Going Live: The Minimum-Viable-Risk Pilot
A strategy that has survived research and forward testing still faces a new test: real capital. The goal of a live pilot is not to maximize profit. It is to discover whether your rules, fills, costs, and behavior remain intact when the outcome actually matters.
1. Live trading changes the experiment
Real money introduces pressures that a spreadsheet cannot model well: hesitation, fear of losing an open profit, the urge to move a stop, and the temptation to skip the next setup after several losses. A small live pilot isolates those problems while keeping the financial consequence deliberately limited.
The first objective of a live pilot is execution evidence, not income. You are asking whether the tested process can be reproduced under real conditions.
2. Define one unit of risk
Traders often express trade risk as R. If the amount you are prepared to lose when the planned stop is reached is $20, then 1R = $20 for that trade. A +2R outcome is +$40; a −1R outcome is −$20. This makes trades with different stock prices and stop distances comparable.
A setup enters XYZ at $50 with a planned stop at $48. Risk per share is $2. If a hypothetical pilot caps planned trade risk at $20, the formula produces 10 shares. If the stop is instead $49, the $1 risk per share would produce 20 shares. The risk budget—not the desire to own a round number of shares—drives the calculation.
Actual losses can exceed planned risk because stocks can gap through stops, orders can slip, and trading costs exist. Position sizing controls intended exposure; it does not guarantee the final loss.
3. Make the pilot meaningfully smaller
A pilot should be small enough that a normal losing streak does not pressure you to abandon or alter the rules. “Small” is relative to the account and the person, so there is no universal percentage. What matters is that the size is intentionally below the level you would use after successful validation.
Predefine pilot boundaries
- One frozen strategy version.
- A fixed risk unit and maximum simultaneous exposure.
- A maximum number of open positions.
- A review point based on a trade count or time window.
- Specific conditions that pause the pilot for investigation.
4. Add portfolio-level risk controls
Five positions do not necessarily represent five independent bets. Owning several semiconductor stocks, for example, may create one large industry exposure. If they react to the same news, losses can arrive together.
Imagine four trades each carrying 1R of planned risk: two chipmakers, a semiconductor-equipment company, and a semiconductor ETF. Treating the positions as four unrelated risks understates the possibility that one industry shock affects all four. A pilot plan can therefore cap total open risk and clustered exposure, not merely risk per trade.
5. Use circuit breakers for investigation—not emotion
A circuit breaker is a predetermined condition that stops new entries temporarily. Examples include reaching a specified pilot drawdown, discovering that live fills differ materially from assumptions, or repeatedly violating the written rules. The pause is for diagnosis; it should not automatically trigger a strategy redesign.
This distinction matters. A statistically ordinary losing streak is not necessarily evidence of failure, while a broken order assumption can invalidate the test even if P&L happens to be positive.
6. Grade execution separately from outcome
After every live trade, record the strategy result in R and an execution grade. A losing trade executed exactly as specified can earn an A. A profitable trade created by chasing an unplanned entry can earn an F. Otherwise, lucky mistakes can teach the wrong lesson.
Track rule adherence, planned versus actual entry, slippage, planned versus realized risk, exit adherence, and any emotional intervention. At the batch review, compare these with the assumptions used during testing.
7. Scaling should require evidence
Increasing size simply because the first few trades won is outcome chasing. A stronger rule requires a completed pilot sample, acceptable adherence, realistic execution costs, and results that remain broadly compatible with the strategy's expected behavior. Scaling can then occur in predetermined steps rather than one large jump.
Likewise, reducing or pausing size should follow predefined risk controls rather than the emotional intensity of the most recent trade.
Build a hypothetical live-pilot card
Without using your real account balance, invent a sample strategy and write:
- A hypothetical 1R dollar amount.
- An entry price and stop price; calculate risk per share and position size.
- A cap on total open R across all positions.
- One correlation rule for closely related stocks.
- One objective circuit breaker and the evidence you would inspect during the pause.
- Three conditions that must be satisfied before the hypothetical strategy can move to the next size tier.
Check your understanding
- What is the main purpose of a small live pilot?
- A stock entry is $80 and the planned stop is $78. If hypothetical planned risk is $30, what is the formula-based position size?
- Why can several positions with identical individual risk still create excessive portfolio risk?
- Should a profitable trade that violated the strategy rules automatically receive a good execution grade?
- Why is “increase size after three winners” a weak scaling rule?
- To test whether the researched process, execution assumptions, and trader behavior hold up with real capital while financial exposure remains deliberately limited.
- Risk per share is $2; $30 ÷ $2 = 15 shares.
- The positions may be correlated and respond to the same market, sector, or event risk.
- No. Outcome and process quality should be graded separately.
- Three winners are a tiny, noisy sample and do not establish that execution or the underlying edge is robust; scaling criteria should be predetermined and evidence-based.
This lesson is educational and does not recommend any security, strategy, position size, risk percentage, or transaction for your circumstances. Live trading can lose money, stops may not execute at their intended prices, and historical or simulated results do not guarantee future performance.