UNCALIBRATEDA heuristic score that ranks setups. It has not been fitted to real outcomes, so it is not a probability.
Methodology
How QuantOS decides what a number means
A trading tool earns trust by being clear about what it doesn’t know. These are the rules every QuantOS page follows.
Principle One
Four questions, scored separately
A losing trade can be a good decision, and a winning trade can be a lucky one. QuantOS never blends these into a single grade.
Setup Quality
Was the market offering a real setup when you looked?
Entry Timing
Did you enter at the setup, or chase after it had already moved?
Execution Quality
How much did slippage, size and fees cost compared with the price available?
Outcome
What the trade actually made or lost, and how much of the move you captured.
Principle Two
Every result carries a label
Labels appear next to the figure itself, not in a footnote.
ESTIMATEDInferred from partial inputs. Dealer positioning taken from open interest is an example.
Derived, Not ObservedCalculated from other data, not reported by the broker or exchange.
PARTIALToo few qualifying trades to score. The page shows how much more evidence it needs.
REFUTEDThe signal failed its control test. It stays visible as context but has no vote.
Principle Three
No evidence, no score
Missing data lowers confidence. It is never filled in or averaged away.
Each behaviour rule needs a minimum number of matching trades before QuantOS will put a cost on it. Until then the rule shows as PARTIAL, with the number of trades it has and the number it needs.
If an input is unavailable, such as an options snapshot that wasn’t recorded at the time, the result says so. QuantOS doesn’t rebalance the remaining inputs to produce a confident-looking number.
Principle Four
Signals must beat a control
An idea becomes a signal only after it outperforms a fair comparison.
The SPX Intelligence Lab tests each candidate signal against a control on the same sessions. Signals that don’t beat it are marked REFUTED. They can still appear on a chart for context, but they can’t confirm, veto or score a trade.
This rule has removed popular ideas from the decision gates when they failed the test.
Principle Five
Hindsight doesn’t rewrite the decision
Review grades a decision by what was knowable when it was made.
Outcome data can grade a decision later, but it can’t change what the gates said at the time. Replays reveal bars one at a time, and markers don’t appear before their bar.
News read by a language model is shown as interpretation, kept apart from the measured price reaction, and always marked Price Confirmation Required. It is never an entry signal by itself.
Data Sources
Where the numbers come from
Charles Schwab
Accounts, positions, transactions, quotes, price history and option chains through the official API, plus your brokerage statements.
Official Releases
Economic calendars and headlines from the BLS, Federal Reserve, Treasury and SEC.
Recorded Sessions
Options snapshots that QuantOS records during the day, so later reviews use what was visible at the time.
What QuantOS Doesn’t Claim
No consolidated options tape, dark-pool prints or proprietary dealer flow. Where a competitor shows these, QuantOS says the data is unavailable.