MQL5 (+4d) -> ES
Fuente MQL5: CONTENT/tools/mql5-promo/2026-08-27-48-hours-backtesting-simulator-plan.html
[CONFIRMAR] que el trial 48h sigue siendo start-on-first-use con ventana de canje previa
(es lo que hace que el consejo de “prepara antes de arrancar el reloj” sea valido).
[CONFIRMAR] precio $149 y que la seccion dlm_trial sigue en la product page.
[CONFIRMAR] URL del post companion sobre el protocolo completo de backtesting: cuando se
publique, sustituir la referencia en texto plano de la seccion “What to walk out with”
por un link real.
============================================================ –>
A 48-hour trial is a strange thing to be handed. It is long enough to validate a strategy properly and short enough that if you open it on a Wednesday evening with no plan, you will spend most of it clicking around, discover the speed controls at hour forty, and end the trial knowing roughly nothing.
I have watched enough trials expire unused to know the failure is almost never about the software. It is about walking in without a question. So this post is the plan I would follow: what to do before the clock starts, what to test in each block, what numbers to walk out with, and the specific ways people waste the window.
Quick Navigation
- Why the clock only starts when you start it
- Block one: learn the controls first
- Block two: the main test
- Block three: the window most people skip
- The four numbers to walk out with
- The four ways people waste the window
- The honest limits of the simulator
First: the clock starts when you start it
The single most valuable thing to understand about this trial is that the 48 hours begin on first use, not when you request it. That means the entire preparation phase is free, and preparation is what determines whether the window produces a decision or a shrug.
So request it now and open it later. Before you start, have three things ready:
- A written strategy. One page. What qualifies as a setup, what triggers entry, where the stop goes, how the target is decided, what you risk per trade. If you cannot write it down, you do not have a strategy yet, and no simulator can test a mood.
- A chosen instrument and a data window you do not remember. Not the last six months of the pair you have been watching live: your memory of what happened next contaminates every decision. Go further back, or pick something you follow less.
- A blank log. A spreadsheet with a row per trade. Date, direction, entry, stop, target, result in R, and one column for “did I follow the rules”. That last column is the one that will teach you the most.
Ten minutes of this before you press start is worth more than ten hours of unstructured clicking after.
Block one, about two hours: learn the controls on a throwaway window
Do not start your real test yet. Spend the first block on data you do not care about, deliberately, so that the mechanics are automatic before they matter.
Three things to get comfortable with:
- Speed control. This is the feature that makes the whole thing possible. You slow down to study price around your levels and fast forward through the dead hours. Find the two speeds you personally want: the “nothing is happening” speed and the “this matters” speed.
- Time jumps. Jumping straight to the next session instead of watching the market sleep. Once this becomes reflex, an evening covers weeks.
- Order placement and management. Place a few simulated trades, set stops and targets, let some hit. You want the mechanics to be muscle memory, because in the real test every second spent fighting the interface is a second not spent reading price.
Two hours feels like a lot to spend on tooling. It is the difference between a test and a fumble.
Block two, the main event: one instrument, one timeframe, thirty trades
Now the actual test. Replay your chosen window and take every setup that meets your written rules, and no setup that does not. Log each one.
Three disciplines make this a real test rather than theatre:
- No peeking. Decide, then advance. If you catch yourself scrolling forward to check before committing, that trade is contaminated and the honest move is to discard the window and restart.
- Log the skips. When something tempting appears that does not meet the rules, note it anyway. The near-misses teach you whether your rules are too tight, and they are the only honest way to know.
- Do not stop on a losing streak. Five losses in a row makes you want to pause and “fix” the strategy. Finish the window. The streak is the data, and specifically it is the data that tells you what you will have to sit through live.
Thirty trades is the minimum before anything you compute means much. If your window does not produce thirty, extend the window rather than lowering the bar.
Block three: the second window, and this is the one people skip
If the first window came out positive, do not stop. Run a second window in a different market regime. If the first was trending, pick a range. If the first was calm, pick something volatile.
This is where most strategies die, and it is much better for them to die here than with your money. A strategy that works in one regime and collapses in another is not a strategy, it is a bet on conditions continuing, and you should at least know that is the bet you are making.
What to walk out with
At the end of the 48 hours you should be able to state four numbers from your log:
- Win rate, which on its own means nothing, which is why it comes paired with the next one.
- Average reward-to-risk, winners divided by losers in R terms.
- Expectancy, the average result of a trade from this system. This is the only number that answers the actual question.
- Longest losing streak, which is your psychological stress test and the input that decides your risk per trade. Whatever the sample produced, live trading will eventually produce worse.
I wrote up the full protocol behind these numbers, including how to compute expectancy and the decision gates for what to do with the result, in a separate article. If you want the method in depth, read that alongside this one.
If you finish the trial with those four numbers written down, the trial did its job whether or not you buy anything, because you now know something specific about your own strategy that you did not know two days ago.
The four ways people waste the window
- Testing five strategies instead of one. Forty-eight hours divided by five is not a test of anything. One strategy, properly, beats five superficially, every time.
- Optimising mid-test. You notice a filter that would have removed three losers, and you start applying it. Now half your sample used one rule set and half used another, and the blend describes a strategy that does not exist. Note the idea, finish as written, test the idea separately.
- Timeframe drift. H1 does not confirm your bias so you check M15, then M5, until something agrees. Pick the execution timeframe in the written rules and stay on it.
- Treating it as entertainment. Replaying markets is genuinely fun, which is the trap. Fun without a log is just a video game with candlesticks.
What you are actually working with
The Backtesting Simulator (MT5 and MT4) replays your broker’s own historical data inside MetaTrader, with no extra software to install. You get speed control from tick-by-tick study to fast forward, time jumps to skip to what matters, realistic order simulation so your log reflects execution rather than intention, and multi-chart mode when you want to test across several instruments or timeframes in the same session.
One caveat repeated from the product page because I would rather you hear it from me: the interface can lag visually inside the strategy tester on some machines. It is cosmetic and does not affect the simulation, but it surprises people who were not told.
It is $149 to own, and the product page has a free demo as well if you want to look before you even start the trial. The free 48-hour trial is claimed here and takes about a minute. Remember the clock only starts on first use, so claim it now and open it when you have the strategy, the window and the log ready.
The honest limits
- A backtest proves the past. What replay testing buys you is cheap failure: a strategy that was never going to work gets exposed in two days instead of four months.
- You are the variable. Manual replay tests your execution of the rules as much as the rules themselves. That is a feature, since you are the variable in live trading too, but it means two people testing the same strategy can get different results, and both can be right about themselves.
- Replay data is your broker’s data. Which is the point, because it is what you will trade against, but it also means a different broker’s spreads could shift a marginal strategy across the line.
- It will not tell you what to trade. It tells you whether the thing you already decided to trade holds up. Bring a strategy.
The close
Forty-eight hours sounds short until you realise that with speed control and time jumps it covers months of market. The constraint was never the time. It was walking in without a written strategy, a clean data window and a log, and hoping the software would supply the discipline.
Prepare first, start the clock second. Then the trial is not a countdown, it is a deadline, and deadlines are the only reason most strategies ever get tested at all.
Trading involves substantial risk of loss. Past performance does not guarantee future results.
