Tight stop: 1-2%
Who: High-frequency intraday, 5-minute charts
Cost: Win rate drops to 30-40%, you get shaken out by noise often, but R per trade stays small
Mid stop: 3-5%
Who: 1h - 4h swing positions
Cost: 45-55% win rate balance, mainstream choice; requires a defined setup trigger
Wide stop: 8-15%
Who: Daily-trend positions, long holds
Cost: 55-65% win rate, rarely shaken, but you must size down to 1/3-1/4 to control R
BTC routinely swings ±0.5% inside an hour. A 0.5% stop gets hit by random noise almost immediately after entry. Tight stops surrender the time needed for a directional thesis to play out — win rate drops from 50% to 30% or worse.
Max single-trade loss is 1% of capital in both — looks equivalent on the surface. But the 1% stop's win rate runs 15-20 percentage points lower than the 2% stop, so expected value diverges. At fixed 1% risk-per-trade, the 2% stop / 50% size combo dominates long-term EV.
No. Expected value = 0.5 × 1R - 0.5 × 1R = 0, and after fees and funding it goes negative. To make money you need win rate above 50% at equal payoff, or payoff above 1 at equal win rate. Strict stop-and-take alone is not enough — the math has to work.
The assumption buried in "tight stop, big size"
The pitch goes like this. If you put your stop 0.5% away and size up so that 0.5% is exactly 1% of your margin, you can run a much bigger notional. When the trade works, the payoff is bigger because the position is bigger. When it fails, you only lose your 1%. Sounds like a free lunch.
It is not. The hidden cost is hit rate. A 0.5% stop sits inside the normal random noise of BTC's price. Most of the time, the market is going to wiggle through your stop before it walks in the direction you want. Your stop-outs go up. Your gross win rate falls. The "bigger position when it works" doesn't fire often enough to compensate.
This is the paradox in one paragraph: tighter stop, bigger size — symmetric outcomes on a single trade, but asymmetric outcomes over a sample. The market does not pay you for tighter stops.
The tighter the stop, the more often noise takes you out
BTC perp 24h realised volatility averages around ±2-3%. That means inside any given trading day, BTC moves through that range with normal-distribution-ish frequency, without anything special happening.
If your stop is 0.5%, you are essentially betting that the next price tick goes your way before the random walk takes you out. Empirically on BTC perp this fires negative against you maybe 60-70% of the time, just from noise.
Widen the stop to 2-3% and the noise stops eating you. Now the win rate reflects your directional edge instead of random fluctuation. Same trader, same view, same setup — the win rate jumps 15-20 points just by giving the stop room to breathe.
For the textbook framing of stop placement, Investopedia's stop-loss entry is the canonical reference. TradingView's Help Center — search "stop order" — covers the practical placement on the chart.
Risk per trade is stop width multiplied by position size
The cleanest model is the "1R" framework. Define one R as your maximum loss on any single trade — typically 1% of your account.
If your stop is 1% away and your position is your full margin, your max loss is 1% of margin. R = 1%. If your stop is 2% away and your position is 50% of margin, your max loss is still 1%. R = 1%. If your stop is 5% away and your position is 20% of margin, R = 1%.
All three setups are identical in terms of "money at risk", but they are not identical in terms of probability that R gets hit. The 1%-stop version takes the most stop-outs. The 5%-stop version takes the fewest. That is the trade-off.
Nine ways to pair a stop with a position size
Below is the matrix. Each cell shows R (% of margin at risk), assuming BTC perp at 10x leverage. Read the rows: tighter stops let you scale the position bigger for the same R; the cost is in the hit rate column you cannot see directly here.
Stop 1% — size 30%: R = 0.3% — size 50%: R = 0.5% — size 100%: R = 1.0%
Stop 2% — size 30%: R = 0.6% — size 50%: R = 1.0% — size 100%: R = 2.0%
Stop 5% — size 30%: R = 1.5% — size 50%: R = 2.5% — size 100%: R = 5.0%
Reading: anywhere R lands above 2% you are running hot and a few consecutive losses can compound into double-digit account drawdown. Anywhere R lands below 0.5% you are probably under-sizing and your account barely moves on winning trades. The 0.5-2% R band is the working zone.
Rewrite the journal in R and wide and tight stops become comparable
The R-multiple framework rewrites your trade log so different stop strategies become comparable. Each trade gets stamped with two numbers: how many R you risked, and how many R you actually made or lost.
A 1% stop, 100% size trade that hit +2% target = +2R. A 5% stop, 20% size trade that hit +4% = also +2R. From an EV standpoint they are equivalent. The journal exposes which setups consistently produce +1R, +2R, +3R outcomes, regardless of how you sized them.
For a working trader, three statistics drive everything: trades per month, average R outcome, win rate. Multiply those three and you get monthly account return. Anything else (your stop policy, your leverage, your entries) is downstream of those three numbers.
Win rate, payoff and risk per trade decide expected value together
The expected value of a strategy is just the three variables multiplied together. EV = win_rate × avg_win − (1 − win_rate) × avg_loss. Plug in any combination and you can see whether the strategy is positive or negative over the long run.
50% win rate, 1.0 average payoff: EV = 0.5 × 1 − 0.5 × 1 = 0. Break-even gross, and after fees and funding you are net negative.
55% win rate, 1.5 payoff: EV = 0.55 × 1.5 − 0.45 × 1 = 0.825 − 0.45 = 0.375 R per trade. Over 100 trades that compounds to a meaningful account return.
The variable most retail traders try to fight is the wrong one. They squeeze the stop to chase a bigger payoff per trade, and watch the win rate crater. The math says: leave the stop where realised vol is comfortable, let the win rate stabilise, accept the payoff multiple the market gives you.
A month of trades on one sheet, and the four numbers to read first
R-multiples only pay off once a full month of trades sits on one sheet. So this section is about building that sheet — the numbers in it are yours to fill in.
Columns you cannot skip. Date, direction, entry, stop (that is what fixes 1R in dollars), exit, PnL, PnL ÷ R for the R-multiple, and one plain-language column for why the trade ended: stop hit, target hit, closed by hand, or trailed out. That last column is the one everybody skips, and it is the only one that tells you whether the loss came from the strategy or from you.
Four numbers to read at month end. Share of winning trades (win rate); average winning R divided by average losing R (payoff); cumulative R; and the longest run of consecutive losses in R. Feed the first two into the EV formula from the previous section and you have your per-trade expectancy. The third is what the month actually delivered. The fourth tells you how much buffer the account needs.
Take the 55% win rate and 1.5 payoff from the section above as an illustration: +0.375R per trade, 40 trades a month, +15R on paper — with R at 1% of the account, that reads as +15% for the month. The reason it looks absurd is that it treats three variables as constants. In a live account the win rate drifts with the regime, slippage and funding come off the top, and monthly variance runs several times the expectancy, so a genuinely positive-EV system will still hand you two or three losing months in a row. The sheet is not there for you to hit the paper number. It is there to show you how far off it you are, and whether the gap sits in the win rate or in the payoff.
Sooner or later the log will hold a trade worse than the R you set: price gaps straight through the stop and fills well beyond it, and the trade books −1.5R or worse. Record it at the fill, not at the planned stop. R is your intended risk, not a guarantee of execution, and the log is the only thing that keeps reminding you of the difference.
One more thing you can only see once the log exists: in almost any month, a handful of trades drives the entire result. Strip out the three biggest winners and a good month often goes flat. That is normal, and it is why sizing the tail properly matters far more than fine-tuning the bulk.
The paradox does not go away, so pick a band and live in it
You cannot wish the noise away. BTC perp has the intraday vol it has. The paradox is permanent. What you can do is the following.
Fix R first, derive size second. Pick a fixed R per trade (0.5-2%), then size the position based on stop distance. Stop distance is not a free variable — it has to clear realised vol.
Anchor stops to volatility, not psychology. Use the BTC daily ATR or a multiple of realised vol as the floor. Anything tighter than ATR × 1 is signal noise.
Track win rate by setup. Tag your trades with the setup that triggered them. Some setups will quietly run 60% win rates; others run 40%. The R-multiple journal tells you which.
For the broader textbook framing, Investopedia's risk-management techniques for active traders covers the same ideas across sizing, stops and risk budgets. From here, the natural next read is the Kelly Criterion piece — which formalises the win-rate × payoff product into an optimal position-size formula.