A passing backtest feels like a finish line. It is not. It is permission to start paper trading — running the strategy against live prices with no real money at stake — so the market gets a chance to disagree with your model before your capital does. This guide covers how to cross that gate deliberately, and what to watch while you do.
Why paper trading exists at all
A backtest, however careful, runs on history. Paper trading runs on the present: live spreads, live liquidity, live latency between a signal and a fill. These are exactly the conditions a backtest can only approximate. The paper stage is where a strategy meets the parts of reality that no historical test fully contains.
Step by step
- Confirm the walk-forward result. Check the strategy passed across out-of-sample folds with the realistic cost model applied — not just on a single tuned window.
- Run it in paper mode. Promote the strategy so it generates live signals against real prices, places simulated orders, and logs everything, all without touching funds.
- Compare paper to backtest. Watch for paper results drifting below what the backtest promised. That gap is the market telling you where your model was optimistic.
- Hold the line on the gate. Only after a sustained, cost-clearing paper run should the paper-to-live confirmation gate even be considered. A few good paper days is not evidence.
What a healthy paper run looks like
You are looking for paper performance that tracks the backtest reasonably closely, signals that still clear their full cost under live conditions, and risk limits that behave exactly as configured. A wide gap between paper and backtest is not a reason to push harder — it is a reason to go back and ask what the test missed.
Make sure the test that earned this step was sound: revisit how to set up a walk-forward backtest and how to spot overfitting. When paper trading proves out, the Kraken bot pillar covers the paper-to-live gate, and the risk pillar covers the limits that travel with the strategy.