Enter your rules. We backtest them against up to 20 years of real market data with full trading costs, then run seven further tests designed to find out whether the result would happen again.
Most backtests flatter you. They ignore the spread, they're tuned on the same history they're measured against, and they can't tell a real edge from a lucky run. These tests exist to catch exactly that — which is why a report can tell you "do not automate this" about a strategy showing a healthy return.
The free report is a real backtest. The full report tells you whether to trust it.
If the tests find your strategy should not be automated, the report says so plainly and explains exactly why, naming the number behind every finding. That is the outcome for a good share of strategies, and it is worth far more than a flattering chart — it is the cheapest possible way to learn it.
Each one targets a specific way a good-looking backtest turns out to be worthless.
| Test | The question it answers |
|---|---|
| Out-of-sample | Rules can be bent to fit the past. Does it work on history it never saw? |
| Parameter sensitivity | If it only works at exactly 50 and 200, those numbers were chosen in hindsight. |
| Statistical confidence | Was this one lucky sample? What is the actual spread of plausible outcomes? |
| Persistence | Did it work throughout, or make everything in one good year? |
| Cost tolerance | Spreads widen at news. Does the edge survive that? |
| Random benchmark | Do your entry signals beat random timing with the same exits? |
| Drawdown reality | The worst drawdown that happened is rarely the worst that could have. What should you plan for? |
No. Everything is dropdowns and numbers. If you can describe your rules to another trader, you can enter them here.
Real hourly candles from Dukascopy’s public institutional feed, going back to 2005 for the majors. Every bar is validated before use, and any data quality problem is reported in your report rather than hidden.
Twenty instruments: all the majors, the main crosses, and gold. Hourly and above — H1, H4, daily and weekly.
The source data is hourly and we will not fabricate finer bars from it — that would invent detail that is not there and produce confident nonsense.
Yes — see below. But test it first. Automating a strategy that fails these checks means automating losses.
No. It is a technical analysis of rules you supply. We do not advise on what to trade, and we are not licensed advisors. Every trading decision remains yours.
Once a strategy passes the tests, the next step is automating it properly.
Custom builds from $1,500, plus $75/month for hosting and monitoring. Email mike@muggle.world with your report reference and I will come back with a scope.
Trading foreign exchange on margin carries a high level of risk and can result in the loss of more than your initial deposit. Leverage works against you as easily as for you. The majority of retail traders lose money.
Backtested and hypothetical results have inherent limitations. They are prepared with hindsight and cannot account for every real-world condition. Past performance is not indicative of future results.
Automated software can fail — broker outages, platform errors, data feed errors, network or server failure, unexpected market conditions, or defects in the software itself. Failures may occur while positions are open. Risk controls reduce but cannot eliminate this; stop losses may not fill at the requested price during gaps or fast markets.
This software is a tool that executes instructions you define. It is not investment advice. All trading decisions and all outcomes are yours. Only trade capital you can afford to lose entirely.
Every field is a choice from a list, so there is nothing to write and nothing to misread. Takes about two minutes.
Conditions are checked when a bar closes, and the trade fills at the next bar's open as a market order. That is deliberate: it makes it impossible for the test to use a price that wasn't available yet, which is the most common way backtests flatter a strategy.
Limit and stop entry orders aren't modelled — every entry is treated as a market order with the spread and slippage charged against it.
Saved in your browser as you type — close the tab and come back if you like.
FX-0000
Your spec is with me. I'll read it properly and reply within one business day, usually much sooner — either with questions, or with a Phase 1 start date.
I read the spec and flag anything ambiguous. We settle those points over email or a short call. Then I confirm the Phase 1 scope and send a single invoice for $600 — nothing is charged before you've agreed the scope.
Quote reference FX-0000 · mike@muggle.world