Paper, Backtest, and Live Differences
Published Jun 22, 2026 · Updated Oct 16, 2026
Same parameters, but backtest, paper, and live results don’t line up? The gap can come from fill assumptions or from a fault. Compare item by item using the tables below before drawing conclusions. For what each stage proves, see Backtesting vs Paper Trading vs Live.
Why do backtest and paper differ?
| Source | What differs | How to check |
|---|---|---|
| Market period | Backtests replay history; paper runs on current prices | Compare only similar market phases |
| Intra-candle prices | Backtests simulate the price path inside each candle from the open, high, low, and close of the “Simulated backtest candle interval”; they are not real finer-grained exchange prices | Use a finer interval and compare again |
| Fill costs | Backtest fees and slippage are recorded in the version snapshot; paper defaults to 0.05% taker and 0.02% maker fees and fills at the last price plus a fixed 0.01% slippage | Check the backtest snapshot under “Saved backtest parameters” |
| Funding | Neither backtests nor paper settle funding | When positions span funding times, live P&L includes funding that the others do not |
| Parameter version | Paper may not be running the version you backtested | Compare the strategy settings field by field |
| Initial funds | The backtest’s “Initial equity (USDT)” does not become the paper balance | Compare the paper account’s initial funds |
Why do paper and live differ?
| Source | What differs |
|---|---|
| Slippage and depth | Paper fills at the last price plus fixed slippage without touching the order book; live fills against the real book, and the gap grows with size |
| Fees | Live uses your actual exchange fee rate, which depends on VIP tier and rebates |
| Funding | Live positions pay or receive real funding; paper does not settle it |
| Exchange rules | Live orders can be rejected for minimum size, price limits, position caps, or position-mode conflicts; paper simulates Binance or Hyperliquid public rules and precision |
When comparing, use the same capital and position limits, and keep simulated costs separate from what the exchange actually charged. Paper results close to expectations still do not guarantee the same live performance.
When is it a real problem?
If you see any of these, send both strategy names, the timestamps, and the “Strategy Events” entries to support@alphafox.app:
- At the same moment, paper and live act in opposite directions on the same signal;
- An event shows a fill, but the exchange account shows no matching position change;
- The parameters match, but the two sides show different config versions.
If a live strategy stops or errors unexpectedly, start with Strategy Not Starting or Stopped.