BackRampage Trader

Most of our strategy ideas fail our tests

New ideas, especially from our computer searches, have to get through these tests. Our listed strategies started in our own trading. Each one's practice record is on its page.

6 stops from idea to running

Each stop can say no. Most ideas stop at the first two.

  1. 1

    Idea

    From our own trading, strategies we've studied, and computer searches.

  2. 2

    Backtest on 2026

    Must make money in both halves of 2026, even with double slippage.

  3. 3

    Beat random trades

    Must beat random entries that use the same stop and target.

  4. 4

    Luck check

    The more ideas we test, the higher the bar gets.

  5. 5

    Paper forward test

    Live prices, pretend money. The paper test is there to see whether the fills look like the backtest.

  6. 6

    Watched while it runs

    While a strategy is running, the plan is to compare it with its tests. If it drifts, it gets fixed or switched off.

Where ideas come from

  • Our own trading. Setups from our own trading.
  • Strategies we've studied. We study other traders' ideas, then rebuild them our way. Never a copy.
  • Computer searches. Our computers try millions of rule combinations on years of minute-by-minute prices.
  • Our strategies' own trades. The trades we review are logged. Patterns in the wins and losses become new ideas.

The backtest: 4 tests, all must pass

Markets change, so we judge on this year's market: 2026. Older years are a side note.

  • Both halves of 2026

    Split the year in two. It must make money in both halves, not one lucky stretch.

    Jan–May May–now = pass Jan–May May–now = out
  • Double slippage

    Real fills are a little worse than the chart. We charge every test trade twice the normal slippage. If it stops making money, it's out.

  • Beat random trades

    We run the same stop and target at random times, hundreds of times. The real rules must beat at least 95 out of 100 random runs. What's this?

  • The luck check

    Test a million ideas and thousands look great by pure luck. So the bar rises with the number of ideas tested. What's this?

One search, start to finish

In November 2026 our computers searched one family of ideas across 8 markets. This is what came out.

Backtest: squeeze search, November 2026, 8 markets, 2026 prices

  1. Made money in both halves of 2026, even at double slippage

    18,204,924
  2. Finished the random-entry test

    9,241,636
  3. Beat random entries, before the luck check

    3,214,519
  4. Still passed after the luck check

    0
  5. Became a new strategy

    0

Bars use a log scale so the small steps stay visible. Source: our lab's own report for this search.

No idea from this search passed

Millions of ideas looked good at first. None got past the luck check, so nothing from this search goes near an account. A search that finds nothing is the system working.

What wasn't scored

8,963,288 ideas couldn't finish the random test, so they weren't scored at all. 948,018 of them were the busiest. Unscored never counts as a pass.

The forward test: live prices, pretend money

  1. Paper first

    A new strategy is meant to run on a paper account for days or weeks first.

  2. Fill by fill

    The check is whether paper fills look like what the backtest said should happen.

  3. Real money last

    Real money last: only after the paper test, at a size set from the account's loss limit. A real-money run, if one is started, has to be checked against the same model. This site shows no real-money results.

What testing can't do: a backtest is a test on the past, and markets change. A strategy that passed every test can still lose, and some do. That's why each strategy's page shows our practice account's losses next to its wins.

See the strategies

About 24 minutes to running on paper. You pay Atomik at step 3. Live prices, pretend money.

Start on paper

Atomik Pro: $129 a month, paid to Atomik. 7-day money-back on your first payment.

Futures trading can lose money.