EA backtesting and strategy optimisation
Before you put money behind a strategy, find out how it behaves on real tick data with realistic spreads and commission, whether its settings survive data they were not tuned on, and how deep the drawdowns could get. You get the raw reports and a written verdict.
Mutual NDA before you share your strategy · reply within one working day
| Equity and drawdown | The full curve, deepest drawdown and how long it lasted |
| Trade list | Every trade, so nothing is averaged away |
| Cost stress | Results with spread and commission increased by half |
| Walk-forward | Optimise on one window, test on the next, repeated |
| Monte Carlo | Trade order shuffled 1,000 times: realistic worst drawdowns |
| Stability map | Do neighbouring parameter values also work? |
| Verdict | Written, in plain English, with recommended risk |
- Starting price
- $120
- Typical delivery
- 3–5 days
- Revisions
- 2 revision rounds
- Handover
- Reports + written verdict
How I test a strategy.
- 01
Data you can trust
Real tick data with variable spreads where the platform allows it, and a data-quality check before any result is read.
- 02
Real costs
Your broker’s commission and typical spread, plus slippage on market orders. Then the same test again with costs increased.
- 03
In-sample and out-of-sample
Settings are chosen on one period and judged on another. Results that only exist in-sample are reported as such.
- 04
Walk-forward
Rolling windows of optimise-then-test, which shows whether the strategy adapts or simply fits.
- 05
Monte Carlo
Thousands of reshuffled trade sequences give a realistic range for drawdowns, which is what decides your position size.
- 06
Breakdown by regime
Results by year, session and volatility level, so you know when the strategy struggles.
Signs a strategy is fitted to the past.
If your current backtest shows several of these, test it properly before risking money.
- A smooth equity curve with very few losing months
- Profit that disappears when spread is increased slightly
- Settings with unusual exact values, such as a 37-period average and a 2.73 multiplier
- Great results on one symbol and losses on similar ones
- Fewer than about 100 trades behind the headline numbers
- Performance concentrated in one short period
What you send, and what you get back.
- The EA (.ex4/.ex5 is enough) or the strategy rules
- The broker and account type you trade, or plan to
- The settings you use now, and any you are considering
- Your risk limits: per trade, per day, overall
- What you want to know: “is it robust?”, “which settings?”, “what risk per trade?”
- The unedited Strategy Tester reports for every run that matters
- Optimisation and walk-forward results as a spreadsheet
- A written verdict in plain English: robust, fragile, or not worth trading
- Recommended settings and risk per trade, with the reasoning
- A list of the conditions in which the strategy struggles
Common questions
Can you test an EA I bought, without the source code?
Yes. Backtesting and optimisation only need the compiled EA (.ex4/.ex5) and its inputs. Changing its logic does need the source.
Which data do you use?
On MT5, real ticks from the broker’s history where they are available. On MT4, imported third-party tick data, because MT4’s own modelling is not reliable enough to judge a strategy on.
Can you make my strategy profitable?
No one can promise that. Optimisation finds settings that are robust, or shows that the strategy does not hold up. Both answers are worth having before you trade it.
How long does a full test take?
Usually 3 to 5 working days for one strategy on one symbol, including walk-forward and Monte Carlo runs. Multi-symbol portfolios or very long tick histories take longer, and I confirm the timeline when I quote.
Do you test Python and TradingView strategies?
Yes. Python strategies are tested in backtrader, vectorbt or a custom engine. Pine strategies are tested in TradingView and, where it matters, rebuilt in MT5 or Python for a more realistic test.
Tell me the strategy. I’ll tell you what it takes.
Fifteen minutes, no pitch. Mutual NDA before you describe anything you care about. You get an honest view of feasibility, timeline and price, even if the answer is “don’t build this”.