PaperPumpPAPER

How to backtest a memecoin trade

By PaperPump team · Updated

Backtesting a memecoin is not the same problem as backtesting a stock. There is no decade of history, the coin may be four days old, and the liquidity that existed during the move is often gone by the time you study it. What you can do is replay the candles that actually printed and trade forward through them — provided the replay charges you honestly for the part that is easiest to fake, which is execution.

PaperPump is a simulator. Every balance is virtual and every trade is simulated against real market data. Paper results do not predict real results. Read the risk disclosure.

Replay the real candles, not a model of them

PaperPump's Replay Mode plays a coin's own history back at a speed you choose, with a bankroll belonging to that session alone — your live balance is untouched, and the balance pill turns blue so you cannot confuse the two. You pause, scrub, and trade into it the way you would live.

The candles are the ones that printed. Nothing is generated, smoothed or invented to make the setup clearer, which matters because the moments worth studying are usually the messy ones.

The assumption that makes most backtests lie

Here is the part almost nobody models: your order does not fill instantly, and how long it takes depends on how contested the coin is at that exact moment. A backtest that fills every order at the candle price is not slightly optimistic — it is optimistic precisely at the moments you are studying, because the moments worth studying are the busy ones.

A coin doing 8 to 30 trades a second is a completely different execution problem from a dead one, and the gap is seconds, not milliseconds. Replay reconstructs the real contention at the cursor from actual per-trade timestamps rather than assuming a constant, so a fill during a rush costs what a fill during a rush costs. If the rate genuinely cannot be established for a moment, it says so and falls back rather than quietly inventing a flattering number.

This is the single biggest reason a strategy can backtest well and lose live. It is worth checking whether any tool you use models it at all — most do not.

The same execution model runs on live coins too, which is what pump.fun paper trading here means: a fill on a contested coin waits, and the wait depends on the priority fee you set.

How much replay you get

There is deliberately no clock on it. A time budget that drains while you think punishes the exact thing replay is for — pausing to actually read the chart — so a session is bounded by TRADES instead: 50 on a free account, 200 on Pro, uncapped on Unlimited. You can sit on one candle for an hour and it costs nothing.

Sessions are the other limit, and they matter more: 2 on free, 10 on Pro, unlimited above that. One bankroll across every experiment means a good result in one hides a bad one in another; a session per idea keeps each answer readable. Free accounts can also only go back three months on any timeframe. See pricing and plans for the rest.

How to actually run one

Pick a coin that already ran and start the replay before the move, not at it — most of what you need to learn is in the part where the setup was still ambiguous. Decide your entry out loud before you scrub forward, then take it and let the fill cost what it costs.

Do it again on the same coin with a different entry. The comparison between two runs on one chart teaches more than ten runs across ten coins, because everything except your decision is held constant.

How to Backtest a Memecoin Trade · PaperPump