The Lab tests a strategy on its devs' past launches before it trades: what it would have bought, where it would have sold, and what it would have made. It's the cheapest way to try a new buy size, a filter or a set of sell rules: no SOL, no waiting for the next launch, an answer in seconds.
Run a test
- Pick the strategy
Strategy is General or one of your flows. The test uses that strategy's saved settings and the past launches of its own dev wallets.
- Say where your buy lands
Our buy lands is First, As usual or Late. It decides how many other buys come before yours in each replay; see below.
- Choose how many launches
Launches is 10, 20, 30 or 50, counting back from the most recent. 20 is the default.
- Compare, if it's a flow
For a flow you can tick Also run General's settings on the flow's launches, to compare the two on exactly the same coins.
- Test
Press Test … on past launches, with the strategy's name in the button.
How the replay works
The Lab replays each launch's first 30 minutes trade by trade. It buys with the strategy's amount, filters and entry point, then sells with the same rules the engine runs live, at the prices that were really there. A coin still held after 30 minutes, or when it migrates, is valued at that price.
- Only pump.fun curve coins are replayed. Mayhem mode coins and coins paired with another token are skipped.
- A launch's trades are collected 31 minutes after it. If some aren't in yet, a notice says how many are still loading; run again later.
- It's an estimate. Other traders replay exactly as they actually traded, so your extra buys and sells don't change what they would have done.
Where your buy lands
On a launch, every buy ahead of yours pushes the price up before you get in. That's why the entry point changes results more than almost any other setting.
Right after the dev's own buys: the best place a snipe can get.
As many buys ahead as your recent snipes usually had. Without data yet, 3.
Five buys later than usual.
Your real numbers come from the Where … snipes land card on the Snipe page. A strategy that only works with First needs a faster speed setting, or a different plan; see Snipe.
Reading the results
- PnLSOL · % on SOL bought
- What the strategy would have made or lost in total.
- BoughtN / launches
- How many launches it would have bought, and how many its filters would have skipped.
- Wonwins / bought
- How many of those buys ended in profit, as a share.
- Peak, usually×
- The typical peak of the coins it bought, with the best and worst single result in SOL.
Under the numbers, a comparison line shows the other settings on the same launches, and an assumptions line spells out the SOL per launch, the number of wallets and the buys ahead that the test used. Then a table, one row per coin: when it launched, the market cap at your entry, its peak, what sold it, and the PnL.
The Sold column tells the story of each coin with chips such as TP #1 ×2.01 · 50% · 12s (which rule, what share, and when), Stop-loss, Trailing, Time limit, Dev sold or Whale. rest held means part of the bag was still there at the end; Held 30 min and Held to migration mean no rule fired at all. A skipped coin says why.
Using it well
- Change one thing at a time and run again, so you know what made the difference.
- Run Late as well as As usual. A strategy that only survives the best entry is fragile.
- Look at the per coin table, not only the total. One huge winner can hide a rule that loses on everything else.
- Before a new flow trades, its page offers Test in the Lab →. Use it, then practise in Dry run; see Flows.