Evgeniy Scherbina / Profile

Happy Holidays and Happy New Year!

Wishing you a profitable year ahead, steady growth, and confident trading decisions in the Forex market.

To celebrate the holiday season, I’m offering 35% discounts on my selected best products:
LSTM Ensemble — new price $210 USD
Chanlun Master — new price $210 USD

May the new year bring strong trends, clean signals, and plenty of profits.
Trade smart, manage risk, and let the market work in your favor.

I am thrilled to announce 📣 that I will soon be releasing the Ensemble Trader, a new advisor that integrates 12 distinct models with stunning results 🚀💥.

Traditional machine learning often relies on a standard 80/20 data split 🤖, but it is strange to think 🤔 that the final 20% can truly "verify" the entire learning process. Over a long history ⏳, the market transitions through vastly different environments; while Brexit and the COVID crash 📉 are the most obvious examples, there are countless other specific trading regimes. A single rigid model simply cannot adapt to all of them effectively 🧊❌.

This adaptive strategy employs an ensemble approach where each sub-model 🧩 is responsible for covering a specific regime found in historical data. Instead of relying on a single logic, the models weigh their inputs ⚖️ to reach a robust consensus decision 🤝✅.

To be clear, this strategy still does not draw a perfectly straight equity line 🚫📈, because that is theoretically impossible in real markets. However, the Ensemble Trader solves many classic stability issues ⚓🛠️. It finally provides solid grounds to believe that earning with a fully automated Forex strategy is absolutely possible 💰🎯.

Here are the first steps of my newest strategy, Chanlun Master.
You can already check out a live demo signal on myfxbook.com.
You’ll notice that the strategy expects large moves and usually opens trades early in the day.

One of the key differences between this strategy and my previous ones is the way it’s trained. It’s based on time-series learning, which usually means you split the data by time — the “right” part of the data, the most recent one, is excluded from training. But after thinking about it, I realized this approach is both illogical and counter-productive.

So I changed it. I split the training into folds, where each fold has its own training and validation sets. Tests show that this method is more stable and reliable. Of course, the real results will become clear in a few months — fingers crossed!

Hi everyone! 👋

My new EA Chanlun Master blends the Eastern logic of Chanlun 🏯 with modern neural networks 🤖 to give you a clean and intuitive trading style.
It focuses on spotting Super Turning Points ⚡ — those moments when a trend finally runs out of steam and a strong reversal is likely.
Smart filtering keeps out the noise, and the system works smoothly across several symbols on higher timeframes 📈.

What really makes it shine is consistency: historical tests show stable accuracy, loss, and—most importantly—profitability 🚀 even on unseen data beyond the training period.
This kind of stability means the model isn’t overfitting, but actually learning real market behavior.

Chanlun Master is simple, focused, and surprisingly powerful.

The EurUsd has been moving sideways for months now, locked in a massive consolidation that started back in the summer. Anyone who's watched this pair long enough knows what that usually means — a big move is coming. The current range looks almost identical to the ones we saw in late 2024 and early 2025, and even to that long, painful consolidation at the start of 2024. History tends to rhyme, and right now it's whispering "trend breakout."

Most likely, that breakout will be to the upside, with potential targets around 1.20 to 1.40. It won't happen overnight, but the setup is already there. So what's been holding EurUsd back? Mainly politics and uncertainty out of France, where the financial and political crises are still dragging on. Germany already went through a similar phase, and its right-centrist forces managed to stabilize things. France, on the other hand, is still trying to figure out what's next — and Macron, being a republican rather than a socialist, has limited room to maneuver.

Once that cloud lifts, the euro could easily regain momentum. Whether it's by the end of the year or even in the next month, a strong rally in EurUsd looks more like a question of when, not if.

My Latte EA operates in a trend-following mode — it's built to catch big moves, not minor noise. And right now, it's been waiting far too long. The market's been quiet, but that silence won't last forever. What's clear is that Latte's cumulative position remains against the USD — and that's the right side to be on. When the breakout finally comes, Latte will already be where it needs to be.

Hi, everyone! I'm currently studying the ideas and potential value of the "Chanlun System". It's a long-standing set of principles, dating back to around 2006, used to find effective entry and exit points in the market.

This post is to announce that I'm integrating a small, yet powerful, component of the system—the inside bars—into my top EAs, Latte and Quantum Pip, in the very near future!

Inside bars, or "bars within bars," are simple but seriously effective in Chanlun analysis. They essentially shout, "The market is taking a breather!" They signal a moment of consolidation, which usually precedes a big move. They're all about potential energy building up in a tight range. Given the large, painful flat stalemate we've been stuck in since August, these inside bars are flashing major warnings. This principle is the perfect mechanism to help my EAs overcome this frustrating, sideways chop. October is finally setting up to fire big, and I want my top EAs to catch the full thrust of the breakout!

What’s up, guys! In this post, I’m going to discuss the first results from the "Dark Mars" EA in June.

The 6 major symbols are currently at 0 profit with the default settings. However, they are winning for both Filter=ATR and Filter=None (though UsdChf drags down performance for ATR, and EurUsd does the same for None - but the overall profit sum remains positive).

The EA isn’t showing profitability on my Myfxbook demo signal because my server wasn’t fit for the task. All my other strategies are daily, so occasional connection losses weren’t an issue. However, "Dark Mars" is an hourly strategy, meaning it missed many profitable exits due to disconnections. I’ll need to switch to a more reliable server for this EA.

Now, here’s the interesting part! I’ve long suspected that a technical scalper system could deliver outstanding results on cross-pairs - and yes, that’s true! You can test it on "real ticks" and "profit in pips for faster calculations" (it may not calculate profit at all without this flag set). Filter=ATR or Filter=None for cross symbols. It shows amazing results for AudCad, AudNzd, GbpChf, and NzdCad. I haven’t checked larger target values for potentially higher profits. The drawdown is close to 0.

Must try a new server on cross symbols!

Cheers! Stay tuned!

Why Transformers Beat LSTMs in Forex Forecasting

1/ Longer Context: Transformers digest more price bars, spotting trends LSTMs miss.

2/ Cleaner Growth: The attached image shows transformer passes (small pics) trending up steadily; LSTMs jump around.

3/ Efficiency Wins: We don’t need many passes—just one solid one. LSTMs can deliver, but transformers do it more reliably.

4/ Key Insight: My "2 numbers" method (forecast + its accuracy score) cuts noise. Transformers leverage both best.

What's up, guys. After running countless tests with LSTM networks - yes, LSTMs are still powerful, even compared to transformers - I stumbled upon a breakthrough in machine learning. Instead of just outputting a single probability (like 0.7 for a positive class), models should provide two numbers: the probability and its expected accuracy. The first number is the prediction; the second tells you how reliable that prediction actually is. This isn’t calibration — it’s a direct measure of confidence per output. I found that LSTMs can learn this naturally, giving not just guesses but self-aware guesses. Turns out, old-school architectures still have some tricks up their sleeve.

I’m attaching a complete test for Dec 2024 – April 2025 (1net + 2nets) that shows this approach is very viable. Historical curves fluctuate but grow steadily. And the best pass - selected by its F1 metric - always makes the sum for the symbols heavily profitable. I’m going to publish this version very soon, and then I’ll update the transformer-based EAs with this approach, too. Cheers, stay tuned!