Risk management on FX Backtest Diary

FX Swap Shock: Is Your EA Losing Money While You Sleep?

Mon, 01 Jun 2026 00:00:00 +0000

A beginner-friendly summary of the verification: “FX Swap Shock: Is Your EA Losing Money While You Sleep?”.

What’s the idea?

When we talk about taking an Expert Advisor (EA) from backtesting to live trading, it’s easy to focus just on the raw profit/loss numbers. But there are crucial “hidden” costs and benefits that can make or break a strategy in the real world. Today, we’re diving into one of those big ones: swap costs. What are swaps? In FX trading, a swap is basically the interest you either pay or receive for holding a position overnight. It’s determined by the interest rate differential between the two currencies in a pair. If you’re holding a currency with a higher interest rate against one with a lower rate, you might earn swap (a “positive carry”). If it’s the other way around, you’ll pay swap (a “negative carry”). For our particular EA, we found that trades are held for an average of 6.8 days. That’s long enough for swap costs and benefits to really add up and impact overall profitability. So, understanding them isn’t just an academic exercise – it’s vital for knowing what to expect when we go live.

Real-World FX: Can Our EA Survive the Hidden Costs of Trading?

Sun, 31 May 2026 00:00:00 +0000

A beginner-friendly summary of the verification: “Real-World FX: Can Our EA Survive the Hidden Costs of Trading?”.

This time, we’re diving into a crucial topic for any automated trading strategy: how well it stands up to the real-world costs of trading. Our focus today is on Core v1.4.0, a low-frequency Expert Advisor (EA) that holds trades for an average of 6.8 days. Over 11 years, it’s executed 6168 trades, which gives us a solid dataset to stress test its resilience.

The AI Sizing Challenge: Why Contextual Bandit Couldn't Optimize Leverage

Sat, 30 May 2026 00:00:00 +0000

A beginner-friendly summary of the verification: “The AI Sizing Challenge: Why Contextual Bandit Couldn’t Optimize Leverage”.

What’s the idea?

We’ve explored various ways to make our algorithmic FX trading systems (EAs) smarter, especially when it comes to deciding how much to trade – what we call “position sizing.” This is like a chef adjusting the amount of spice based on the ingredients and the diners’ preferences. Get it right, and you could boost returns while managing risk. Get it wrong, and… well, let’s just say it’s not good! Previously, we tried using Machine Learning (ML) for this (Research 100), aiming to dynamically adjust leverage (how much borrowed money we use for trades) based on market conditions. But that experiment didn’t pan out; the ML system performed no better than a “placebo” (a shuffled version of its own output). It seemed the underlying market structure it was trying to learn was either too complex or already captured by our existing, simpler methods. This time, we’re trying a different flavor of AI: a contextual bandit. Think of it as a simpler, more focused type of reinforcement learning (RL). Instead of learning a complex sequence of actions, a contextual bandit learns the best single action to take in a given “context” or market state. Our goal was to see if this approach could learn to pick optimal leverage levels (0.5x, 1.0x, or 1.5x) based on different market scenarios, outperforming our existing, hand-coded sizing logic.

Crisis Alpha: Why Our Safe Haven EA Strategy Didn't Deliver

Fri, 29 May 2026 00:00:00 +0000

A beginner-friendly summary of the verification: “Crisis Alpha: Why Our Safe Haven EA Strategy Didn’t Deliver”.

What’s the idea?

Sometimes, even the most robust Expert Advisors (EAs) have downtime. Our v1.4.0 system, for instance, is designed to reduce its core positions when the stock market signals “risk-off” – that’s when investors are nervous, stocks are falling, and they’re looking for safer places to put their money. This makes sense for capital preservation, but it also leaves a bit of an “empty slot” in our trading activity. So, we wondered: could we fill that slot? The hypothesis was to introduce a “crisis alpha” strategy. This meant actively trading traditional safe-haven assets (like buying Japanese Yen or Gold) specifically during these risk-off periods. The goal? To generate profit when our main system is taking a cautious stance, essentially acting as a complementary hedge. Think of it like a smart backup generator that kicks in exactly when the main power grid gets shaky!

Our EA Just Broke the Monthly Profit Ceiling! Find Out How.

Tue, 21 Apr 2026 00:00:00 +0000

A beginner-friendly summary of the verification: “Our EA Just Broke the Monthly Profit Ceiling! Find Out How.”.

This time around, we explored a fascinating way to potentially boost our algorithmic trading systems (EAs) beyond what we thought was possible: by combining an existing, reliable strategy with a completely new one. The goal was simple but ambitious: break through a perceived ceiling on our monthly profits.

Was It Just Hindsight? Our Forward Validation Delivers a Decisive Conclusion

Mon, 30 Mar 2026 00:00:00 +0000

A beginner-friendly summary of the verification: “Was It Just Hindsight? Our Forward Validation Delivers a Decisive Conclusion”.

What’s the idea?

Today, we’re diving into a crucial experiment to see if we can find a truly sustainable edge in algorithmic FX trading. The big question is: can we pick a winning portfolio of currency pairs for the coming year solely based on their past performance, without any “peeking” into the future? To answer this, I designed what’s called a complete Out-of-Sample (OOS) testing approach. Think of it like this: imagine you’re picking a fantasy sports team for the upcoming season. You’re only allowed to look at player stats from previous seasons. You pick your best team, and then you see how they perform in the new, unseen season. That’s exactly what we did, but with currency pairs and trading strategies.

Weak Alone, Strong Together? Unlocking Stability with Combined Micro-Edges

Sat, 28 Mar 2026 00:00:00 +0000

A beginner-friendly summary of the verification: “Weak Alone, Strong Together? Unlocking Stability with Combined Micro-Edges”.

What’s the Idea?

Ever wonder if you need a single, super-powerful trading strategy to succeed, or if combining smaller, less flashy ones could be the secret? That’s exactly what I’ve been exploring! My latest research dives into the concept of building a robust trading system by stacking up multiple “weak” but uncorrelated trading edges. The goal isn’t to hit a home run, but to create a highly stable, low-risk portfolio that consistently generates profits, making it ideal for things like prop firm challenges where capital preservation is key.

Prop Firm Challenge: Unmasking Our EA's True Expected Value

Thu, 19 Mar 2026 00:00:00 +0000

A beginner-friendly summary of the verification: “Prop Firm Challenge: Unmasking Our EA’s True Expected Value”.

What’s the idea?

We’ve been looking into the Fintokei Quartz challenge, a popular prop firm offering where traders aim to prove their strategy on a simulated account to eventually manage real capital. The specific challenge we focused on was for a ¥1,000,000 account, with a ¥12,500 participation fee and a 1% risk set for each trade. Our goal was to calculate the “overall Expected Value” (EV) of taking on this challenge with a specific EA (Expert Advisor, or automated trading strategy). In simple terms, EV tells us, on average, how much profit or loss we expect to make each time we attempt the challenge, factoring in the costs and potential payouts.