I have had the same conversation about a dozen times now, and it always arrives at the same sentence: “Wait — so that number isn’t the stock?”
No. Not one number on TradeDM is about the stock. Every chart, every tile, every ratio is measuring one thing: how our signal did. The reason so many people are surprised by that is completely reasonable, which is why it is worth writing down properly.
Where the confusion comes from
Everywhere else you have ever seen a ticker next to a chart, the chart was the stock. That is the whole convention of financial screens. So when you land on a card headed NVDA with a green line going up and a row of numbers underneath, decades of habit tell you the same thing: NVDA went up.
On TradeDM the ticker is only there because a signal has to be about something. The green line is our call record on that ticker. The company itself could be compounding beautifully or heading for bankruptcy court, and the line would look exactly the same either way.
The cleanest proof I can give you
Roughly half of what our models publish are SHORT calls — a call that the price will be lower at the close than it was at the open. A short is profitable precisely when the stock goes down.
So take every short call the platform has ever published and score it two ways. First: what did the signal do? Second: what did those same stocks do, on those same days?
Across 23,813 short calls, the signal column reads +$22,159 and the stock column reads −$22,159. Identical, to the dollar, with the sign flipped. That is not a coincidence or a clever result — it is arithmetic. A short’s return is the negative of the move. But it makes the point better than any explanation I can write: the signal number and the stock number are not two views of the same thing. On half our calls they are exact opposites.
What this looks like on a real card
There is a model on NXDR — Nextdoor Holdings — whose card currently shows a cumulative +$78.19 on our standard $1,000-per-call basis, with 21-day alpha of +$237.67 and a win rate around 67%.
The solid green line is the signal’s cumulative P/L. The faint dashed line below it is the stock. They are going opposite directions on the same card, on the same days.
Over that window the stock fell about 20%.
Both of those are true at the same time, and neither is an error. If you read that card as “Nextdoor is doing well” you have read it exactly backwards. What it says is that a model kept calling a falling stock down, and kept being right.
Every instrument, restated
The fastest way to read our numbers correctly is to mentally add four words to the end of each one: …of the signal. P/L is what the calls made, not the price change. Win rate is how often the direction matched — a stock does not have a win rate. Sharpe, Sortino and Max Drawdown describe the shape of the signal’s daily results, never the share price. A model that shorts persistently can even run a negative beta on a stock whose own beta is strongly positive.
Alpha is the one that trips people up, because the stock does appear in its arithmetic — alpha is the signal’s return minus the stock’s. But alpha is still a statistic about the signal: it is how far the calls beat simply owning the thing. The stock shows up on the card exactly once, as that faint dashed line, and that line is the only stock information on the page.
The part that is easy to oversell, so I won’t
Here is the whole platform, every scored call since 27 April 2026, on the same $1,000 basis:
44,515 calls across 1,095 live signals. The signals: +$11,884. The same stocks, held over those same days: −$32,433. Directional accuracy: 50.27%.
Notice that this one is not a mirror — the two bars are nothing like each other, because the calls are a mix of longs and shorts across a thousand signals. Which is the point: you cannot derive either bar from the other.
And that last figure is the number a marketer would bury. 50.27% is barely distinguishable from a coin flip, and we do not claim otherwise — we have looked hard for durable directional skill and have not been able to demonstrate it to our own satisfaction. You will not find a word on this site promising that our models predict the market.
What those two bars do show is the thing this post is actually about: they measure different objects. The stocks lost money. The calls on them did not. You cannot get from one figure to the other, in either direction, and anyone who reads our scoreboard as a view on companies is reading the wrong scoreboard.
So what do you do with that?
Two things follow, and both are practical:
- A strong signal is not a recommendation of the company. Our best performer might be a model that keeps correctly calling a dying stock down. Nothing in its record says the business is sound.
- A company you like is not a reason to expect a strong signal. A model on a stock that tripled can still have a miserable record, if it kept getting the individual days wrong.
Pick a signal on the signal’s record. If you have a view on the company, that is a separate decision, and it is yours to make.
All of this now lives in the User Guide as its own chapter, and the full call-by-call record is public on the Ledger — including the losing days.
TradeDM is a financial publication, not an adviser. Nothing here is investment advice or a recommendation to buy or sell any security. Past performance does not guarantee future results.