Apollo
Plain English in. A strategy you can trust out.

Trading strategies withno code.

Describe a strategy in plain English. Apollo builds it, backtests it without lookahead bias, and tells you — honestly — whether it holds up.

You
Apollo · ValidateBuild
compiled
MeanReversionsymbolAAPLlookback20entry_z-2.0
Walk-forward verdictPass
Retention60%
Sharpe (oos)1.30
AAPL mean-reversionPASS· retains 62%NQ momentum · weeklyREJECTED· collapses out of sampleSPY 50/200 crossoverPASS· Sharpe 1.18 oosGold ATR breakoutREJECTED· fitted to 2021TSLA reversion · 5% stopPASS· retains 55%EURUSD carryREJECTED· no edge out of sampleBTC trend · dailyPASS· Sharpe 0.94 oosPairs: KO / PEPREJECTED· overfit to the spreadAAPL mean-reversionPASS· retains 62%NQ momentum · weeklyREJECTED· collapses out of sampleSPY 50/200 crossoverPASS· Sharpe 1.18 oosGold ATR breakoutREJECTED· fitted to 2021TSLA reversion · 5% stopPASS· retains 55%EURUSD carryREJECTED· no edge out of sampleBTC trend · dailyPASS· Sharpe 0.94 oosPairs: KO / PEPREJECTED· overfit to the spread
01How it works

Three steps from a sentence to a strategy you can trust.

01

Describe it

Write what you want in plain English. Apollo compiles it into a real, parameterized strategy — no code required.

buy 2σ below the 20-day average
02

We build & backtest it

It runs on real data through a simulator where future prices physically don't exist. Lookahead bias is impossible, not just discouraged.

03

We try to break it

Apollo tunes on past windows, then judges on data it never saw. An edge that only worked in hindsight gets caught here.

02Inside the builder

Your whole workflow, in one place.

Every strategy you describe becomes a tracked run. Watch Apollo build and validate it in real time, then keep a history of what held up — and what didn’t.

apollo / builder
MeanReversion · AAPL
lookback 20 · entry_z -2.0 · exit_z 0.0
Pass
Equity · in-sample vs held-out+18.4%
Train Sharpe2.16
Held-out Sharpe1.30
Retention60%
OOS trades13
03Why you can trust the result

Most backtests lie. Apollo tells you when yours does.

Every strategy is judged on data its tuning never touched. If the edge was just fitted to the past, the verdict says so — in plain language.

Walk-forward · 4 windowsPass
Train Sharpe2.16
Held-out Sharpe1.30
Retention60%

verdict: held-out Sharpe 1.30 retains 60% of train Sharpe 2.16 across 4 walk-forward windows (13 out-of-sample trades)

Walk-forward · 4 windowsRejected
Train Sharpe1.74
Held-out Sharpe-1.20
Retention-69%

verdict: performance collapses out of sample — held-out Sharpe -1.20 retains -69% of train Sharpe 1.74; the edge looks fitted to the training windows

Across recent runs, roughly 1 in 3 strategies that looked great in-sample were rejected out of sample.

04What you can build

If you can say it, Apollo can test it.

›Mean-reversion on AAPL: buy 2σ below the 20-day average, sell at the mean.

Passwalk-forward

›Moving-average crossover on SPY — long when the 50-day crosses above the 200-day.

Passwalk-forward

›Momentum on the Nasdaq 100: weekly, hold the 10 strongest of the last 3 months.

Rejectedwalk-forward

›Mean-reversion on TSLA, max 100 shares, 5% stop loss.

Passwalk-forward

›Pairs trade: KO vs PEP when the spread diverges 2σ.

Rejectedwalk-forward

›Bollinger breakout on BTC, daily, 20-period band.

Passwalk-forward

Describe your first strategy.

A sentence is all it takes. Apollo handles the rest — and won’t let a fluke through.