From Insights to Strategy Candidates (2/8)

The tools and mindset to sketch alpha ideas before you ever write a line of code.

In the previous newsletter, we explored the first step of the strategy building process: data analysis. That phase wasn’t about building models or signals, it was about observing, understanding the asset, and collecting raw insights.

Now, it’s time to turn those insights into ideas.
Not full strategies yet. Just clear hypotheses that might lead somewhere. Think of it as going from “this is weird” to “I wonder if I can trade that”.

In this newsletter, I’ll walk you through how I generate strategy ideas based on early observations, without falling into the trap of over-engineering too soon.

We’ll cover how to structure a trading hypothesis, assess whether it’s worth pursuing, and keep track of everything in a clean, testable way.

In the next issue, we’ll dive into data processing, where we’ll clean, align, and transform the data to actually test those ideas.

But first, let’s talk about how to go from curiosity to concept

---

1. From Raw Insight to Trading Hypothesis

The first step in idea generation isn’t to invent something out of thin air. It’s to look at what the data is already suggesting.

Maybe you noticed that after a spike in volatility, the price tends to revert more often than not. Or that when volume dries up, breakouts fail more frequently. Maybe the label distribution shifts under certain conditions, like we saw in the percentile analysis from the previous newsletter.

That’s your fuel. Your job now is to turn it into a clear hypothesis.

Something like:

“When X happens, Y becomes more likely.”

The best hypotheses are simple. Not simplistic, but clear. They’re testable. They leave room for nuance later, but they start with one question:
Is there a relationship I can build on?

And don’t forget the reciprocal:

If X increases the chance of Y, then not-X should reduce it.

If you’re only seeing one side of the pattern, it might be noise. True edges tend to have contrast: a situation that creates opportunity, and one that doesn’t.

At this stage, you don’t need a full strategy. You just need a believable story, grounded in something you saw, that could explain future behavior.
That’s your hypothesis.

---

2. Sketching the Trading Logic

Once you’ve spotted something interesting in the data, a shift in behavior, a pattern under certain conditions, the next step is to imagine how that could turn into a tradable idea.

You’re not writing rules yet. You’re just sketching the logic.

Start simple:
Is the signal directional or does it indicate a reversal?
Does it behave more like mean reversion, or is it a breakout pattern?
Does it suggest price continuation, exhaustion, or volatility expansion?

You might also ask:

• Would this idea make more sense on equities, crypto, futures?

• Is it likely to work better in trending environments, or choppy ones?

• Is this something you could time intraday, or more like a swing setup?

At this point, you're still allowed to explore with instinct, as long as it’s backed by something seen in the data.

You can even mix insights. Maybe you saw a pattern on BTC, but something similar showed up on a tech ETF. Don’t hesitate to connect the dots across assets. Good ideas often emerge from that kind of cross-asset intuition.

This sketch is your first blueprint.
You’re beginning to imagine what kind of signal might emerge, what kind of market conditions it might thrive in, and how it might behave across timeframes.

You’re not committing to anything yet, just giving shape to the idea.

---

3. Qualifying the Idea

Before jumping into backtests or building signals, take a step back. This is the moment to pressure-test your idea, without a single line of code.

Ask yourself a few critical questions:

• Can I actually exploit this in live trading?
  If the signal shows up once every three months, or requires acting in 2 milliseconds, it might be interesting, but not practical.

• Do I have enough data in those specific situations?
  An effect that appears only during high-volatility, low-liquidity selloffs might look good, but with 10 samples, you’ll never get a robust model.

• Is the behavior stable across markets or assets?
  Maybe the pattern looks great on Bitcoin, but disappears on ETH or equities. Does it generalize, or is it too narrow?

This is not the time to run regressions or test models. You’re just doing a sanity check:
Is this idea even worth developing?

A good idea isn’t just clever. It’s practical, testable, and grounded in enough data to be worth the effort.

If the answer is yes, then you’re ready to move on.

---

4. Organizing Your Strategy Leads

This phase isn’t about finding the strategy. It’s about collecting many leads that seem promising, and making sure you don’t lose them.

Each idea should go into your strategy notebook (or Notion, Obsidian, Excel, whatever fits your workflow), and it should follow a clear format:

• The core intuition
  What did you observe that sparked this idea?

• The pattern
  “After large drops, we often see a bounce,” “High volatility regimes tend to reverse,” etc.

• The potential target
  Is this a direction-based setup? A volatility shift? A breakout probability?

• Timeline
  Is this something you feel like testing now? Or a longer-term idea to explore later?

Over time, you’ll build a real library of alpha candidates.

And try to extract more than one idea per asset. If you’re analyzing BTC, don’t stop at “post-volatility breakout.” Look for effects around options expiry, daily seasonality, volume surges, etc.
Same goes for equities, ETFs, FX...

Most people only document the idea they end up coding. That’s a mistake.

By capturing all the possible leads, even the weird or niche ones, you give yourself room to revisit, combine, and refine ideas later.

The real edge doesn’t always come from the first draft. It comes from the version you sharpen after re-reading it weeks later.

You don’t need a perfect strategy at this stage, just a solid lead. Something that feels connected to the data, grounded in logic, and worth testing.

In the next newsletter, we’ll move on to Data Processing: cleaning, transforming, and preparing your dataset to actually test the ideas you just collected.

See you there.

👉 If you want to go deeper into each step of the strategy building process, with real-life projects, ready-to-use templates, and 1:1 mentoring, that’s exactly what the Alpha Quant Program is for.

It’s the full roadmap I use to turn ideas into live strategies.