I built the research operation I always wanted and could never buy.
Four models, a data pipeline, ten years of testing and the discipline to run it every trading morning. Inside a fund that is a hire and a budget line. Outside one it has mostly not existed at any price, and I never accepted that this was a law of nature.
That is where this is going. The coverage widens every month, the record gets longer including the months that go badly, and the work lands before the US open. I would like a dated, checkable record to be an ordinary thing to expect from anyone selling research, and the only way to argue for that is to keep one.
I learned to answer for a position before I learned anything else about markets.
My first job in markets was making prices. I sat on a G10 desk at NatWest in London quoting two-way risk in FX swaps and short-term rates, which means anyone could ask me for a price in either direction at any moment, and I had to give one. You find out what you really believe fairly quickly when the alternative is a number on a screen with your name against it.
What shaped everything afterwards was not the trading. It was the review. Someone senior would ask why a position was on, and the answer had to be reconstructable rather than merely persuasive. Vague did not survive that desk, and neither did clever.
Before the desk I trained as an ICAEW Chartered Accountant, which sounds like a detour and was not. Accountancy teaches you to read a set of numbers backwards, from the published figure to the judgement that produced it. Spend enough time doing that and you stop taking a headline number at face value. You start asking who chose it, and what it was chosen instead of.
Since 2021, I have also led the FOMC Press Conference commentary at Seeking Alpha. That meant publishing a read on the most scrutinised event in the calendar, with a timestamp on it, in front of an audience that told me immediately when I was wrong. It is a useful thing to have done. It removes the option of quietly revising a view afterwards, and I have kept that constraint here on purpose.
I got tired of being the least reliable part of my own process.
There is a particular feeling I would rather never have again. Someone asks why you are holding something, you begin to answer, and halfway through the sentence you hear that what is coming out is not reasoning. It is a preference that has learned some vocabulary. On a rates desk that gets caught the same afternoon. Almost everywhere else it does not, and people carry views around for years without ever being made to show the working.
I could have tried to become more disciplined. Plenty of people do, and it holds right up until a genuinely bad month arrives. I spent the past two years building something that does not need me at my best instead. Four independent models read every market I follow before the US open. They agree or they do not, and you see which. A portfolio is built from what they agree on and published every Monday with exact weights. Afterwards the record says what came of it, and it cannot be edited.
The part I did not expect was the relief. My role has switched from someone who wears their views on the sleeve and try to defend them all the time, to a quieter observer of what the models tell me about the markets. This means I am spending more time learning than trying to confirm a cognitive bias, aka my ego.
You get the same page I do, at the same time. You are not getting a summary, or a version prepared for subscribers, or a conclusion with the working removed. If it helps you, it will help in the same unglamorous way it helps me. The work happens on the difficult mornings as well as the easy ones, nobody has to be right for it to keep running, and there is a dated record either way.
Some people use it as a portfolio. Some use it as a second opinion to argue with. Both are fine, and neither requires you to take my word for anything. What you do with your money stays yours, and so does the credit for it.