Paper Trading Before Real Money

Paper trading is placing simulated orders in a broker's practice account, with live market prices and imaginary money. It proves you can work an order ticket, follow a written plan, and keep a journal. Two things stay untested inside a simulator: the cost of crossing the spread on every fill, and how a person behaves when the balance on the screen is their own.

What paper trading proves

A practice account is a mechanics lab. Inside it a beginner learns what a limit order does when the quote moves away from the stated price, how a stop behaves on a fast tape, what a $2,000 position looks like beside a $20,000 one, and how a written journal turns a hunch into a rule that can be tested. Those lessons transfer intact. The order ticket at a funded account is the same ticket.

A practice log also exposes plan defects cheaply. A strategy with no exit rule, a sizing scheme that puts a third of the account into one name, a routine that needs four screens watched during working hours: each of those shows up in a few weeks of simulated trading, at a price of zero.

What the simulator never charges you

Most simulators fill at the last printed trade or at the midpoint of the quote. A live marketable order pays the ask when it buys and receives the bid when it sells, and the gap between those two prices, the bid-ask spread, stays with whoever posted the quote.

Hypothetical arithmetic makes the size of the omission plain. A stock quoted $20.00 bid and $20.05 ask has a midpoint of $20.025. A simulator filling at the midpoint books the entry at $20.025. The live buy fills at $20.05, two and a half cents worse, and the exit gives up the other half of the width. One round trip, five cents a share, on a position the practice log recorded as free.

That toll is not uniform across names. The panel below measures the median quoted spread in basis points of the midpoint, where one basis point is one hundredth of one percent, across six widely held tickers over the most recent completed sessions on file as of July 2026.

SELECT ticker,
       round(quantileExactIf(0.5)(toFloat64(ask_price - bid_price) / (toFloat64(ask_price + bid_price) / 2), bid_price > 0 AND ask_price > bid_price) * 10000, 2) AS typical_spread_bps,
       round(quantileExactIf(0.5)(toFloat64(ask_price - bid_price), bid_price > 0 AND ask_price > bid_price) * 100, 1) AS typical_spread_cents,
       round(count() / 1e6, 1) AS quote_updates_m
FROM global_markets.cache_stocks_quotes
WHERE ticker IN ('SPY', 'AAPL', 'MSFT', 'NVDA', 'TSLA', 'F')
  AND sip_timestamp >= toDateTime(today() - 10)
  AND sip_timestamp < toDateTime(today() - 3)
  AND (toHour(sip_timestamp) * 60 + toMinute(sip_timestamp)) BETWEEN 810 AND 1199
GROUP BY ticker
ORDER BY typical_spread_bps

The tight end of the board, SPY, measured 0.27 bps, a median resting on 18.8 million quote updates. The wide end, F, measured 6.96 bps, a quoted gap of 1 cents. On a $10,000 order, one basis point is one dollar. A practice log of 200 round trips carries a real bill that the practice log itself never printed. What it costs to trade a stock prices that ladder across a wider board.

Fill quality moves with the clock

A simulator hands out the same fill at 9:31 a.m. and at 1:00 p.m. The live quote does not. Grouping every consolidated quote update for one mid-priced stock into half-hour buckets shows the shape of a session, measured over the same recent window.

SELECT formatDateTime(toStartOfInterval(toTimeZone(sip_timestamp, 'America/New_York'), INTERVAL 30 MINUTE), '%H:%i') AS et_time,
       round(quantileExactIf(0.5)(toFloat64(ask_price - bid_price) / (toFloat64(ask_price + bid_price) / 2), bid_price > 0 AND ask_price > bid_price) * 10000, 2) AS spread_bps,
       round(count() / 1e6, 2) AS quote_updates_m
FROM global_markets.cache_stocks_quotes
WHERE ticker = 'F'
  AND sip_timestamp >= toDateTime(today() - 10)
  AND sip_timestamp < toDateTime(today() - 3)
  AND (toHour(sip_timestamp) * 60 + toMinute(sip_timestamp)) BETWEEN 810 AND 1199
GROUP BY et_time
ORDER BY et_time

The 09:30 bucket, the first half hour of the regular session, measured 7.01 bps. The 12:30 bucket measured 6.94 bps, and the closing bucket at 15:30 measured 6.98 bps. A practice strategy that trades the first fifteen minutes and a practice strategy that trades midday are priced differently in a funded account, and identically in most simulators.

How much market one practice window covers

Four weeks of practice samples about twenty sessions. The panel below counts every regular session of a broad market fund over five calendar years, measured open to close, with the share of sessions that closed above their own open and the size of a typical daily move.

SELECT toString(y) AS year,
       count() AS sessions,
       round(100 * countIf(ret > 0) / count(), 1) AS pct_up_sessions,
       round(quantileDeterministic(0.5)(abs(ret) * 100, cityHash64(d)), 2) AS median_abs_move_pct,
       round(min(ret) * 100, 2) AS worst_session_pct,
       round(max(ret) * 100, 2) AS best_session_pct
FROM (
    SELECT d,
           toYear(d) AS y,
           toFloat64(session_close) / toFloat64(session_open) - 1 AS ret
    FROM (
        SELECT toDate(toTimeZone(window_start, 'America/New_York')) AS d,
               argMin(open, window_start) AS session_open,
               argMax(close, window_start) AS session_close
        FROM global_markets.delayed_stocks_minute_aggs
        WHERE ticker = 'SPY'
          AND toDate(toTimeZone(window_start, 'America/New_York')) >= toDate('2021-01-01')
          AND toDate(toTimeZone(window_start, 'America/New_York')) <= toDate('2025-12-31')
          AND (toHour(toTimeZone(window_start, 'America/New_York')) * 60
               + toMinute(toTimeZone(window_start, 'America/New_York'))) BETWEEN 570 AND 959
        GROUP BY d
    )
    WHERE session_open > 0
)
GROUP BY year
ORDER BY year

In 2021 the fund traded 252 regular sessions, 54.8% of them closing above their own open, with a median absolute move of 0.38%. In 2025 the same share ran 52%, the median absolute move 0.46%. The extremes of that year sit far outside the median: the weakest session moved -4.81% open to close, the strongest 10.12%.

Read the row counts as a coverage question. A twenty session practice window is roughly eight percent of one year, and the sessions it happens to contain decide what the log says. Practising across a quiet stretch and practising across a violent one produce different numbers from the identical rule set. Traders who want a verdict rather than a sample tend to run a practice log across several months, and to keep it running in parallel after real money starts.

The metrics that outrank win rate

A practice log reports win rate first, and win rate is the least informative line on it. Consider a hypothetical record of ten closed trades: nine wins of $50 each and one loss of $600. The win rate reads 90%. The account is down $150.

• Maximum drawdown, the deepest peak to trough fall in account value, is the number that describes what a strategy asks a person to sit through.
• Average win against average loss, paired with the win rate, gives expectancy per trade, which is the figure the equity curve actually follows.
• Trade count, where a log of twelve trades measures luck and a log of two hundred begins to measure a rule.
• Cost per round trip, taken from the spread panels above rather than from the simulator's fee field, which is often set to zero.

Sizing the first real trades

The common convention among trading educators is a graduated start: the smallest workable position size, the same written rules the practice log used, and a fixed number of trades before the size changes at all. One share of a $200 stock puts $200 at risk and produces the identical emotional data a hundred shares produce, at one percent of the exposure. The mechanics carry over from the simulator, the discipline does not, and small size is how the second one gets measured.

Size also interacts with cost. Fixed per trade frictions land heavier on small orders, and the spread lands proportionally the same at every size. A trader who plans forty small trades a month prices forty round trips, not one. Readers weighing a rules based alternative to discretionary entries can compare with dollar-cost averaging, where the schedule replaces the decision.

FAQ

Does paper trading actually work?

It works for what it measures: order mechanics, plan structure, and the discipline of journaling. It does not measure fill quality or emotion. A practice log and a funded log of the same rules over the same period usually differ, and the two gaps above are where the difference lives.

How long do people paper trade before using real money?

There is no standard interval. A useful frame is sample size rather than calendar time: a month covers roughly twenty sessions, and a broad market fund traded 250 sessions in 2025 alone. Many traders run a practice log for several months and keep it going alongside a small funded account.

Why are paper trading results better than real results?

Three mechanical differences account for most of the gap. Simulated orders often fill at the midpoint or the last trade rather than at the bid or the ask, simulated size never moves the quote, and a simulated loss carries no emotional weight. A paper trader holds through drawdowns that a funded trader closes early.

Can you paper trade options the same way?

The mechanics carry over, and the fill gap is wider. Options quotes are typically far wider in percentage terms than stock quotes, assignment and early exercise rarely appear in a simulator, and expiration handling varies between practice platforms. Treat a simulated options log as a mechanics test.

What does paper trading not teach?

Slippage, borrow availability for short positions, partial fills on larger orders, and the behaviour of a person watching real money fall. Those four are the standing gaps between a practice log and a funded account.

Every panel above ships with the exact SQL that produced it. Open one, swap in the tickers from your own practice log, and price your fills on the Strasmore terminal.