Start here

1. What this is

Two things under one name. A private research program builds and tests a trading model. This site publishes the record of that model’s decisions — the part worth having — while the internals stay private, because they would stop working the moment they were public.

GSS is a free paper-trading platform and paired research engine that scores, tests, and records stock trades. It works in two stages, and they are separate. First the pre-market scan reads roughly 1,500 stocks and picks the day’s strongest names, which join a hand-curated list to form that day’s watchlist. Then the model works that watchlist and nothing else: it builds a statistical signature for each name and scores it against the structures we favor, using price trends, trade volume, and moving averages to measure momentum that is already under way.

2. What the model does

It follows that day’s watchlist on a fixed schedule — it never hunts outside the list it was handed, because finding names is the scan’s job, not the model’s — and asks one question: is the trend aligned, and is the alignment strengthening? When it is, the model records a setup and scores it 0 to 1. Then it checks a longer timeframe — if the bigger trend disagrees, the setup is vetoed rather than taken.

It follows trends; it does not predict them. The refusals are as interesting as the entries, and every refusal is written down too.

3. Why the record is worth trusting

Before the model sees new data, we write down what we expect and what would prove us wrong. Then every live decision — trade and no-trade alike — is logged to a file that cannot be edited afterwards, before the outcome is known.

Everything on this site can be checked: the board and the public feed show every decision, delayed 15 minutes, free. The white paper spells out the evidence standard we hold ourselves to — and says plainly where our own evidence falls short of it, including why we withdrew the reliability dataset we used to offer for download. If you find a claim the evidence does not carry, tell us and we will correct it in public.

4. Where this stands

The model went live in August 2026 and has been logging forward evidence since. The record is young — too young to support performance claims, and we would rather say that plainly than dress it up. Watch it accumulate on the model’s own book.

5. Try it with play money

The paper trading desk gives you $100,000 of play money. Pick a stock, buy or sell, and the fill uses exactly the price you can already see on the board. Nothing is real and nothing is at risk.

Signing up takes one tap — passkeys, so your device’s own fingerprint or face unlock, with no password to invent and no email to hand over. Your trade log keeps every fill, so you can compare your calls against the model’s.

6. Words you will see

Watchlist

The fixed set of stocks the model follows. It does not go hunting outside it.

Timeframe

How much market time each reading covers. Trading decisions run on a short timeframe; the permission check runs on a longer one.

Regime gate

The longer-term trend check that can veto a setup. “Good setup, wrong market” is what it exists to catch.

Exit watch

A flag that a rising trend is showing signs of stalling. An early warning, not a prediction.

Delayed / stale

Free readings run 15 minutes behind live prices. Outside market hours the board freezes and is marked as last session’s data.

For the board’s card states — entry, no trade, vetoed, earnings blackout, unavailable — see the signal board itself.

7. Not advice, not a prediction

The model does not know you, your situation, or your goals, and nothing here is a recommendation to buy or sell. A higher score has not been shown to mean a better outcome — which is why the score is presented as the model’s opinion, not a forecast. The white paper has the full argument, the evidence standard, and what is not built yet.