Full-size versus starter — and what each is for
Starter sizing exists for trades where the thesis is right but the entry is uncertain. You put on 25–50% of your intended size at the first valid signal, then add the rest only if the trade proves itself by moving favorably to a defined add point — never by moving against you. Full-size is reserved for trades where both thesis and entry are crisp at the moment of execution: the strategy fires cleanly, the conditions are A++, the structure on the chart leaves no ambiguity about where you are wrong. If you find yourself going full-size on a setup that "looks pretty good", you are using full-size as a substitute for conviction, which is a different and worse trade.
Why adding to winners is usually wrong
Adding to a winner sounds disciplined — "I am pyramiding into strength" — but the math usually betrays the intuition. The original entry was at the technical level; every subsequent add is at a worse price with a wider stop relative to the new average. After two adds, your effective stop is often well beyond what the original strategy allowed, and your R has silently doubled — R being an R multiple: profit divided by risk, so one R is the risk you planned to take. The exception is a discrete, pre-planned add at a confirmation level (e.g., the first higher-low after a breakout retest) where the new stop is structurally tighter. If the add is just "the trade is working, I want more", it is chasing dressed up as a system.
Scale-out only — preserving R
A simpler and more durable approach is to size correctly at entry and only scale out. Take partial at 1R, another partial at 2R, leave a runner with a trailing stop. This shape locks in R progressively, reduces the emotional cost of giving back open profit, and keeps a tail on the trades that extend. It also resolves the "should I add?" question — you do not add, you only reduce. For most discretionary traders, scale-out-only is the position-sizing model that survives the most market regimes.
R-based sizing math
The math is the same every time. Decide R (e.g., $500). Identify the stop distance from your planned entry (e.g., $0.40). Position size = R / stop distance = 1250 shares. Notional value is whatever falls out — $25 stock means $31,250 of buying power, $80 stock means $100,000. Notional does not drive the trade; risk does. The discipline is to never invert the formula: never start with "I want a $50,000 position" and back out the stop. That is how a 0.5R trade silently becomes a 3R trade because the stop moved.
Account heat across open positions
Heat is the sum of R at risk across every open trade simultaneously. A common rule: total heat capped at 2R to 3R at any one moment. If you have three positions on, each at 1R, you are at 3R heat and you take no new trades regardless of how good the next setup looks. This rule prevents the scenario where four "small" positions all hit stop together on a market flush and the day is suddenly down 4R. Heat caps are boring and they save accounts.
Correlation risk — three oil names equal one oil bet
Sizing is meaningless if you ignore correlation. Three long positions in different oil and gas names are not three independent 1R trades; they are one 3R trade on the energy sector. The same applies to small-cap momentum baskets, AI-adjacent tech, regional banks, and any thematic cluster. The fix is to count correlated names as one position for heat purposes. If you are already long XOM, the next energy long counts against XOM's R, not as fresh R. Most blowups in discretionary trading are correlation events, not single-trade losses.
Confidence-tier sizing
Tie size directly to your written grade of the setup. A practical scheme: A++ = 1R, A = 0.5R, B = 0.25R or paper only, anything below B = no trade. The tiers force the question "how good is this, really?" before sizing, rather than after the fact. They also create a long-run statistical record: if your A++ trades win 70% and your A trades win 50%, the sizing tiers automatically allocate more capital where edge is highest. Without tiered sizing, traders default to roughly equal size on every trade, which is mathematically equivalent to claiming every setup is equally good — which they are not.
Key takeaways
- Starter size when the entry is uncertain. Full size only when both thesis and entry are crisp.
- Adding to winners usually widens effective R. Prefer scale-out-only as a default.
- Position size = R divided by stop distance. Never invert the formula.
- Cap total open heat at 2–3R, regardless of how good the next setup looks.
- Correlated names share R. Three oil longs is one oil bet, not three.
- Tier size to setup grade: A++ = 1R, A = 0.5R, B = 0.25R or skip.