You cannot backtest this EA. Not because I am hiding anything: because it makes decisions by sending live market context to external AI models, and the strategy tester cannot replay an API call that has not happened yet. Historical simulation is simply not available for this category of product.
Which leaves an awkward question if you are thinking about buying one. How exactly are you supposed to evaluate it?
The industry’s answer is “look at my track record”. Mine is public and I keep it that way. But I want to be honest about what someone else’s account can and cannot tell you, and then point you at a better way to find out: an AI trading bot free trial that runs on your own chart, on your own demo account, for seven days. It costs you nothing but ten minutes of setup.
Quick Navigation
- What my account tells you, and what it does not
- What this AI trading bot free trial actually gives you
- How to start it
- How to spend the seven days
- Why one week cannot tell you whether it works
- Who should try it, who should buy, who should do neither
What my account actually tells you, and what it does not
A public forward-tested account is the minimum standard for trust. It shows the system exists, that it runs live, and how it has behaved across a real sample including the bad stretches. That is worth a lot, and a vendor who cannot produce it should not get your money.
Here is what it does not tell you, and this is the part usually left unsaid:
- It is my broker, not yours. Spreads, execution quality and slippage differ, and on a system whose average winner is not enormous, execution costs are not a rounding error.
- It is my preset and my risk settings. Change either and you have a different system with different behaviour.
- It is my AI provider configuration. Different providers, different costs, different analysis.
- It is my account size. The API costs that are background noise on one balance are a structural drag on a smaller one.
So the honest position is that my track record earns the EA a place on your shortlist. It cannot tell you how the thing will behave in your hands, with your settings and your AI key. The only way to find that out is to run it.
What this AI trading bot free trial actually gives you
The free demo on the MQL5 Market can’t help here: it only runs in the strategy tester, the same place where the AI can’t be called. So I built a working one. You get the full Alpha Pulse AI for seven days, on your own demo account, with the AI model you already have a key for.
- The seven days start the first time you load it on a chart, not when you sign up, so setting up your API key does not eat into them.
- Demo accounts only. The point is to see it work, not to trade real money for free.
- One engine, Auction & Liquidity, on the chart’s symbol, with any model from the list. That keeps seven days comparable for everyone.
- No card, and one trial per trader and per machine.
At the end of it you know things that no amount of reading can give you. Whether the EA runs cleanly on your setup. Whether your VPS holds. What your actual API bill looks like at your chosen model, which is the single most underestimated number in AI trading. What the AI decides on your market and the reasons it gives for each decision. And most importantly, whether you can personally leave it alone, which is the variable that decides more outcomes than any setting.
That is a genuinely different kind of knowledge from reading a stats page, and this time it costs you nothing to acquire.
How to start it
- Request it on the free trial page with your email.
- Open the sign-in link you receive and download the MT5 or MT4 file from your account.
- In MetaTrader, allow WebRequest for
https://doittrading.comand for your AI provider. The setup guide has the full list with a copy button. - Load it on a demo chart, put your email in the license field, pick your model and paste its API key.
A countdown appears in the top-right corner of the chart. The AI bill is yours and goes straight to your provider: usually around $50 a month in normal use, with a daily cost limit you set in the inputs.
How to spend the seven days so they actually answer something
A trial with no plan produces the same nothing as an unopened download. Four parts:
Day one: verify the plumbing, not the profits
You are not testing whether it makes money, you are testing whether it works: does the API connection hold, does the VPS stay up, do the logs make sense, does the panel show the decisions. A boring first day of verified operation is exactly the outcome you want.
The rest of the week: one model, unchanged
Pick one model and leave it alone. Do not switch models every few hours, do not adjust risk mid-stream. If you change three things you will have three explanations for whatever happens and therefore none. The discipline sounds obvious and almost nobody manages it, which is precisely why most people learn nothing from their trials.
Throughout: measure your costs, not just your trades
Log the API spend per day. Multiply it out to a month and divide it by the account balance you would actually trade. That percentage is a permanent headwind on your returns, and it is the number that decides whether this product makes sense at your account size at all. Nobody else can tell you this figure, because it depends on your provider and your model.
At the end: judge fit, not edge
Open the panel and read three of its decisions. Would you have taken them? This is the important question, and it is where I have to argue against my own interest.
Why one week cannot tell you whether it works
The live account’s results are demonstrably not random in sequence: wins cluster near wins and losses cluster near losses. That is a structural property of the system, and it has a direct consequence for anyone running a trial.
A single week can easily land inside a good cluster or a bad one, and neither is representative. A great trial week does not prove the system works, and a poor one does not prove it is broken. Demo fills are also usually kinder than live ones. So evaluate the things a week can answer, which are stability, cost, the quality of the reasoning and your own temperament. Leave the question of edge to the long public sample, which is exactly what it is there for.
Any vendor telling you a week is enough to validate a strategy is telling you something they know is not true.
Who should try it, who should buy, and who should do neither
- Try it if you have never run an AI EA, you are unsure about your VPS or your provider, or you want your real API cost figure before committing.
- Buy if you have already watched it long enough on the public account and decided this is your setup for the coming year.
- Do neither if your account is small enough that a monthly API bill is a material percentage of it. That is a maths problem and no trial fixes it. Grow the account first, and I would rather say that plainly than sell you something that cannot work at your size.
The practical details
The product page is DoIt Alpha Pulse AI, for MT5 and MT4: one purchase, no subscription, and the current price is always there rather than in whatever an old post says.
You will need an API key from your chosen AI provider, a MetaTrader demo account for the trial, and realistically a VPS for continuous operation. The public forward-tested account is linked from the product page. Go and read the bad months on it, not the good ones: the good months are the ones that tell you nothing.
The close
Selling licenses is better for me than handing out free weeks. I am pointing at the trial anyway, because the alternative is people spending several hundred dollars on a product they have never seen run on their own chart, discovering the fit is wrong, and both of us ending up in a refund conversation neither of us wanted.
Spend ten minutes setting it up. Generate your own data. Then decide with it.
Trading involves substantial risk of loss. Past performance does not guarantee future results. The trial runs on demo accounts only. API costs are paid to your AI provider and are controlled by your configuration.
