The question a backtest answers
Imagine you decide on a simple rule today: buy five large companies that make real profit, check them once a month, and sell any stock that falls 10% below what you paid. That is a plan. The honest question is whether it would have survived the last five years, and you cannot answer that by thinking about it.
A backtest answers it by replaying the rule day by day across real market prices. It starts on a date years ago with a set amount of pretend cash, walks forward one trading day at a time, and does exactly what your rule says to do on each of those days. At the end you get an account balance, a list of every trade, and the reason each trade happened.
That is the whole idea. You are not predicting the future. You are checking whether your rule has ever worked when markets were real, including the weeks nobody enjoyed.
A worked example you can follow
Say the test starts with $10,000 five years ago. On the first trading day the plan buys five stocks, putting $2,000 into each. Every trade costs a small fee, so the account does not start at exactly $10,000 of stock; the fees come out immediately, the same way they would in a real account.
A month later the plan reviews its five holdings. One of them has dropped 12%, past the 10% line you set, so the plan sells it that day and buys a replacement that passes your rules. That sale is recorded with its reason: the stock fell past the selling line you chose.
Repeat that for sixty months. At the end the test might report that the account finished at $13,850, that its worst stretch took it down $2,100 from a previous high, and that it made 214 trades along the way. Now you know three real things about your rule instead of zero.
Why the fill price matters more than anything
Here is where most free backtesting tools quietly lie. If a test decides to buy a stock using the closing price of the same day it made the decision, it is using information that did not exist when the decision was made. That is not a test, it is hindsight, and it makes every rule look brilliant.
A trade in a Stax backtest fills at the next trading day's opening price. The plan decides after the market closes, the order goes in for the next morning, and it fills at whatever the market opens at. Sometimes that is better than the closing price and sometimes it is worse, exactly as in life.
Whenever you read a backtest result anywhere, this is the first thing worth asking about. A test that fills on the same day it decides is worth nothing.
The costs that quietly eat results
Every trade costs something. A plan that trades twice a year barely notices; a plan that trades every week can hand over a meaningful slice of its gains without you seeing it happen.
A backtest that ignores fees will always flatter a busy plan. In a Stax test the fee comes out of the simulated account on every single trade, which means a rule that trades constantly has to earn enough to cover its own activity before it looks good.
This is why the number of trades is worth reading as carefully as the final balance. Two plans that both finish at $13,850 are not equal if one got there in 40 trades and the other needed 400.
What a backtest genuinely cannot tell you
It cannot tell you what next year holds. Markets change, whole industries fall out of favour, and a rule that suited the last five years may suit the next five badly.
It also cannot protect you from a rule that was invented by staring at the past until something fit. If you try two hundred versions of a plan and keep only the one that performed best, you have not found a good rule. You have found the version that matched history most closely, which is a different and much less useful thing.
What a backtest gives you is evidence: a plan behaved this way across real years including bad ones. Evidence is worth a great deal more than a hunch, and considerably less than a promise. Anyone selling you the second one is not being straight with you.
How to read a result without fooling yourself
Look at the worst stretch before you look at the final number. If the test says the account fell $2,100 at its lowest point, ask yourself honestly whether you would still have been holding on that day. If the answer is no, the plan is wrong for you regardless of how it finished.
Then open the trades. A result you cannot inspect is a number you are being asked to trust. Every trade in a Stax test carries the rule that fired, the price it filled at, and what it did to the account, so you can read the story rather than take the summary on faith.
Eight independent verification checks audit each run before you see it, covering the rules the test followed, the prices it used, the fees it charged, and the limits it obeyed. If something looks wrong, that shows up too.
Running your first one
In Stax you can describe a plan in a sentence, watch the rules come back in plain words, and run the history test in a few seconds. The free tier covers this, and no card is needed to do it.
A good first exercise is deliberately testing a rule you suspect is bad. Watching a plan you do not believe in fail across five years of real prices teaches you more about what a backtest is worth than watching a flattering one succeed.