How to size your position after receiving a ROBUST verdict from Kiploks
Turn a ROBUST validation verdict into a live sizing plan: start small, scale with evidence, cap by drawdown budget, and avoid jumping from backtest notional straight to full capital.
A ROBUST verdict means your evidence stack is comparatively strong. It does not mean you should deploy full capital on day one.
Sizing is where many good strategies die from operational risk: slippage, outages, partial fills, and psychological overrides.
Rule 1: separate research notional from live notional
Your backtest used a notional that may ignore capacity, leverage constraints, and exchange rules.
Pick a starting live fraction such as:
- 10-25% of your target long-term size for the first 2-4 weeks
Then scale only when live behavior matches expected ranges for:
- fill quality
- fee drag
- drawdown path shape
Rule 2: size from drawdown budget, not from greed
Define the maximum dollar drawdown you can tolerate for this strategy. Convert that to maximum exposure using a conservative worst-case drawdown estimate from OOS windows, not the best window.
If the math says your desired size implies an unacceptable loss, the size is wrong even if the edge looks strong.
Rule 3: scale in steps with prewritten rules
Example ladder:
- week 1-2: 10% size, tight kill-switch
- week 3-4: 25% if execution metrics are stable
- month 2: 50% if OOS-like live behavior holds
- month 3+: 100% only if nothing material changed in logic or market structure
Write this ladder before you trade. After losses, humans rewrite ladders.
Rule 4: freeze logic during ramp-up
No parameter tweaks during ramp-up unless you explicitly reset the validation clock.
Keep paper trading in parallel for one cycle
A verdict is historical evidence. Live is a different measurement instrument.
Run a short overlap where paper and live should match within tolerances on fills and fees. If they diverge, fix execution assumptions before scaling.
Kelly-style sizing is optional, not automatic
Aggressive growth sizing can violate your drawdown budget even when edge exists.
If you use Kelly thinking, cap it hard and treat it as an upper bound, not a target (Kelly after validation).
Document what would invalidate the verdict
Write three falsifiable checks: fee drift, trade count drift, drawdown shape drift.
If any trigger hits, you return to review size and possibly pause, even if the old verdict still "feels" valid.