Four strategies. The code. And the documentation to run them.

What this package contains.

---

You see the potential before you open a file

The first thing shown for every strategy is its performance. Profit, maximum drawdown, recovery factor, trade count, win rate and the longest losing streak, on one page. Nothing is hidden: two of the four win less than half of their trades, and one of them survived fourteen losses in a row.

Underneath sits the part that makes the package usable. Every strategy documents how it enters, how it exits and what it does that a quick read of the code would miss, together with the parameter set we shipped. Paste the code, paste the parameters, and your first backtest reproduces our numbers. From there the documentation walks you through incubation on live data and on to a real account.

---

What they did on Gold

Same instrument, four different ways of trading it. Two catch reversals, two follow trends, which is why they are worth holding together.

---

What you can do with it

The .py and .mq5 files are not black box containers. They are readable, parameterised code that you own the moment you download it.
You can edit them as you like and apply them wherever you like.

• Reproduce the numbers today. Paste the .py file and its .json parameters into the builder and run the backtest. For the trend reversal you can instead drop the .mq5 into MetaEditor and use the Strategy Tester.
• Re-optimize it for your symbol. Every input is exposed. Point the optimizer at your instrument, your timeframe and your broker's spread instead of ours.
• Swap the entry signal. The entry logic sits in one place in every file. Replace it with your own idea and keep the position management, the parameter set and the risk handling that already work.
• Read how a model exposes parameters. The most useful part is often not the strategy — it is seeing which knobs the AI builder chose to leave open, and why that makes a strategy optimizable at all.
• Use it as a starting template. Let your AI read it and prompt it for the next idea instead of asking a model to build the same scaffolding from scratch every time.