METHODOLOGY

Evidence is only as good as its assumptions.

Here is exactly how APAIIR tests a system, and where history stops telling you the truth.

Two layers

Statistics

Every time the rules fired since 2010, entered and exited, without capital limits. Good for judging whether an idea has an edge at all. Commissions are not included at this layer.

Simulation

History replayed day by day from the first day, with a starting capital, position size, your limits and commissions. The one to judge a real account by.

What we model

  • Every trade in order, from the first day of history
  • Capital that grows and shrinks with results
  • Position size, limits per asset and per system
  • Commissions in simulations
  • Stock splits and delistings
  • Several systems sharing one capital

What we don't model

  • Past index membership. Tests use today’s lists of the S&P 500, Nasdaq-100 and US 1000.
  • Order book and slippage. Fills are assumed at the modelled price. In liquid stocks the gap is small, but it is never zero.
  • Your broker. Real fills, fees and taxes differ. That is why you log every real order.

How to read a result

  1. How many trades?

    Fourteen trades prove nothing. Thousands start to.

  2. Every year, or one lucky year?

    Look at the year-by-year and month-by-month views, and the losing periods.

  3. Does a small change break it?

    If moving one number a little changes everything, the rule fits the past, not the market.

  4. Which size?

    Check the 20 ready-made simulations, or run your own, before choosing how much to put in each position.

  5. How often did you change the rules?

    The more you tune, the more you fit to history.

AI and your rules

AI drafts and edits rules. It does not validate them. Every rule it suggests goes through the same tests, and nothing runs until you approve it.

Past results, real or simulated, do not guarantee future results. Tell us when something looks wrong at support@apaiir.com.