# A $100 trading account or a prop firm challenge? One year, simulated - specification

Specification written and **frozen 2026-10-16 (approved by the owner the same day, draft 2)**, before any simulation of this study was run and before the tick data was exported. (Draft 1: a $10,000 challenge and traders defined in R only. The owner asked for a realistic scalping case on both accounts, a $5,000 challenge, 0.01 lot on both, no stop, a $5 target, trades closed within 40 seconds, 15 trades a day.) Once frozen, any change is documented; bugs may be fixed.

## Why this study, and what it builds on
- **Chosen by the owner** on 2026-10-16, after the question "why is a $100 account better than a prop firm?": an article that answers it with numbers, in two parts: a general comparison by trader edge, and a realistic case, the owner's own style of gold scalping, on real tick data.
- It reuses the trader profiles and the challenge engine of "How to Pass a Prop Firm Challenge" (`pc_spec.md`, `pc_core.js`, frozen 2026-10-14). New here: the $100 live account, the funded year with payouts, the comparison of the two, and the tick-based scalper.

## The challenge (both parts; generic, not any firm's rules)
- **$5,000, two phases.** Phase 1: target +8% ($400); phase 2 (a new $5,000 account): target +5% ($250). Maximum loss 10% static ($4,500 floor), daily loss limit 5% ($250) from the day's starting balance, at least 4 trading days per phase. Checked on the balance after each closed trade, as in the earlier study.
- **Fee $35 per attempt** (illustrative: $5,000 two-step challenges were offered for roughly $29 to $36 in October 2026); the article shows how the results scale with the fee.
- **Funded account** after both phases: $5,000 with the same loss limits. **Payouts:** every 10 trading days after funding, if the balance is above $5,000, the profit is paid out, **80% to the trader**, and the balance goes back to $5,000. The account ends at a breach. Payout rules differ by firm; these are generic.
- **The year counts from the first day of phase 1.** **B, one attempt:** one fee. **B-R, buying again:** a new attempt after every failure or breach while time is left in the year (at most 12 attempts); fees counted.

## Part 1: traders by edge (the general case)
- **Traders (as in `pc_spec.md`):** every trade risks R and wins 1.5 R or loses 1 R after costs. Negative edge: 38% winners, -0.05 R a trade; no edge: 40%, 0 R; small edge: 42%, +0.05 R; good edge: 46%, +0.15 R. 3 trades a day, 250 trading days.
- **A, a $100 live account:** each trade risks 1% of the current balance, never less than the minimum the smallest position allows: **A1** $1 (0.01 lot of EUR/USD with a 10-pip stop), **A2** $10 (0.01 lot of gold with a $10 stop). Below that minimum the account stops (wiped out). No deposits or withdrawals.
- **B and B-R:** the challenge above with **$25 risk a trade** (0.5% of $5,000).
- **Reported:** for A, the final balance (median, mean, 5th and 95th percentiles), the share down by half or more and wiped out; for B and B-R, the share passing, the median days to pass, the share with a payout, the money paid to the trader (mean, median), fees paid, the net result and the share ending the year ahead.
- **100,000 simulated years per cell**, fixed seeds.

## Part 2: a gold scalper on real ticks (the owner's style)
- **Data:** the broker tick history of XAUUSD (bid and ask) exported from the owner's MetaTrader 5 terminal by `ExportTicks1s.mq5` as one record per second (bid and ask open, high, low, close), October 2025 to October 2026 as far as the server provides it. The file's hash, span, gaps and spreads are recorded before anything is simulated. (MetaTrader 4 keeps no tick history, so the 1-minute files of our other gold tests cannot show what happens inside 40 seconds.)
- **The trades:** **0.01 lot (1 ounce: $1 for every $1 of price) on both accounts.** **15 trades a day**, Monday to Friday, at random seconds between 08:00 and 17:00 London time, one at a time. **No stop.** Exit at **+$5** (the bid reaches entry + $5 for a long, the ask reaches entry - $5 for a short), otherwise **at 40 seconds** (the owner closes "by eye" within 40 seconds; we use the latest moment). Buys fill at the ask and sell at the bid, shorts the mirror: **the real spread is the cost**; a variant adds $0.07 a trade (a $7-a-lot commission).
- **The traders:** direction at random (no skill), and traders who know the direction of the next 40 seconds part of the time: with probability s they take the side the mid price will move to over the next 40 seconds, otherwise a coin flip; s = 10%, 20% and 30% (right about 55%, 60% and 65% of the time). The article reports the accuracy a scalper needs to break even.
- **A, the $100 account:** trades stop if the balance falls below the margin of one 0.01 lot at 1:500 (price x 1 ounce / 500).
- **B and B-R:** the $5,000 challenge above, with the same 0.01-lot trades.
- **Reported:** the same as Part 1, plus the cost per trade, the share of trades closed at +$5 and at 40 seconds, the worst 40-second result, and the days the scalper needs to reach each target.
- **A note for the article, not simulated:** some firms do not pay profits from trades held under one or two minutes or ban "tick scalping"; under such a rule this style is not paid at all. Readers must check their firm's rules.
- **10,000 simulated years per trader** (each year uses the same real days with new random entry times and directions), fixed seeds.

## Data note, added 2026-10-16 before any simulation
- The ticks were read through the terminal's own read-only tick-history function (its MCP server, `get_chart_ticks_history`: time, bid, ask) by `sa_ticks_export.js`, which writes the same one-record-per-second file `ExportTicks1s.mq5` would have written; the owner opened MetaTrader 5 for it. Server time is New York + 7 hours (the daily break falls at 00:00 to 01:00 server time).
- ~~The server's tick history for XAUUSD starts on 5 February 2026, so Part 2 covers 5 February to 16 October 2026.~~ **Corrected the same day, after the export and before any simulation:** the server says its tick history starts on 5 February 2026, but it returns no ticks before **24 March 2026** (none in the 09:00-21:00 server window on any weekday from 5 February to 23 March; whole days asked again, still none). Part 2 therefore covers **24 March to 16 October 2026: 148 trading days** (Good Friday, 3 April, has no ticks) instead of a full year. Part 2's results are for that period; nothing is extrapolated to a year. Part 1 keeps its 250-day year.
- **The data check** (`sa_data_check.js`, run before any simulation; output in `data_ticks/xauusd_quality.txt`): sha256 `1f5e4e7edba4cc269c04a6239024d4e41f96354d5dc7006edd6925d29bbae818` (matches the export log); 5,913,574 seconds with ticks, 23,021,456 ticks; prices with 2 decimals; no record out of order or inconsistent (the ask never below the bid). Inside the London window, 75% to 99% of the seconds have ticks (median 93%) and the longest stretch without a tick is 25 seconds. The spread at the first quote of a second averages $0.22 per 0.01 lot (median $0.19, 95th percentile $0.37; $0.32 on average in April, $0.17 in September and October). 23 one-second moves of the mid of $5 or more (the largest $31.93 on 11 September at 08:30 New York time, the usual time of US data releases). No day is left out for quality: all 148 days are used.

## Checks before any result is shown
1. **Part 1, path A without the minimum:** the mean final balance equals $100 x (1 + 0.01 x the mean result per trade in R)^750 within the simulation error.
2. **The challenge loop is `pc_core.js` unchanged:** with the published settings of the earlier study it reproduces its 0.5%-risk pass rates with the same seeds.
3. **Payouts, resets, breaches and the end of the year by hand** on made-up sequences; **the scalper's fills, the +$5 exit and the 40-second exit by hand** on made-up seconds.
4. **An independent re-implementation** of both parts, compared on every cell.

## Before the article
The numbers go to the owner first; one article, published only with the owner's agreement. No firm is named.

## Reading the result
Simulations: real trading has varying size, streaks, news, psychology, execution, slippage beyond the quoted spread and each firm's fine print. The study compares two ways to use a small amount of money; it is not advice and does not rate any firm.