Aspen Analytics
A behavioral economist named Alex Imas has been running experiments on why people want things — not the surface answer, but the deeper one. His finding, backed by experimental data: a growing share of what people spend money on isn’t about the thing itself. It’s about provenance, exclusivity, human involvement. Willingness to pay roughly doubles when people know something can’t simply be reproduced at scale. More striking: when AI is involved in producing something, that premium gets cut in half. Not because the output is worse. Because replicability changes what the object means.
This is an economics concept. But it describes something traders are about to feel very directly.
What happens when the cost of production collapses
Here’s the mechanism worth understanding. When the cost of producing something collapses, supply doesn’t just increase — it overwhelms. This is what happened to recorded music when distribution went digital, to photography when phones replaced cameras, to written commentary when anyone could publish. The market didn’t get better at finding good music or honest photography or rigorous analysis. It got worse, because the noise floor rose faster than any individual’s ability to filter it. AI doesn’t change this dynamic — it accelerates it by roughly an order of magnitude. The cost of generating plausible-looking market analysis is now functionally zero. Which means the question a trader faces every morning isn’t just “what’s the market doing” — it’s “how do I find anything worth reading in a sea of content that’s been optimized to look credible without the cost of actually being so.” That’s not a technology problem. It’s a judgment problem. And judgment is precisely what can’t be manufactured at scale.
The signal services feel this first. An alert system selling entries and exits is a commodity — the value is in the output, detached from the person or process that generated it. It can be packaged, replicated, undersold. AI doesn’t need a business model to produce a convincing entry signal. It just needs a prompt and a price feed. The floor of the commodity tier is dropping, and it isn’t coming back.
What three years of following signals actually builds
The typical retail trading arc goes like this: someone discovers trading, gets overwhelmed by noise, and reaches for a shortcut. A signal service, an alert system, a Discord group calling entries and exits. This feels like progress — the alerts arrive, the trades get taken, and for a while it even works.
But here’s what’s actually happening: every alert consumed is a repetition that builds someone else’s pattern recognition, not yours. After three years of following signals, you’re three years older but not meaningfully better at the thing that matters — reading what the market is doing and making a decision about risk. You’ve rented access to a process without building one.
This isn’t a moral failing. It’s a rational response to a market that makes signal consumption easy and skill development slow and uncomfortable. The problem is that the trade-off is about to get worse. Because the thing that makes signals vulnerable to AI is exactly what makes them a commodity in the first place: they’re replicable, scalable, detachable from human judgment. The cheaper AI makes commodity analysis, the more signal consumers compete with machines rather than learning to do something machines can’t.
What the migration actually requires
“Develop your own system” is advice that appears in every serious trading book and helps almost no one, because the obstacle isn’t knowledge — it’s the discomfort of operating without a net during the period when you’re building something real. The migration from consumer to practitioner runs through five stages, none of them fast.
The first is understanding why, not just what. The signal consumer knows what to do — buy here, sell there. The first real step is understanding why a setup has edge. What is the market structure condition that makes this entry valid? What breaks the thesis? This is the difference between executing a rule and understanding the logic behind it.
The second is developing structural literacy — a framework for reading the market that doesn’t depend on someone else’s interpretation. Where has the market made decisions before? Where does liquidity live? What does a regime change look like? Without this foundation, every setup floats in isolation.
The third is thinking about trades as risk decisions rather than opportunities. The question is never “where is the profit?” It’s “where is my risk defined, and does the potential reward justify taking it?” Until this is internalized, you’re still in the commodity tier — executing someone else’s risk decisions without building your own.
The fourth is building a feedback loop. Logging trades with reasoning, not just outcomes. Reviewing setups that worked and ones that didn’t with equal rigor. Asking not “did I make money?” but “did I execute the process correctly?” This is the stage almost no one does and the one that separates genuine development from expensive noise consumption. A trader who does this seriously for a year has data that tells them where their actual edge lives. That data belongs to them in a way no signal ever will.
The fifth is owning a methodology you can defend in plain language — not because a rule says so, but because you understand the market condition, the risk, and the statistical expectation behind it. The judgment, the process, the pattern recognition built from actual trades — these can’t be copied at scale.
None of this can be rushed. The time isn’t the obstacle — it’s the mechanism. The reps are the thing. You can accelerate the quality of your reps with good methodology. You cannot buy the reps themselves. That’s the part human nature keeps trying to route around, and AI is about to make the routing feel even more plausible while delivering even less of the actual thing. Time is a feature, not a bug.
Why domain expertise appreciates as the commodity tier collapses
The skilled cobbler isn’t threatened by the factory shoe. The factory shoe threatens the unskilled cobbler — the one whose only advantage was being cheaper than handmade. For the skilled cobbler, the factory shoe actually clarifies value. When everyone can have cheap footwear, what people are paying for when they commission something handmade is precisely the thing that can’t be manufactured: the judgment of someone who has spent years developing their craft, applied to your specific situation.
People don’t just pay more for handmade — they pay more because they viscerally understand it can’t be reproduced. The provenance is part of the product. Twenty-five years of live trading, a methodology stress-tested across multiple market regimes, the ability to look at a setup and distinguish the textbook version from the one where subtle regime conditions change the thesis entirely — this is not replicable at scale because it wasn’t produced at scale. It was produced by accumulated reps, losses, and the specific pattern recognition that comes from having real money at risk over time.
AI doesn’t threaten this. AI is the filter that finally makes the distinction between practitioners and consumers economically visible. When AI can produce adequate commodity analysis for free, adequacy stops being a selling point. What remains valuable is the judgment that knows when adequate is wrong.
The diagnostic
There’s a single question that tells you which side of this line you’re on, and it has nothing to do with your P&L.
Before you enter a trade: can you identify where your risk is defined, and is at least 1R of potential reward available from that structure?
Not “does the alert say buy.” Not “does the setup look like the one from the course.” Can you answer the question, in real time, with your own read of the structure?
A trader who can answer that honestly — yes or no, not approximately — is developing something that compounds. A trader who needs someone else to answer it first is in the commodity tier. And the commodity tier, as of roughly right now, is the most competitive it has ever been.
The noise is going to get louder. That’s not a reason to despair — it’s a reason to get serious about the thing that doesn’t scale.
Reflections from a long career in trading.
Aspen Trading Group is a registered Commodity Trading Advisor (NFA #0576114). Nothing published here constitutes trading advice.
