Why Tradoki will never sell signals

The single most asked question I get from prospective Tradoki students is some version of "do you have a signals room?" or "will I get the trades?" or "do you publish your daily picks?" The answer is no, the answer will continue to be no, a

The single most asked question I get from prospective Tradoki students is some version of "do you have a signals room?" or "will I get the trades?" or "do you publish your daily picks?" The answer is no, the answer will continue to be no, and I want to spend a long-form piece explaining why — because the reasoning is the foundation of everything else we do, and the choice is the one I am proudest of and the one that costs us the most revenue. We do not sell signals because we have never seen a consumer signals product produce a sustained edge for the median customer, and we are not in the business of selling people things we have not seen work.

This is a personal piece, in the first person, about a business decision. It is not a marketing pitch, although the choice it describes is the reason the company is what it is. I am writing it because the question keeps coming up and I want a single place to point to when it does. Educational only, none of it investment advice.

The economics of selling signals

I want to start with the part of this that does not get said often enough in trading education: signals are the easiest revenue product in the industry. The unit economics are excellent. Marginal cost per subscriber is approximately zero (you publish the same signal to all of them). Customer acquisition cost is recoverable in two to three months at typical pricing. Churn is high but predictable, and the inflow of new subscribers — driven by the perpetual hope of finding the system that works — is reliable enough that the LTV/CAC math closes comfortably.

The problem is not that signals are a bad business. They are a great business. The problem is that they are a great business in the same way that lottery tickets are a great business — the product captures the customer's hope rather than producing the outcome the customer is paying for, and the cumulative effect on the median customer is negative.

I will not pretend I have never been tempted. The first product spec I drew up for Tradoki, in early 2024, was a hybrid course-plus-signals-room model, because the spreadsheet for that model is significantly better than the spreadsheet for pure education. The reason we did not ship it was not financial. It was that I could not square it with the data on consumer signals services and the personal experience I had with the question.

The personal experience that ended the conversation

In 2023 I was running a small trading group as a side project. The group was free, the participants were friends and friends-of-friends, and I would post an idea once or twice a week — what I was watching, what I would consider doing, why. The idea-posts were not signals. They were narrated reads, with the bias, the level, the trigger I would want to see, and the conditions under which I would skip the trade.

After about six months I noticed a pattern. The participants who treated my posts as signals — who entered when I described an idea and exited based on my levels — produced a worse trading record than the ones who used my posts as starting points for their own analysis. The latter group were learning. The former group were copying, with a one-bar lag that compounded over time, on a platform with execution different from mine, sized arbitrarily, and in the absence of the higher-timeframe context I was reading.

The participant who flagged this to me was, with embarrassment, the one who had blown up a small live account doing exactly what I had described, on the bar I had described it. He had not done anything wrong. He had done what the format invited him to do. The format was the problem.

I shut the group down two weeks later. The lesson I took away was that the act of publishing a level to be acted on is structurally different from publishing analysis to be learned from, and that the difference is large enough to matter even when the underlying analysis is identical. The signal format is the problem, not the signal content.

When the spreadsheet for Tradoki said "add a signals tier and the LTV/CAC closes in 90 days," I remembered the participant in the 2023 group, and I said no. The spreadsheet has been broken in our favour ever since.

What signals do to the customer

The technical reason signals are the wrong product is that they short-circuit the only part of trading that matters: the trader's own decision-making.

A trader who follows a signal feed is not learning to read the market. They are learning to read the signal feed. The skill that develops is the skill of executing other people's analysis on a platform, on a schedule, with a sizing rule. It is not the skill of analysis. It is not the skill of risk management. It is not the skill of regime recognition. None of those skills transfer when the signal source goes away — and the signal source always eventually goes away, because no signal source survives forever.

The trader who completes a structured education in trading, on the other hand, is doing the exact opposite. They are building the analysis muscle. They are learning to recognise the regime that calls for the strategy. They are learning to size correctly. They are learning to journal. The skill survives the absence of any specific tool, vendor, or service. The compounding of the skill is the entire game.

A signals customer who has been on a service for a year has a year of subscription fees and a portfolio that is at best break-even (in our records). An education customer who has spent the same money on a course and ninety days of structured practice has a year of accumulated competence and a journal that explains how their account moved. Both have spent the same money. One has a transferable skill; the other has a habit of clicking buttons.

This asymmetry is the entire reason we sell what we sell.

What the market wants vs. what the market needs

The honest tension in our positioning is that what the market wants is signals, and what we sell is education. The market is louder about wanting signals because the desire for signals is the desire for an answer that does not require the work, and the desire for an answer that does not require the work is universal.

Education is harder to sell than signals because it does not promise the avoidance of the work. The marketing pitch for signals is "subscribe and follow." The marketing pitch for education is "subscribe, study for 60 hours, journal for 90 days, paper-trade for 60 sessions, and you might be ready to risk a fraction of an account on a small position." The first sells faster. The second is the only one that produces an outcome the customer would, on reflection, actually want.

The compromise I have watched competitors make is the hybrid model — sell education and signals, with the signals as the carrot that keeps the customer paying while the education is consumed. The hybrid model is the worst version. The customer takes the signals; the education sits in the LMS unread; the trader does not develop; the subscription continues until the next bad month forces a cancellation.

We do not run the hybrid because the hybrid is the model that takes the customer's money the longest while teaching them the least. The full education-only model is harder to sell and produces worse short-term unit economics. It is also the only model I am willing to attach my name to.

What we do publish, and why it is not a signal

We do publish trading content. We publish setups, frameworks, and walkthroughs of historical sessions. We publish analytical pieces like the one you are reading. We publish weekly desk notes describing the regime we are in.

None of that is a signal in the sense that matters. The setups describe patterns. The frameworks describe analytical approaches. The walkthroughs explain how a discretionary trader could read a session. The desk notes describe what we are watching, not what we are doing.

The student who reads our desk notes and then asks themselves "given this regime read, what would I want to see on EUR/USD before considering an entry, and what would my plan be if it formed?" is doing the work. They are using the desk note as a thinking aid for their own analysis. The student who reads the same note and asks "what is Arthur trading today?" is asking the wrong question, and the answer would not help them in the way they think it would even if I gave it to them.

The structural difference between content and signals is the locus of the decision. Content moves the analysis to the student. Signals try to outsource it to us. We are not in the outsourcing business.

The honest cost of the choice

I do not want to pretend the choice has been costless. It has not. Every quarter I look at the conversion rates and see a number that would be 30–50% higher if we ran a signals tier. Every cohort, a meaningful fraction of new applicants leave the funnel at the moment they discover we do not run a signals room. The unit economics are demonstrably harder than the alternative, and the runway implications are real.

The choice has also produced a different kind of customer. The students who join Tradoki, knowing what we sell and what we do not, self-select toward a population that wants to actually learn — not a population that wants to be told what to click. That self-selection is a quieter compounding effect than the LTV/CAC spreadsheet, but it is the one I am betting the company on. Cohort outcomes have been meaningfully better than industry comparables in our records, and I attribute most of the difference to the kind of person who decides to join us in the first place.

For the broader picture of why we built the company at all, see why we built Tradoki — an AI trading academy, not a signals room. For what we actually teach, see inside the eight-week syllabus. For why the cohort model produces those outcomes, see the cohort model and why trading students finish.

We sell education because we sell what works. The harder business is the only honest one available, and the cumulative compounding of student outcomes is the only marketing I am willing to fund.

— Author's note

The lesson in here for the reader

If you take one thing from this piece, let it be this: the next time someone offers to sell you signals, ask them what skill you will have at the end of a year that you do not have today. If the answer is "you will have a track record of following our signals," that is not a skill. It is a subscription history. If the answer is "you will be able to read the markets we cover well enough that you would not need to subscribe to us anymore," ask them whether they have ever produced a customer who chose not to renew because they no longer needed the service. The answer will tell you whether the product is education or rent.

The product I am willing to sell is the one whose success criterion is the customer's eventual independence. That is education by definition. Signals are the opposite by definition. The two products live in the same market and use the same vocabulary, and they could not be more structurally different in what they do to the customer over a year.

For the underlying risk math that makes "small enough to learn safely" a defensible position size, see the risk-of-ruin pillar. For the structured arc that produces an independent trader at the end of it, see the ninety-day deliberate practice plan.

● FAQ

Why don't you sell signals?

Because we have never seen a consumer signals product produce a sustained edge for the median customer, and selling something we do not believe works would be incompatible with the kind of company we want to run. We sell education, which compounds; signals fees compound the wrong direction.

What is the difference between a signal and a setup?

A signal tells you to enter at a specific price. A setup describes a pattern you are responsible for recognising and acting on. The first transfers no skill; the second is the skill itself.

Are all signals services bad?

I have not seen a consumer-priced one that produced sustained net-positive results for the median subscriber over a one-year window. Institutional research products are a different industry; they are not what the retail market is buying.

What does Tradoki sell instead?

Structured education — an eight-week curriculum, a deliberate-practice program, and journaling tools — designed to make our students self-sufficient. We measure success by how well students do without us, not by their renewal rate.

How does this affect your business model?

It makes the unit economics harder. Education has higher dropoff than subscriptions and a longer payback. We made the trade because the alternative — selling something we do not think works — is not a business we want.