Flat 1 percent on every trade is the right way to start, and for plenty of profitable careers it is also the right way to finish. But a mature journal eventually shows something worth acting on: your best setup earns two or three times the expectancy of your average one, and you are paying both the same salary. Sizing by setup quality is the legitimate version of the confidence every trader already feels. The catch is that it only works when the confidence is written down and audited, not felt in the moment.
Earn the right first
Tiered sizing is an upgrade, not a foundation. The prerequisites are strict: at least 100 journaled trades with per-setup expectancy calculated (the expectancy article covers the method), and several months of flat-fraction sizing with no violations. A trader who sometimes risks 3 percent on a feeling has nothing to tier; the variance of the rule-breaking swamps any signal from the grading. Sizing up by conviction without a written basis is just impulse wearing a lab coat. The flat-sizing period is not wasted time either; it is the control group your tiers will be measured against later.
A three-tier framework
Grade every setup against a short written checklist created when calm. Four binary criteria work well for intraday trading: aligned with the higher-timeframe trend, located at a pre-marked level, occurring in the setup's proven session, and clean structure with no imminent scheduled news. Then the tiers on a $10,000 account: four of four is an A trade at full risk, 1 percent, $100. Three of four is a B trade at 0.6 percent, $60. Two or fewer criteria is not a discount trade; it is no trade. Two details keep the system honest. The grade is computed before entry, from the checklist, and if you cannot name which criterion is missing, it is not an A. And the tier ratios stay modest: the best trade risks less than double the ordinary one, because grading is an estimate, not a prophecy. Audit the tiers quarterly against results: if B trades outperform A trades over a decent sample, the checklist is measuring the wrong things, and the fix is rewriting criteria, not shuffling percentages. Grading systems drift, and only the journal notices.
Size down when conditions degrade
The second layer is regime, and it multiplies the tier. Halve size when the day's fuel is spent: if EURUSD averages a 60-pip daily range and 55 pips are already traveled by early New York, continuation trades are fighting arithmetic. Halve it in holiday liquidity, when spreads widen and structure gets noisy, and in the minutes after major news while spreads normalize, and on your first day back from any break. An A-grade setup in a degraded regime is a 0.5 percent trade. This one habit, cutting size in chop instead of grinding full risk through it, quietly removes a whole category of death-by-a-thousand-cuts weeks. The multiplier stacks with the tier and rounds down: a B trade in thin liquidity computes to 0.3 percent, $30, and is more often simply a pass, because a discounted trade in a discounted market rarely justifies the attention.
The direction you never scale
Size never increases in response to losses. The seduction is always the same: down $300, double the next trade, one winner makes it all back. Follow the doubling ladder from $100 and watch it: $100, $200, $400, $800, $1,600. Five straight losses, which a 45 percent win rate delivers about once every 20 sequences, now costs $3,100, or 31 percent of the account, requiring 45 percent just to recover. The ladder converts a routine streak into the recovery table's worst rows. Losses move size in exactly one direction, down, on the step-down schedule from the drawdown article. Sizing up is reserved for new equity highs and setups whose graded expectancy has earned it in the journal. Watch for the stealth version too: normal size but doubled frequency after a loss, forcing marginal trades to hurry the recovery. The ledger reads exactly the same.