The experiment
Under the
hood.
Three independent ideas, the same limits, and a separate guard that can always say no. Here is the experiment without the jargon.
Scout.
Scout follows an established opening-range-breakout checklist. Its picking logic stays frozen so changes in Quant can be compared with a consistent reference.
- 01
Scan the session
Look for a completed opening-range breakout in sufficiently active stocks.
- 02
Check the odds
Replay comparable historical moments. A research score, not a verified prediction.
- 03
Ask the risk guard
Reject anything that fails data, buying-power, time-window or stop-loss rules.
- 04
Paper trade and close
Place simulated orders with exits and go flat before the session closes.
Quant.
Quant is the research challenger. It builds a point-in-time dataset, tests alternatives and records virtual trade outcomes. It shadow trades while the evidence is evaluated.
- 01
Collect the evidence
Prices, volume, market context, earnings and every candidate signal.
- 02
Only what was known
Use only information available when a signal fired. Missing data stays missing.
- 03
Test on unseen periods
Judge ideas after modeled trading costs, not before them.
- 04
Challenge, don’t self-promote
Compare with Scout and require owner approval before the running strategy changes.
Atlas.
Atlas is the patient one. Instead of trading the same day, it holds a small set of stocks for months, when slow, lasting signals line up. Its rules come from the research lab and must pass a final exam first.
- 01
Wait for a reason
Only buy when a slow signal fires, the kind that plays out over months, not minutes.
- 02
Size it small
Spread money across many picks so no single stock can sink the account.
- 03
Hold, don’t fidget
Sell on a fixed schedule, not on daily swings. Fewer trades means lower costs.
- 04
Beat the market or bust
Score every result against simply owning the whole market. If it can’t beat that, it retires.
Non-negotiable guardrails
The answer can always be
“no trade.”
- Market access
- Paper only
- Direction
- Long only
- Leverage and options
- Disabled
- Stop-loss
- Mandatory
- Data errors
- Block new entries
- Strategy changes
- Owner approval
No stop guarantees an exact maximum loss in a live market. Paper fills are simulations and do not prove real-world execution. Trading runs in a standalone Python system; no chatbot is in the execution loop.