Confidence Scores Decoded

Confidence is a probability hint, not a promise

The confidence on a swipe card signals how strong the setup is relative to recent history. It’s not a guarantee; it’s a guide to how much you should risk, not whether you should abandon discipline.

How to translate confidence into size

• Set a base risk/unit: e.g., 0.75% of account per trade on normal days.
• Scale inside a band:
  • 50–60: 0.5× base
  • 61–75: 1.0× base
  • 76–85: 1.25× base
  • 86–90: 1.5× base
  • 90+: Rare. Keep it at 1.5× base, not “all-in.”
• Never exceed your daily/drawdown brakes: If you’re near limits, size down regardless of confidence.

What makes a confidence score rise or fall

• Confluence of signals (structure + momentum + liquidity health).
• Recent hit rate of similar setups.
• Volatility regime and spread quality.
• Time decay (TTL): stale ideas lose edge and confidence.
• Venue reliability: rejects or throttling drag confidence.
• Market microstructure: wide spreads or shallow book reduce usable edge.

How to avoid overbetting

1. Cap per-trade risk: Hard ceiling (e.g., 1.5% max) regardless of confidence.
2. Cap same-idea stacking: If two cards are correlated, treat them as one position.
3. Respect venue/asset limits: Thin books? Even high confidence gets smaller size.
4. Honor daily brakes: Confidence never overrides risk stops.
5. Adjust for volatility: If ATR doubles, halve size even on high confidence.
6. Match RR: If RR < 1.3, keep size on the lower band no matter the score.

Quick read checklist before you swipe

• Does the stop sit outside noise? High confidence with a too-tight stop is fake safety.
• Is the RR acceptable? If target/stop is ~1:1, size small even at high confidence.
• Is liquidity normal? Wide spreads reduce real edge.
• Are you correlated elsewhere? If you’re already long similar beta, downsize.
• Is TTL fresh? A stale high-confidence card is worse than a fresh medium one.
• Is funding sane? High funding eats edge; size down.

How to react to outcomes without tilting

• Hit TP at high confidence: Log it; confirm the model’s calibration. Don’t increase size beyond your band.
• Hit SL at high confidence: That’s data, not betrayal. Stay within the band; revisit structure quality.
• Hit SL at low confidence: Working as intended. Small risk protected you.
• Expire without fill: Good—no fill at bad price. Wait for the next card.
• Slip more than expected: Trade fewer cards during wide-spread windows or reduce band by 0.25×.

Calibration drill (15 minutes, once a week)

1. Take last 20 cards. Bucket by confidence decile.
2. For each bucket, note TP%, SL%, average RR.
3. If a bucket underperforms, cap it to 0.75× base next week.
4. If a bucket overperforms with tight stops, check slippage to ensure it’s real edge.
5. Adjust band rules; write them down so you don’t improvise mid-week.

Common confidence myths

• “High confidence means bigger stops are okay.” No—stop distance is structure-driven; size absorbs distance.
• “Low confidence = skip all.” Not always; low can be fine with tiny size if RR is strong.
• “Confidence replaces RR.” Never. Bad RR + high confidence is still a bad bet.
• “One big win fixes variance.” Overbetting one card usually breaks the curve.

Mini playbook: news days

• Treat all bands as half-size on major news days.
• Only take cards with RR ≥ 2.0 and stops outside the news wick.
• If spreads blow out, pause until they normalize; confidence scores assume sane microstructure.

Mini playbook: low-vol chop

• Expect more fake breaks; prioritize cards with wider structure stops.
• Keep size at 0.5–1.0× base unless volatility picks up.
• Be comfortable passing more; confidence drops faster when TTL decays in chop.

FAQs

• Can I ignore low confidence if I “feel” it? Feelings aren’t calibration. If you must trade it, trade it small.
• Why not go 3× at 90+? Edge isn’t linear at the tails; capacity and variance explode. Protect longevity.
• Does confidence equal hit rate? It’s related but also factors liquidity, vol, and recency of similar setups.
• Should I stack two high-confidence cards on same asset? Treat them as one idea; split size, don’t double total risk.
• What if confidence drops after I’m in? If structure and RR still hold, stick to plan. Avoid panic exits without signal.
• Do funding/spread changes affect confidence? They should affect your size. High cost = smaller size even if confidence is high.

The takeaway

Confidence should tune your size, not override your rules. Pick a base risk, apply a simple banded scale, and never let a high number talk you into breaking your brakes. That’s how you stay in the game long enough for any edge to matter.