Learn

A discretionary trader decides each trade by feel. A rule-based (systematic) strategy follows the exact same logic every time — no mood, no hunches, no “this time is different.” Every entry and exit comes from conditions you could point to on a chart.

Why does that matter? Because if you can’t say what triggers a trade, you can’t know when it will stop working, you can’t tell a rough patch from a broken edge, and you certainly can’t size it with confidence. “It just knows” is not a plan.

It also matters because of what’s being sold. A lot of “black-box” robots hide their logic for one of two reasons: there’s nothing underneath, or it’s a martingale that doubles down on losers — smooth gains for months, then one bad run wipes the account. Hidden logic is a red flag, not a feature.

Our systems trigger on things you can see: a volatility squeeze (the market coiling before it moves), a range break, agreement with the higher-timeframe trend. You don’t have to take the logic on faith.

Give a strategy enough adjustable settings and you can make its backtest look flawless on any chart — by tuning it until it fits every wiggle of that specific slice of history. That’s called overfitting (or curve-fitting), and it’s the number-one reason bought strategies fail the moment real money is on them.

The analogy: it’s like memorising the answers to last year’s exam. You’ll ace that exact paper and look like a genius — right up until the questions change. A market that has moved on is a new set of questions.

The test that separates a real edge from a memorised one is out-of-sample validation. You build and tune the strategy on one chunk of history (in-sample), then run it — completely untouched — on a different chunk it has never seen (out-of-sample). If the edge holds up, it might be real. If it collapses, it was a curve fit all along.

We split every strategy this way and reject the ones that only worked on the data they were built from. That’s most of them — and that’s the point. The rejections are how you know the survivors mean something.

A backtest is a simulation, and simulations are generous. Many assume you get filled at the perfect price, ignore the spread you actually pay on every trade, and quietly skip slippage — the gap between the price you wanted and the price you got when it mattered. Those small costs, ignored, can turn a losing system into a “winning” one on paper.

There’s a subtler problem too: cherry-picking. A seller can quietly run a thousand variations and only show you the one that looked spectacular. You never see the 999 that didn’t. That’s survivorship bias, and a single gorgeous equity curve tells you nothing about how it was chosen.

The only honest proof is live money on a real account, tracked by an independent third party the seller can’t edit — like Myfxbook. Real fills, real spread, real drawdown, all timestamped. It can’t be photoshopped after a bad week.

That’s why every Huskai system runs live and is publicly tracked before it’s offered. And it’s why you should be sceptical of anyone selling on backtests alone, no matter how pretty the curve.

New traders obsess over returns. Experienced ones obsess over drawdown — the peak-to-valley drop in the account along the way. Here’s the maths that makes it matter: lose 50% and you now need to make 100% just to break even. Losses compound against you faster than gains compound for you.

Blow-ups almost never come from low returns. They come from oversizing — risking too much per trade — and then riding a normal losing streak until the account is too small to recover. That’s “risk of ruin,” and it has ended far more accounts than any bad strategy.

The fix is boring and it works: cap the risk on every single trade to a small, fixed percentage, so no one trade — and no short streak of them — can do real damage. Then size the whole strategy so a bad month is a dent, not a death.

It also changes how you should judge a system. The honest measure isn’t return, it’s return relative to drawdown — the Recovery Factor. A system that makes 10% with a 4% drawdown is far better than one that makes 30% with a 40% drawdown, because you can safely size the first one up to whatever return you want, and you can never safely size the second.