ARMFA history trading simulator for forex
A history trading simulator solves a task otherwise solved with money and time: accumulating statistics on the rules before they start costing a lot. A hundred trades in a real market with intraday trading is months, in a simulator an evening. We take apart three ways to run the rules over past quotes, how they differ and why the result of a run is always better than the real one.
Simulators are a training tool: they take no part in real trading and calculate nothing on your account. Their task is to give a sample of trades by the rules that would otherwise take months to accumulate.
Why run the rules over history
Risk management is built on numbers a new system does not have: the win rate, the average ratio, the length of a typical losing streak. Until they exist the risk percentage is chosen at random, and the very first drawdown is taken as a breakdown, although it may be the norm.
| Question | The answer the run gives | What to do with it |
|---|---|---|
| What the win rate of the rules is | the share of profitable trades in the sample | compare it with the break-even win rate |
| What the average ratio is | the average profit against the average loss | check how realistic the targets are |
| What a typical losing streak is | the longest chain of stops in a row | choose the daily limit and the risk share |
| How many trades a month | how often the rules trigger | estimate how long it takes to accumulate statistics |
| How the rules behave in a range | the result over stretches without a trend | decide whether a market regime filter is needed |
| How much the costs eat | the difference in the result with and without the spread | recalculate the expectancy under real conditions |
None of the answers is a forecast. A run tells you about the past behaviour of the rules on a specific stretch of history — that is enough to filter out obviously losing ideas and not enough to guarantee a result.
Forex Tester 5 and Soft4FX Forex Simulator: how they differ
Two paid tools with the same idea and different implementations. The choice between them is usually decided by whether you are ready to leave your familiar terminal.
Forex tester 5 and its versions are a standalone application with its own data, charts and statistics. The plus is independence from the broker and a detailed report, the minus a second interface that has to be learned.
its own environmentSoft4fx forex simulator embeds into the built-in MetaTrader tester: the same charts, the same indicators, the same markup. There is nothing to learn, everything is familiar.
inside MTThe mt5 strategy tester on its own runs advisors but does not let you trade by hand. As a training tool it works only together with a plugin.
no extra costThe bar playback mode on web platforms: fast and free, but without the spread, the swap or a model of execution.
superficialFor testing risk management rules the difference between the first two is small. The third point matters more: if what you are running is an advisor rather than your own decisions, there is no need to pay at all — the built-in tester copes.
What distorts the result of a run
The result in a simulator is almost always better than the real one, and the reasons are systemic. Knowing them matters more than knowing the functions of a particular program.
The antidote is simple and unpleasant: the run is done once, without restarts, and the result is halved in your head. And it is always checked on another stretch of history that was not used during the tuning.
How to run the rules so that the sample can be trusted
The order below separates a test from entertainment. It also turns the result of a run into numbers usable for calculating the risk share.
The entry condition, the idea invalidation level, the exit rule and the size of the risk. If the rules are formulated as you go, what is tested is not the system but intuition.
before the startA year to eighteen months of quotes with different market regimes. Tuning on one stretch and testing on another is a mandatory separation.
before the startThe spread, the commission and the swap in the run settings. Without them the expectancy is inflated and short targets look workable.
compulsoryWithout restarting and without skipping. It is exactly the continuity that makes the sample resemble real trading.
one runThe win rate, the average ratio and the length of the worst streak go into the calculators to give the risk share and the expectancy over distance.
after the runSimulators: short answers
- How a simulator differs from a strategy tester
- A tester runs the code of an advisor, a simulator lets you trade by hand. For testing your own decisions it is the second one you need.
- Is a demo account enough instead of a simulator
- A demo is more honest about emotions but accumulates trades in real time. Usually both are used: the simulator for the rules, the demo for the habit.
- How many trades should be run
- The reference is a hundred for first conclusions and several hundred for stable ones. Over thirty trades the result is decided by chance.
- Is high-quality tick data needed
- For rules with targets of tens of pips minute quotes are enough. Tick data matters for scalping, where the stop is several times shorter than the spread.
- Can several pairs be run at once
- In separate programs and in a multi-currency tester, yes. That matters if the rules assume simultaneous positions and a total risk.
- What to do with the result of a run
- Carry the win rate and the average ratio into the calculation of expectancy and the risk share. The run exists for those numbers rather than for the final curve.
Frequently asked questions
What a history trading simulator is
A program that plays past quotes back at a chosen speed and lets you open trades as if the trading were happening in real time. The right-hand part of the chart is hidden, so the decisions are taken without knowing the future.
How Forex Tester differs from a plugin in MetaTrader
Forex Tester is a standalone application with its own data and reports, while a plugin works inside the terminal and uses the familiar charts and indicators. Their task is the same, what differs is the environment and the depth of the statistics.
Can you trade by hand in the built-in MT5 tester
On its own the tester runs advisors rather than manual trades. Manual trading on history appears when a simulator plugin is added to the tester.
How close is the result of a run to the real one
It is usually noticeably better. The reasons are perfect execution, the absence of emotions and a subconscious knowledge of how the stretch of history ended. It is sensible to allow for that and not to treat a run as a promise.
Do the spread and the swap have to be included in a run
Definitely. Without costs short targets look profitable, while on a real account a system with a ten-pip target can go into the red on the spread alone.
How do you avoid fitting to history
Split the data: tune the rules on one stretch and test them on another that was not opened during the tuning. If the results diverge strongly, the system describes the past rather than the market.
How long does a run of a hundred trades take
From two to five hours with intraday rules and accelerated playback. That is the main benefit: the same hundred trades in a real market would take months.
Does a simulator replace the risk calculation
No, it provides the input data for the calculation. The win rate and the average ratio from a run are substituted into the formulas for expectancy, the losing streak and the risk share.
Is bar replay on a chart a substitute for a simulator
For practising markup, yes; for assessing statistics, no: it does not account for the spread, the swap or the model of execution, and the trade sizes are not recorded anywhere.
Can the daily loss limit be tested in a simulator
It can and should be: a run will show how often the limit would have fired and how many trades you would have skipped. That is the best way to pick the threshold without finding it out on a live account.
Is paid quote data needed
For most tasks the free history the terminal or the program itself provides is enough. Paid tick archives are needed when the rules are sensitive to every pip.
What to do if a system is profitable in the simulator and not on the account
Compare the execution: the real entry points, the actual stops and the number of skipped trades. The discrepancy is almost always in discipline and in costs rather than in the rules themselves.
Read next
