THE ALKEMIX 2025 - MT5 EDITION

This is not the kind of trading I ever do, but like any trading system, if you can master it, it will pay.
Name of the game is to 'always be on the right side' using Alkemix trend, and then use the golden ratio to pick the next lot size, until we have decent move, starting the sequence at the right time might be the edge. Then either lock in profits by closing all (so far the best way when fully automated) or let runners go (next version will have runners be able to use 55 ema as a stop, or switch to larger timeframe automatically and use the stepma as the stoploss). The setting are there, just gotta find appropriate risk parameters for each session and timeframe.

This was just meant to make the process of buying and selling Alkemix trend signals automatic so I don't have to be looking at the screen. Todays first full 24h test netted 2% on demo, requiring me to intervene to take profits and de-risk 2 times on the 3-4-5min charts. I switch time frames (3m ->5m)to give the step a little more or less room to breathe based on my own bias.

The next step is to clean up the code and work out some of the take profit exits. Eventually for fun ill try to build a python backend with ml.

Grok:
What it does
This MT5 EA is a visual trading panel combined with a fully automated hedging system based on a signed net Fibonacci ladder.
It maintains one clean net exposure (e.g. +0.21 or -0.13) that advances or retreats along the fib sequence (0.01 → 0.02 → 0.03 → 0.05 → 0.08 → 0.13 → 0.21 → 0.34 …).

BUY signal → moves net one level higher (positive direction)
SELL signal → moves net one level lower (negative direction)
Lot size opened = |target net – current net| (big trades when crossing zero are intentional)

Main features:

Clean graphical panel with manual BUY/SELL, STEP, BANK, CLOSE ALL, micro trades, and lot override
Full automation via Alkemix Trend signals (can run on seconds chart while trading the real symbol)
Smart deleverage mode (signals close positions instead of opening new ones)
Multiple profit-taking tools: auto step/bank on P&L, BB band exits, per-position takes, trailing, swap drain, hold-days, equity DD protection
Depth-aware skin (scaled runner left after banking) + x2 step latch at deeper levels
Extensive filters (ICT kill zones, spread, session risk caps, etc.)
Commission-aware close targets and robust risk management

In short: A structured fib-ladder hedging engine with strong manual + automated capabilities.
Strengths

Precise and mathematical net management
Intelligent handling of hedge books (closes most profitable first, opens shortfall when needed)
Excellent visual panel with real-time data
Rich automation and protection suite
Works well with external signals and different instruments

Weaknesses

Can open large lot sizes during zero-crossing or at depth (requires good risk control)
Complex system with many inputs and modes — steep learning curve
Equity can diverge from balance in deep hedges (swap & open P&L)
Depends heavily on Alkemix Trend for full auto mode

What needs improvement

Better visual cues for skin target and pending actions
Stronger safeguards against oversized trades
Easier debugging / state logging
Simpler presets for different symbols

Why it’s different from other similar systems
Most fib or hedging EAs are basic grids, martingale, or simple retracement bots that add legs independently.
This EA is fundamentally different: it treats the entire hedge book as one single signed position on a Fibonacci number line. Every action (signal or step) moves the net exposure exactly one fib level with correct lot sizing — including intentional large crosses through zero. It combines precise ladder logic, smart deleveraging, depth-scaled banking, and a full automation/protection layer while keeping full manual panel control.

Is the source code for similar HA stepma available? I read HA is intensive per tick, maybe we can make it calculate every second or 2, the ea only reads the signal on bar close.

On extreme volatility days, I've manually traded this on 23-180 second charts with success... and even in chop if the waves are decent on smaller timeframes with quick profit taking it works quite well.
Overall its safe to say the less whip saw signals the better so the 1hr works really well too.

Would be interesting to see if I can train a model to pick the most optimal timeframe for the system.

Im working on being able to use some of the functions while in visual mode, for now live test best works on 3-4-5min so there some action.
this is a set file for 1hr xau that works full auto on backtesting (+visual)

that last 1hr set works well with 100,000 demo as its set, but you can change start fiblevel to 0, and closeall target at 30-70 probably and run it with 5k ... the next logical step is to have the ability to put in any indicator that has arrows.. should be easy enough. Also a note, when you have bank enabled instead of close all, the running total calculation (because profitable leg gets closed) get one sided. I have to work on the automatic profit taking calculation, its pretty good now but can be better.

Hey Cagliostro, can you give me a description of what your Oracle 3.6 is calculating? also dynamic vs persistent? morph?
Is that the latest oracle you made? It's been a main stay on my charts for a while, but I would love to know what it calculates exactly.
Thanks in advance

5Rivers wrote: Tue Apr 21, 2026 8:40 pm Hey Cagliostro, can you give me a description of what your Oracle 3.6 is calculating? also dynamic vs persistent? morph?
Is that the latest oracle you made? It's been a main stay on my charts for a while, but I would love to know what it calculates exactly.
Thanks in advance

Sorry it won't happen. Feel free to test whatever you want but am not inclined to open the code or math details. I did in in the past and it resulted in a badly coded clone of my work that did not even mention me as the original creator of the indicator. Oracle 3.6 is the latest publicly released version.
I am working on a completely different code since last year, you can read about some innovation / science behind it on the public whitepapers published on our website.

Hope you understand.

Best.

No worries, I understand, its highly praised.

Are you open to share the alkemix trend source?
I need to change it to calculate ha position every second or 2 instead of every tick so it doesn't hog much resources.
I also use your XU RSI MA which is great, but it would be nice to have it not recalculate so often as well - (i made a clone that Ill try to clean up to see if its even possible to save resources)

I forgot xard made XU... most of my tools come from the great creators on forex station... keep up the great work!

for todays London/NYC session, again I moved timeframes slightly to give the stepma some room for losers(bigger TF 4min->12min), or less room when deciding to take profits (smaller timeframe5-13min->3-1min)).... I did use my own bias to counter signal trade for quick scalps... but took the last signal when it was clearly slipping and joined the short for an explosive move down.

5Rivers wrote: Wed Apr 22, 2026 1:31 am I forgot xard made XU... most of my tools come from the great creators on forex station... keep up the great work!

for todays London/NYC session, again I moved timeframes slightly to give the stepma some room for losers(bigger TF 4min->12min), or less room when deciding to take profits (smaller timeframe5-13min->3-1min)).... I did use my own bias to counter signal trade for quick scalps... but took the last signal when it was clearly slipping and joined the short for an explosive move down.

HI,

the panels are too big how can I make the smaller?