Conviction-Based Sizing: High-Conviction Bets
Conviction-Based Sizing: High-Conviction Bets
How Do You Size Positions Based on Conviction?
Conviction-based position sizing is a method that calibrates your bet size to your actual confidence level in a trade. Instead of using a fixed percentage of capital per trade—which treats all opportunities equally—conviction sizing scales your exposure up when your edge is strongest and scales down when you have lower confidence. This approach recognizes that not all trades merit equal capital allocation; a setup with 75% historical win rate and clear, repeatable patterns deserves larger risk than a borderline setup with 52% edge.
Professional traders and hedge fund managers use conviction sizing because it maximizes returns per unit of risk taken. When you can quantify why you're confident in a trade—through backtested edge, macroeconomic fundamentals, or statistical probability—you can size accordingly. The method requires honest self-assessment: you must differentiate between hope-based conviction (emotional bias) and evidence-based conviction (pattern recognition, technical structure, fundamental catalysts). Done correctly, conviction sizing compounds returns by deploying more capital when your probabilistic edge widens and preserving capital when uncertainty is high.
Quick definition: Conviction-based position sizing allocates position size proportionally to your confidence level in a trade, scaled by historical win rate, pattern strength, or probability of success. Higher conviction produces larger positions; lower conviction produces smaller positions.
Key Takeaways
- Conviction sizing ties position size directly to edge strength and confidence probability, aligning capital deployment with actual skill
- Three conviction levels (low, medium, high) provide a simple framework: 50% base size, 100% standard size, 150% max size
- Quantified conviction—measured by backtest win rate, pattern frequency, or fundamental catalyst weight—prevents emotional over-sizing
- Kelly Criterion offers a mathematical foundation: f = (bp − q) / b, where position size scales with edge probability and payoff ratio
- Real conviction sizing separates successful traders from breakeven traders because it compounds edges into measurable outperformance
- The method requires discipline: conviction must be assigned before entry, not retroactively after price moves
Understanding Conviction as a Sizing Driver
Conviction is measurable, not intuitive. A trader with genuine conviction can answer: "Why am I 80% confident in this trade, and what data supports that confidence level?" The answer might reference a pattern that works 78% of the time in backtests, a fundamental catalyst with historical precedent, or a technical setup that has triggered profitable trades 23 out of 31 times. Without this quantification, "conviction" becomes anchoring bias dressed up in trader language.
Consider a currency trader who has tested a mean-reversion bounce off the 200-day moving average. Historical data shows this pattern wins 62% of the time with a 1.3:1 payoff ratio. That 62% win rate is evidence-based conviction. A separate setup—a broad momentum trend with weaker confirmation—might show only 52% win rate across similar conditions. The first setup warrants a larger position than the second because the edge is demonstrably stronger.
Size the low-conviction trade at 0.5% risk per trade. Size the high-conviction setup at 1.0% or 1.5%. The trader with the discipline to follow this discipline compounds over time because capital is systematically deployed when odds are favorable and conserved when odds are marginal.
The Three-Tier Conviction Framework
The simplest conviction-based approach uses three tiers: low conviction, medium conviction, and high conviction. Each tier corresponds to a position-size multiplier applied to your base risk unit.
Low conviction (multiplier: 0.5x) applies to setups with marginal edge, untested patterns, or high ambient uncertainty. If your base risk per trade is 1%, a low-conviction position risks 0.5%. This tier includes trades in new markets you're learning, setups during high volatility spikes, or patterns you've seen only a handful of times. The risk is real, but you're not risking meaningful capital until you've validated the edge.
Medium conviction (multiplier: 1.0x) is your standard position size. This covers trades that meet your normal criteria: established patterns with decent track records, technical setups with clear support and resistance, or fundamental plays with moderate catalysts. If your base is 1%, medium conviction is 1%.
High conviction (multiplier: 1.5x) applies to your highest-edge setups. These are patterns with 70%+ historical win rates, setups with multiple confluences (price action + momentum + fundamental catalyst), or trades into chart levels that have held multiple times historically. High conviction positions risk 1.5% of capital.
This framework prevents the emotional trap of "revenge sizing" (adding risk after losses) and the greed trap of equal-sizing all trades. Instead, you mechanically scale to conviction assigned beforehand.
Mathematical Foundation: Kelly Criterion
The Kelly Criterion offers a mathematically optimal position size given an edge and payoff ratio:
f = (bp − q) / b
Where:
- f = fraction of capital to risk per trade
- b = payoff ratio (win size / loss size)
- p = probability of win
- q = probability of loss (1 − p)
Suppose a trader has a 60% win rate, risking $1 to make $1.20 (1.2:1 payoff). The Kelly calculation is:
f = (1.2 × 0.60 − 0.40) / 1.2
f = (0.72 − 0.40) / 1.2
f = 0.32 / 1.2
f ≈ 0.0267, or 2.67%
Kelly suggests risking 2.67% per trade. However, full Kelly is aggressive and can lead to deep drawdowns. Most traders use "fractional Kelly" (half Kelly, third Kelly) to reduce volatility. Half Kelly suggests 1.33% risk per trade—still higher than the 1% standard rule but mathematically justified by the edge.
The beauty of Kelly is it quantifies what conviction should mean: your position size is directly proportional to your true edge. If your edge shrinks (win rate drops to 55%), Kelly size shrinks. If edge widens (win rate climbs to 68%), Kelly size grows. This mathematical framework transforms conviction from emotion into calculation.
Conviction Assignment Framework
Assigning Conviction Levels Objectively
Discipline requires assigning conviction before you enter the trade. Here's a practical rubric:
Backtest frequency: If a pattern appears 3+ times per week in your backtest with 65%+ win rate, assign medium or high conviction. If it appears once per month with 55% win rate, assign low or medium conviction.
Confluence count: Count the number of independent signals confirming the trade. Price at moving average + oscillator overbought + fundamental earnings catalyst = three confluences. This warrants high conviction. Price at round number alone = low confluence, low conviction.
Recent track record: In your trading journal, how many consecutive wins or losses from this pattern? Five wins in a row suggests strong current conviction. Two wins and three losses in the last ten trades suggests medium conviction.
Chart structure clarity: Can you draw support and resistance levels with high confidence? Do they align with round-number psychological levels? Clear structure = higher conviction. Murky structure = lower conviction.
Personal emotional state: Are you chasing the setup because you're frustrated with recent losses, or did you identify it dispassionately? Frustrated traders systematically overestimate conviction. Cool-headed traders are more reliable.
Once you assign conviction, the position size follows mechanically. This removes the moment-to-moment emotional override that destroys capital.
Example: A Three-Conviction Trade Day
A currency trader uses conviction-based sizing with a 1% base risk per trade.
Trade 1: GBP/USD pullback to 1.27 support. This level has held three times in the past eight months. Recent price action is clean; volume confirming the move. Backtest shows 58% win rate at this level. Conviction: medium. Position size: 1.0% risk.
Trade 2: EUR/USD breakout above 1.10 resistance. This is a new pattern the trader is testing, based on a novel technical setup from a trading book. No backtest data yet. Conviction: low. Position size: 0.5% risk.
Trade 3: USD/JPY short into an FOMC rate decision. The trader has tracked this setup 12 times over two years; it has a 72% win rate. Multiple technical confluences align (price at moving average, bearish divergence on RSI, and fundamental catalyst of rate pause). Conviction: high. Position size: 1.5% risk.
The trader's total risk for the day is 0.5% + 1.0% + 1.5% = 3.0% of capital. If all three trigger in the same direction on the same day, drawdown exposure is bounded. But capital is deployed heavier into the highest-edge setup (trade 3), which is where historical edge is strongest.
Real-World Examples
Renaissance Technologies: The legendary quant fund sizes positions by Sharpe ratio and correlation, a data-driven conviction proxy. Positions with higher Sharpe ratios (edge strength) receive larger allocations. This is conviction sizing in algorithmic form.
Paul Tudor Jones: The macro trader has stated that position sizing reflects confidence. He scales into major macro bets when multiple time-frame analysis aligns (daily, weekly, monthly charts all confirming the same directional thesis). When only one time frame confirms, he sizes much smaller. This is conviction-based scaling in practice.
Retail daytrader example: A daytrader tests a momentum setup over three months (63 trades). Win rate is 61% with 1.2:1 payoff. In month four, the trader increases position size from 0.5% to 1.0% per trade, reflecting higher conviction after validating the edge. Over the next year, this single adjustment—matching size to proven conviction—increases account growth by 18% compound because capital is deployed heavier when edge is demonstrably stronger.
Common Mistakes in Conviction Sizing
Mistake 1: Emotional conviction vs. statistical conviction. A trader feels confident about a breakout trade but has no backtest data supporting the setup. Emotional conviction often precedes drawdown. Assign low or medium conviction until you have evidence.
Mistake 2: Assigning conviction retroactively. After a winning trade, a trader mentally boosts the conviction level assigned before entry, creating selection bias. Discipline requires conviction assigned before entry; after entry, honor the original assignment regardless of early price movement.
Mistake 3: Stacking convictions into overleveraged positions. A trader believes 72% in trade A and 68% in trade B, so sizes both at 1.5% high conviction for a combined 3% daily risk. The two trades are correlated (both long USD), creating hidden leverage. Conviction sizing is per-trade; overall portfolio risk must still be bounded.
Mistake 4: Ignoring conviction decay. A pattern worked well for six months (high conviction assigned). Then it fails 12 times in a row. The trader keeps the original high-conviction assignment. Conviction must decay with recent empirical failure. Update conviction downward as pattern success rate declines in real time.
Mistake 5: Confusing conviction with aggressive sizing. Conviction sizing is about matching size to edge, not about pushing size to the maximum. A trader who sizes 2% on a 55% win rate is not showing conviction; they're gambling. True conviction sizing sizes 0.5% on low edge, 1.0% on medium edge, 1.5% on high edge—and stops there.
Frequently Asked Questions
Q: How do I know if my assigned conviction is accurate? A: Track it in your trading journal. Write down the conviction level (low/medium/high) before entering, then log the result (win/loss). Over 50+ trades, calculate the actual win rate for each conviction tier. If low-conviction trades really do have worse win rates and high-conviction trades have better win rates, your conviction assignment is calibrated. If all tiers have similar win rates, you're assigning conviction emotionally, not objectively.
Q: Should I adjust conviction mid-trade if price moves against me? A: No. The conviction level assigned at entry is locked in. If price moves 2R against you and the trade is still valid per your original thesis, stick with the original size. If price invalidates the original thesis, exit per your stop loss. Do not retroactively downgrade conviction and reduce size mid-trade; that's emotional and destroys discipline.
Q: Can I use conviction sizing with fixed-dollar sizing? A: Yes. Instead of scaling risk percentage (0.5%, 1.0%, 1.5%), scale the dollar amount. If your base position is $10,000, scale low conviction to $5,000, medium to $10,000, and high to $15,000. The principle is identical: match absolute capital deployment to edge confidence.
Q: How many conviction tiers should I use? A: Three tiers (low, medium, high) are optimal for most traders. More tiers add decision fatigue without improving returns. Some advanced traders use five tiers, but this requires more precise backtesting and can introduce analysis paralysis at trade entry.
Q: What if I have zero conviction in a setup but my system generates a signal? A: Do not trade it. A system signal and genuine conviction are different things. If you have low conviction, size it at 0.5%, and then be okay losing that trade. If you have near-zero conviction, skip it. Over a career, skipping low-conviction setups (even if they occasionally win) compounds into better risk-adjusted returns.
Q: How does conviction sizing compare to Kelly Criterion sizing? A: Conviction sizing is qualitative and discretionary; Kelly Criterion is quantitative and mathematical. Both aim to match size to edge. A trader using conviction sizing should, over time, converge toward Kelly Criterion size once enough data is collected. The tiers (0.5x, 1.0x, 1.5x) can be calibrated to approximated Kelly sizes for your typical edge profile.
Related Concepts
- ./20-the-1-percent-rule.md
- ./01-fixed-dollar-sizing.md
- ./22-sizing-for-trend-following.md
- ../chapter-02-the-risk-of-ruin-equation/01-what-ruin-means.md
Summary
Conviction-based position sizing is a professional approach that scales capital deployment to edge strength and confidence probability. By assigning conviction levels (low, medium, high) to each trade before entry and sizing accordingly (0.5x, 1.0x, 1.5x), traders align position size with actual edge instead of deploying equal capital across unequal opportunities. The approach is grounded in Kelly Criterion mathematics but implemented through discretionary assignment, making it practical for traders who combine quantitative backtesting with qualitative pattern recognition. The discipline required—assigning conviction beforehand, not retroactively, and resisting emotional override—separates long-term profitable traders from breakeven traders. Over years, traders who scale positions to proven conviction systematically outperform traders who risk equal amounts on every setup, because capital is automatically deployed heavier when odds are favorable and conserved when odds are marginal.