A trading result is only as believable as the number of trades behind it. Ten trades can show almost any win rate by luck. A few hundred trades, recorded by an independent platform over many months, begin to show what a strategy really does. That is why Sonic AI reports its gold (XAUUSD) results as a long, independently verified record: 18 consecutive winning months and an 80% win rate, with live performance you can read for yourself.
This guide explains sample size in plain terms, shows the numbers, and gives you a short method for judging any track record, including audited trading results.
TL;DR
- Sample size is the number of trades behind a result. A small sample says very little, even when every trade wins.
- A win rate is a range, not a single number. The more trades, the narrower the range.
- Count trades and months, not screenshots. A record is stronger when a third party tracks it and it covers different market conditions.
- Win rate alone is incomplete. Pair it with drawdown and the size of average wins and losses.
- Sonic AI publishes a verified record on MyFXBook: 18 consecutive winning months, an 80% win rate and drawdown below 1%, as we state it.
Table of Contents
- What sample size means in trading
- Why a win rate is a range
- How many trades is enough?
- Win rate is not the whole result
- Audited trading results: what makes a record count
- Reading a gold track record
- Common sample size mistakes
- Our perspective
- See the record and start copying
- FAQ
- Sources
What sample size means in trading
Sample size is the count of independent observations in a result. In trading, that is usually the number of closed trades. If a strategy shows 8 wins in 10 trades, the sample size is 10.
The same logic applies to time. Twenty trades taken in one calm week come from a single market mood. Twenty trades spread across trending, quiet and news-driven months say more. So a good sample has two parts: enough trades, and enough different conditions.
Why does this matter? A coin flipped ten times can land heads eight times. Nobody calls that coin special. A trader who wins eight of ten trades has shown the same thing: a result that luck alone can produce.
Why a win rate is a range
When someone reports an 80% win rate, they report what happened in one sample. The true long-run rate of the strategy is unknown. Statisticians describe it with a confidence interval: a range that very likely contains the true rate.
The table below uses the Wilson score interval, a standard method that behaves well for small samples. It shows the 95% range for a strategy that wins 80% of the time in the sample, at different trade counts. The figures are our own calculation from the standard formula.
| Trades in sample | Wins at 80% | Likely true win rate (95% range) |
|---|---|---|
| 10 | 8 | 49% to 94% |
| 30 | 24 | 63% to 91% |
| 100 | 80 | 71% to 87% |
| 200 | 160 | 74% to 85% |
| 500 | 400 | 76% to 83% |
Read it row by row. With 10 trades, an 80% result is compatible with a coin-flip strategy. With 100 trades, the low end is above 70%. With 500, the range is tight enough to plan around.
The range shrinks with the square root of the trade count. Going from 100 to 400 trades roughly halves the uncertainty. That is why the first few dozen trades add a lot of information and each later batch adds less.
A worked example
A trader posts 30 wins in 50 trades. That is 60%, which sounds like an edge. The 95% range is about 46% to 72%. The true rate could sit below 50%, where the strategy loses more often than it wins. The result is not wrong. It simply cannot yet separate skill from luck.

How many trades is enough?
There is no single number that fits every strategy. Common rules of thumb help as a starting point:
- Under 30 trades: treat the result as a first look. It cannot support a decision.
- 30 to 100 trades: a rough signal. Useful for spotting a strategy that clearly fails, not for proving one that works.
- 100 to 200 trades: a workable base for a win rate and an average win and loss. Our guide on verifying gold trading performance with 200 trades walks through this level.
- 200 or more trades across several months: the range where a win rate starts to carry real weight, particularly when the trades cover rising, falling and sideways markets.
The number you need also depends on what you measure. A win rate needs fewer trades to estimate than a drawdown. Drawdown is the largest fall from a peak in account value. It depends on the worst stretch in the record, and a short sample may never have met a bad stretch. So a long record matters most when you want to know how deep a strategy can fall.
If you want to see how many trades you need to prove expectancy, see how to prove expectancy with R and 50 to 100 trades.
Win rate is not the whole result
Sample size tells you how much to trust a number. It does not tell you whether the number is the right one. A high win rate can hide small wins and rare, large losses. A low win rate can be very profitable when winners are large.
The metric that joins these is expectancy: the average amount you expect to make per trade, after wins and losses. A simple form is:
Expectancy = (win rate × average win) − (loss rate × average loss)
Take two strategies, each with 200 trades:
| Strategy A | Strategy B | |
|---|---|---|
| Win rate | 80% | 40% |
| Average win | $10 | $40 |
| Average loss | $45 | $20 |
| Expectancy per trade | −$1.00 | +$4.00 |
Strategy A wins four of five trades and still loses money. Strategy B loses more often than it wins and earns. The example is invented to show the arithmetic. For a longer treatment, read win rate versus expectancy in a trade journal and what an 80% win rate means and how to check it.
This is why drawdown belongs next to win rate. A strategy with a high win rate and a drawdown below 1%, as Sonic AI reports, shows that losing trades stay small relative to the account.
Audited trading results: what makes a record count
A sample can be large and still be unreliable if you cannot confirm that it is real. Regulators warn investors about performance claims for this reason. The U.S. Securities and Exchange Commission has published an investor bulletin on investment performance claims to help people read them with care.
A record carries the most weight when these points hold:
- A third party records the trades. Results come from the connected trading account through an independent tracking platform, not from a spreadsheet the trader edits. Sonic AI's results are verified on MyFXBook.
- Every trade is included. The record counts all closed trades, with no selection of the best ones.
- The period is long. Months of continuous history cover more market conditions than a single strong week. Sonic AI reports 18 consecutive winning months.
- Live accounts back it. A live record reflects real execution, including spreads and fills. A backtest is a simulation. The gap between the two is covered in backtest versus live trading.
- The numbers tie together. Win rate, drawdown, trade count and growth should describe one consistent account.
A step-by-step process for this check is in the 7-step third-party trade verification for XAUUSD copy traders.
Reading a gold track record
Gold behaves differently across conditions. It can move sharply on interest-rate news, trend for weeks, or drift in a range. A record that includes all three tells you more than one that covers a single phase.
When you read any gold strategy's record, work through these questions in order:
- How many closed trades are there? Aim for 100 at the very least, and more is better.
- How many months does the record span? Several months is the minimum for a gold record. A year or more is stronger.
- What is the maximum drawdown? Compare it with the profit. A strategy that earns a lot while falling very little has controlled its risk.
- How is the win rate paired with average win and loss? Check expectancy, not the win rate alone.
- Is the data independent? Look for a public page run by a tracking platform.
- Does the live account match the signal? For copy trading, the results you receive depend on execution. Automatic copying keeps your account close to the master account.
The reasoning behind these questions applies to any strategy. If you want to see how a full gold approach is built, see our best gold trading strategy for automated investors and the practical guide to algorithmic gold trading.
Common sample size mistakes
Stopping at a lucky streak. A run of wins at the start of a record is exactly what chance produces. Wait for the sample to grow.
Mixing strategies. Results from different setups or timeframes belong in separate samples. Pooled together, they hide what each one does.
Changing rules mid-sample. Every change to entries, exits or position size starts a new sample. Trades before and after the change are not the same strategy.
Ignoring costs. Spreads, slippage and fees reduce live results. A record taken from a real account already includes them.
Treating the sample as a promise. Past results describe what happened. They do not guarantee what comes next. Trading gold involves risk of loss, and you should only trade with money you can afford to lose.
Our perspective
We built Sonic AI around one idea: a result means something only when enough trades stand behind it and someone independent records them.
That is why the strategy focuses exclusively on gold. One instrument, one methodology and one account history make the sample consistent. Every trade belongs to the same strategy, so the record is easy to read.
We report an 80% win rate, drawdown below 1% and 18 consecutive winning months, verified on MyFXBook. We put the verification first because a number without a source is only a claim.
For a reader with little trading knowledge, this matters more than any indicator. COPYX copies the trades to your account automatically, so you do not need to read charts or time entries. The sample you review is the sample your account follows.
Performance fees are shared from trading gains, so our interest follows the account's results. Past performance does not guarantee future results, and all trading carries risk.
Paulo
See the record and start copying
Sonic AI is an automated gold (XAUUSD) strategy managed by an experienced team. It uses COPYX to copy trades to your account automatically, so you can follow the strategy without extensive trading knowledge. You can review the live results on the Sonic AI live page, read how profit sharing works on the performance fee page, or begin at Sonic AI.

Network marketers can look at the COPYX IB Partner Program. Investors who want larger exposure can look at the 12X and 24X amplified accounts.
FAQ
How many trades do I need to trust a trading result?
At least 100 closed trades gives a first usable estimate, and 200 or more across several months is stronger. With fewer than 30 trades, even a perfect record can come from luck. The more trades and the more varied the market conditions, the more the result tells you.
Is a high win rate enough to judge a strategy?
No. A win rate shows how often trades win, not how much they earn or lose. Check expectancy and maximum drawdown alongside it. A strategy with an 80% win rate, small losses and a drawdown below 1% gives a more complete picture than the win rate alone.
What are audited trading results?
Audited trading results are records of real account trades kept by an independent party, not by the trader. The platform logs each trade, so the history cannot be edited. Sonic AI's results are verified on MyFXBook, and you can read them on the live results page.
How long should a gold track record be?
Several months at minimum, and a year or more is better. Gold moves through trending, ranging and news-driven phases, and a long record includes them. Sonic AI reports 18 consecutive winning months.
Does a large sample guarantee future profits?
No. A large sample lowers the chance that the result is luck, but markets change. Past performance does not guarantee future results, and you can lose money trading gold. Only use capital you can afford to lose.
