The single most important variable in my trading results isn't stock selection, entry timing, or even options strategy. It's position sizing. I've seen traders with brilliant market read destroy their accounts by sizing too large on a single bet. I've seen mediocre stock pickers generate consistent compounding returns because they never risked more than they could absorb.
My sizing system is mechanical, conviction-tiered, and has a built-in sanity check. Here's the complete framework.
The 2% Rule as a Starting Point
The 2% rule states that you should never risk more than 2% of your portfolio on a single trade. On a $100,000 portfolio, that's $2,000 max loss per position.
I use this as my base — but I modify it by conviction level.
Conviction Tiers: 1%/2%/3%
Every trade I put on gets classified into one of three conviction tiers before I size it:
Tier 1 — Low Conviction (1% risk):
- Ideas I like but don't have a strong view on
- Setups where I'm buying the technicals but the fundamental story is unclear
- Exploratory positions in new sectors I'm still learning
- Any position where I'm less than 65% confident in the thesis
Tier 2 — Medium Conviction (2% risk):
- Standard thesis where both technicals and fundamentals align
- Setups I've traded before in similar conditions
- Positions in sectors I understand well
- Confidence level between 65% and 80%
Tier 3 — High Conviction (3% risk):
- Strong thesis with multiple confirming signals (technical + fundamental + catalyst)
- Names I've researched deeply, ideally including reading the last two earnings calls
- Setups with a clear near-term catalyst (earnings, product launch, sector rotation)
- Confidence level above 80%
I never exceed 3% on any single trade. Even if I'm 95% confident. The reason: correlation. My high-conviction ideas tend to be clustered — similar sectors, similar macro themes. If I'm wrong, I'm often wrong on multiple positions simultaneously. The 3% ceiling ensures that even a cluster of four simultaneous losses doesn't destroy more than 12% of the portfolio.
The Kelly Criterion Sanity Check
After I classify a trade by conviction tier, I run a Kelly Criterion check as a sanity filter. The Kelly formula is:
Kelly % = W − (1 − W) / R
Where W = win rate and R = win/loss ratio (average win / average loss).
For a Tier 2 trade with my historical 68% win rate and average win/loss ratio of 1.6: Kelly % = 0.68 − (1 − 0.68) / 1.6 = 0.68 − 0.20 = 0.48, or 48% of portfolio
Full Kelly would say to risk 48% of my portfolio on this trade. That's obviously insane. Full Kelly is theoretically optimal for maximizing long-term geometric growth but produces catastrophic drawdowns in practice.
I use half-Kelly as a ceiling: 24% in this example. My actual 2% sizing is well inside half-Kelly, which confirms the bet is not over-sized. If my conviction-tier sizing ever exceeded half-Kelly, that would be a warning sign to reduce.
The Kelly check is useful in the opposite direction: it flags when I might be under-sizing a genuinely exceptional opportunity. If half-Kelly is 5% and I'm sizing at 1%, I'm leaving edge on the table. This is rare — I'm usually in line with my tier system — but it's a good second opinion.
AAPL Real Example: Walking Through the Math
High conviction trade on AAPL in Q3 2025. My thesis: Apple's Services business was accelerating faster than street estimates, the stock had pulled back 8% from highs on market rotation, and the next earnings call was 6 weeks away. Multiple confirming signals.
- Portfolio size: $100,000
- Conviction tier: 3 (high conviction) → max risk $3,000
- AAPL price: $185
- My stop loss: $175 (support level, recent consolidation zone)
- Risk per share: $185 − $175 = $10.00
Shares to buy: $3,000 / $10.00 = 300 shares
300 shares × $185 = $55,500 notional exposure. That's 55.5% of portfolio in one stock — which sounds extreme, but my actual risk is only 3% because of the tight stop. The stop defines the risk, not the notional exposure.
I bought 300 shares at $185 and set a hard stop at $174.80 (slightly below $175 to avoid noise triggering). AAPL reached $204 before earnings. I sold 150 shares at $202 (locking in 50% of the position as it neared my pre-earnings trim target) and held 150 shares through the report. Post-earnings, stock gapped to $211. I closed the remaining 150 at $209.
Total trade P&L: (150 × $17) + (150 × $24) = $2,550 + $3,600 = $6,150. Return on risk: $6,150 / $3,000 = 205%.
Portfolio Delta: Never More Than 30% in One Sector
Beyond individual position sizing, I track portfolio-level exposure by sector. My rule: no more than 30% of portfolio capital deployed in any single GICS sector.
In practice, I check this weekly during my Sunday ritual. I look at my total notional exposure by sector and compare it against my portfolio size. If I have three tech positions totaling $35,000 out of $100,000 — that's 35%, above my threshold. I won't add a fourth tech position until one is reduced or closed.
I also track portfolio delta — the aggregate directional exposure across all my positions, expressed in S&P 500 equivalent terms. My target: total portfolio delta between −0.05 and +0.05 (near-neutral). This means I can have individual long bets and short bets, but the net isn't dramatically skewed in one direction.
Correlation Adjustment
The 30% sector rule is a blunt instrument. The more refined adjustment is correlation-based. Two positions in the same sector aren't necessarily correlated — a beaten-down regional bank and a high-growth fintech are both "financials" but behave very differently.
I group positions by their actual correlation to SPX over the trailing 90 days. Any position with 90-day correlation above 0.85 to the index is treated as an "index proxy." I don't want more than 40% of my portfolio in index proxies at once.
This matters because in sharp drawdowns, high-correlation names fall together regardless of their fundamentals. In March 2020 and again in early 2022, virtually every position I had with SPX correlation above 0.85 declined simultaneously. Correlation management is the one adjustment that would have meaningfully reduced my 2022 drawdown.
The One Rule I Never Break: No Sizing Up After Wins
The most dangerous moment in trading is after a winning streak. I've tracked my own behavior carefully and found that I consistently take larger risks in the 2–3 weeks following a significant win. It's pure psychology — the brain interprets recent wins as evidence of skill, reduces perceived risk, and nudges you toward larger bets.
My counter: I do not adjust position sizing based on recent results. At all. My tier system runs on thesis quality, not recent P&L. After a $15,000 winning month, I'm still taking the same 1%/2%/3% sized positions. The equity curve doesn't get a vote on what the next trade should be worth.
The only time I adjust size is when my portfolio capital changes by more than 10% — then I recalculate absolute dollar amounts based on the new capital base.
— Ruslan Averin, averin.com