Paper trading and forward testing, done properly

After this lesson you'll run paper trading as an instrument instead of a video game, and you'll know the promotion criteria between paper and real money.

The chain of evidence

Course 2 ended with backtests as arguments. The next link toward believing a method is forward testing: running the rules on data that didn't exist when the rules were written. History can be memorized, which is what overfitting is, and that's the whole reason a few months of forward results carry a different kind of weight than years of backtest. Paper trading is forward testing with you in the loop, execution and nerves included, and run properly it's the best instrument a developing trader has. Most people run it casually, and casual paper trading mostly builds confidence that nothing has tested.

"Properly" comes down to a short list:

Same rules, written first. The paper campaign trades the exact checklist, ritual, caps and management scheme you intend to trade live: School V/VI's documents, dated before the campaign starts. A paper trade that improvises tells you about your mood that morning, and the method goes untested.
Real costs, charged by hand. Paper fills arrive free and instant, while real ones pay everything School I, course 9 listed. Charge every paper round trip its realistic cost in the journal, and fill at the prices you'd actually get, which means the ask with the spread in it rather than the mid-quote.
No do-overs. The mulligan is what usually ruins casual paper trading. A paper trade that would have stopped out gets quietly forgotten, and a missed entry gets marked as taken after the fact. One rule handles all of it: the journal entry gets written before the outcome exists, per School V, and once something is in the journal it stays there.
Full size in R-terms. Trade the paper account at the same fractional risk you'll trade live, so every statistic transfers directly.

Two things paper can't test, said plainly: the emotional weight of real money, since School VII's biases run at a fraction of live intensity when the losses are theoretical, and real execution friction at size. That's why the chain has one more link after paper, which is going live at pilot size on School VI's ladder. The money is real, so the psychological instruments come online, and the amounts are small enough that the tuition stays cheap.

Promotion criteria

Between each link, written gates — decided before the campaign, like everything in this curriculum:

Paper → live pilots: a completed campaign (the 30-trade capstone is sized as one) with positive expectancy after hand-charged costs; average loss holding at ~1R (the contract, audited); process grades predominantly A (School VII, course 3 — paper exists to practice the discipline, and a paper campaign full of F-grades promotes nothing); and at least one drawdown handled per protocol.

Pilots → full size: School VI's ladder, on evidence — pilots surviving, live results consistent with paper (a large gap between them means costs, fills or psychology are unmodeled: find which before scaling), heat discipline intact.

Failing a gate sends you back a link with one change made (School V's rule). The loop is cheap to run, which is the whole argument for running it, and the mistakes that would have been expensive stay theoretical. Impatience with the sequence is School VII, course 4 playing out in slow motion.

Check yourself

Why is one month of forward results different in kind from ten years of backtest? (The rules couldn't have memorized data that didn't exist — overfitting is structurally impossible forward.)
Name the mulligan's damage. (Selective memory turns the sample into fiction, and every statistic downstream of it is measuring the editing rather than the method.)
Paper says +0.4R/trade; live pilots say −0.1R. What are the three usual suspects? (Costs undercharged on paper, fills modeled too kindly, or psychology arriving with real money — the journal's process grades isolate the third.)

The idea this lesson installs

Nothing journaled is ever unjournaled.

Next: Course 4 — "Simple statistics for traders."