---|---|---|
| 0.5% | 95.1% (−4.9%) | +5.1% |
| 1% | 90.4% (−9.6%) | +10.6% |
| 2% | 81.7% (−18.3%) | +22.4% |
| 5% | 59.9% (−40.1%) | +67.0% |
| 10% | 34.9% (−65.1%) | +186.8% |
Each loss multiplies the balance by one minus the risk: after two 10% losses the balance is 0.9 × 0.9 = 0.81 of the start, and after two 1% losses it is 0.99 × 0.99 = 0.9801. Repeating that 10 times gives 0.99^10 = 0.904382075, about 90.4% of the start at 1%, and 0.9^10 = 0.3486784401, about 34.9% at 10%. The same streak is a setback at one size and close to the end of the account at the other. How likely a streak is depends on your actual win rate and whether outcomes are independent; the losing streaks guide calculates that separately.
Seven written risk rules
These are example rules for a learning account, not universal standards. The point is to write yours down before the next trade.
- Risk a small fixed share per idea, such as 1% of the account, or 0.5% while learning. Size from the stop distance, not from a feeling.
- Set a daily loss limit, such as three full-risk losses, and stop for the day when it is reached.
- No plan without an invalidation. If you cannot say where the idea is wrong, there is no trade.
- Keep liquidation well beyond the stop. As a working rule, keep the liquidation price at least twice as far from entry as the stop; if it is not, lower the leverage.
- Do not add to a losing leveraged position to average down.
- Check funding for anything held longer than a day.
- Treat signals as prompts to look, never as instructions to act.
Hyperliquid’s margining documentation states that “Leverage can be set by a user to any integer between 1 and the max leverage.” The maximum is a limit the venue allows, not a size the account should use. Choosing the number of contracts from the loss you can accept, rather than from the largest position the margin permits, keeps one losing trade from becoming an account-level event.
Ten common mistakes and the habit that prevents each
| # | Mistake | What it looks like | Prevention habit |
|---|---|---|---|
| 1 | Treating a signal as an order | “The indicators lined up, so I went long” | Run the check, plan and alert loop first |
| 2 | Choosing leverage first | “20× because I’m confident” | Size from risk and stop distance; leverage comes last |
| 3 | Confusing an alert with a stop | “I had an alert at 144” | Place the stop order on the venue; an alert only notifies |
| 4 | Moving the stop further away | “It will come back” | Keep the original invalidation in a written record |
| 5 | Revenge trading | A bigger trade straight after a loss | A daily loss limit; record “no trade” as a decision |
| 6 | Chasing stretched moves | Entering after a large daily rise with RSI above 80 | Wait for the next setup with a defined invalidation |
| 7 | Ignoring funding | Holding a crowded position for weeks | Convert the rate to a daily cost before holding |
| 8 | Trading one-minute noise | Many tiny trades consumed by fees | Use longer timeframes while learning; compute the fee break-even |
| 9 | Holding through scheduled events without a plan | Surprised by a macro release | Check the event calendar and plan before, not during |
| 10 | Never reviewing | The same mistake every week | Record every decision and review weekly |
Mistake 3 deserves a closer look, because three different mechanisms are often confused.
An alert sends a message. A stop order is an instruction to the venue that may fill, partly fill or slip. Liquidation is the venue’s own process: Hyperliquid’s documentation says a liquidation occurs when “the account equity falls below the maintenance margin,” and advises that “traders can place stop loss orders or exit the positions before the mark price reaches the liquidation price.” The stop loss vs liquidation guide works through a gap that moves through both.
Mistake 7 is easy to underestimate. Hyperliquid’s funding documentation states that “The funding rate on Hyperliquid is paid every hour.” The same page explains that “funding is paid every hour at one eighth of the computed rate for each hour”, so a crowded position there pays or receives a funding payment every hour. For illustration only, assume that each hourly funding payment equals 0.002% of notional and stays the same for 30 days; this is an assumed input, not a quoted or typical figure. Under that assumption, and only as arithmetic rather than a sourced figure, 30 days would cost 0.002 × 24 × 30 = 1.44% of notional, and the cost keeps growing for as long as the position stays open. Before holding a position for weeks, check how often your venue charges funding and convert the rate to a daily cost.
Trading less is not a weakness
Barber and Odean’s study of 66,465 households with accounts at a large discount broker during 1991 to 1996 found that those that trade most earn an annual return of 11.4 percent, while the market returns 17.9 percent. That is a historical stock study, not a crypto or perpetual-futures result, but it is a useful reminder that activity is not the same as progress. A recorded “no trade” is a valid decision.
Practise before risking money
Use a paper account to rehearse the full loop, from signal to plan to review, and record your stated confidence for each decision. After a few weeks, compare your confidence with what happened. If they do not match, you learned something without paying for it. The signal workflow guide shows one way to set that up.
Hyperliquid’s liquidation documentation states that “A liquidation event occurs when a trader’s positions move against them to the point where the account equity falls below the maintenance margin.” Rules like the ones above exist so that one bad week cannot end the learning process. Educational material only; the percentages are illustrative guidelines, not recommendations for your situation.