What do you check before buying an EA? These are the 7 things I look at
After a lot of bought EAs, prop challenges and backtests that looked great, I ended up with a short checklist I now go through before I pay for anything. Backtests weren't the problem. They just never told the whole story.
Here's my list. Curious what you'd add or disagree with.
1. Deposits dressed up as growth
A growing balance isn't the same as profit. I check total deposits vs. current balance, and when money was added. Top-ups after a losing stretch can make a struggling strategy look like it's compounding.
2. Equity vs. balance
Balance only counts closed trades. If the equity drawdown is much deeper than the balance drawdown, the EA is sitting on losing positions and waiting for them to come back, the typical grid/martingale pattern. This one cost me a challenge: the balance looked fine while the equity was deep in open losses, and a single bad day broke the daily loss limit. If a signal doesn't show equity at all, that's a warning on its own.
3. Backtest far better than live
Some gap is normal (spreads, slippage, broker conditions). But a much deeper live drawdown or a much weaker live profit factor usually means the backtest was fitted to the past. I put PF, max DD and monthly return side by side.
4. Numbers that are too perfect
A near-straight equity curve or a profit factor that seems unreal rarely means a perfect edge. More often it means heavy optimization. Does the curve look like trading or like a ruler?
5. Track record doesn't match the product
Signal younger than the product, changed start date, replaced signal, a carefully chosen time window on the sales page. I compare the release date with the signal start and look for older signals from the same developer that aren't linked anymore.
6. Statistics that don't add up, or are missing
Winning + losing trades should be 100 %, long + short too. If basics don't match or are hidden, I ask why.
7. Too short to mean anything
With less than ~3 months or ~200 trades it's very hard to tell skill from luck. I look at trade count and time span before I look at the return.
None of these proves an EA is bad on its own. Two or three together make me ask a lot more questions.
(Just my own checklist, not advice.)
What's on your list? Anything you check that I'm missing?
- Testing the model on new data
- Determining Expert Advisor parameters
- Introduction to MetaTrader 5 Strategy Tester
This check lisk is good in case someone is going to buy the product from anyone unknown - just about this "someone"/buyer is having a lot money or does not care about the seller he is investing to for example.
My personal experience is below:
My personal experience is below:
Forum on trading, automated trading systems and testing trading strategies
...
Sergey Golubev, 2021.01.14 09:18
- If you bought the product from your friend or from your relative or from very active users on this forum so you can fix anything related with the sellers who are responsible about what they are selling (and because you both know each other for the long time).
- But if you bought something from "anyone" just because you have a lot of money so - sorry:
it is your risk, and this risk can not be turned to anyone else.
----------------
I am posting this suggestion all the time on the forum. It is related to the Signals, the Market, and to the Freelance as well -
Forum on trading, automated trading systems and testing trading strategies
...
Sergey Golubev, 2020.01.16 06:35
There is one rule which some people follow:
subscribe to the signal or create an order in Freelance or purchase the product from the user/seller/coder/provider you know in person or you know by activity on the forum.
Because if you purchased something so I may ask:
"Do you know the seller in person, or you are familiar with him on this forum?"
If you say: "Not" so it will be very strange case for me ...
----------------
There are a lot of professional and honest users/coders on the forum, and some of them are very active on this forum helping to the other people and replying on the forum posts/requests.
Always choose the person you know in this forum (who is very active here), and you will have less problems with anything.
Just my opinion sorry.
it is your risk, and this risk can not be turned to anyone else.
Sergey Golubev #:
Because the buying the product is about the following: do you trust this seller or not. And if you are buying from very well-known user of this forum so there is one main quesyion to be asked: "when I can lose money with your EA and why?"
Because the buying the product is about the following: do you trust this seller or not. And if you are buying from very well-known user of this forum so there is one main quesyion to be asked: "when I can lose money with your EA and why?"
Thanks Sergey,
that's a good first filter, and probably the strongest one: someone who has been helping people here for years has a reputation to lose.
that's a good first filter, and probably the strongest one: someone who has been helping people here for years has a reputation to lose.
And your question is a great one to add to the list: "When can I lose money with your EA, and why?"
A seller who can answer that clearly, with the market conditions and the drawdown to expect, has usually thought about the risk
One who can't, or only talks about the upside, tells you something too.
A seller who can answer that clearly, with the market conditions and the drawdown to expect, has usually thought about the risk
One who can't, or only talks about the upside, tells you something too.
The checklist is more for the cases where you don't know the seller yet, or where you want to check that the answer matches the track record.
I would disagree with point 7, or rather with how low the bar sits. I back-tested my own six-setup index book over 11.4 years, 66,511 trades. Over the whole period it made +5,243 R, yet 2015 to 2017 together came out at +15 R, essentially zero, and the entire maximum drawdown of 449 R sits inside those three years. From 2018 on the same rules made about 621 R a year. So three years and thousands of trades would not have told that book apart from a dead one, and 200 trades or three months can land entirely in either regime. Only 46 percent of its days were green, which is also why a short signal history says so little.
On point 3 I would add a question for the seller. The gap between a simple bar-based backtest of my rules and the same rules run as real EA code on real ticks was a factor of three to five per trade, so I would want to know which of the two the screenshot shows.
Timon-pascal Krueger #:
I would disagree with point 7, or rather with how low the bar sits. I back-tested my own six-setup index book over 11.4 years, 66,511 trades. Over the whole period it made +5,243 R, yet 2015 to 2017 together came out at +15 R, essentially zero, and the entire maximum drawdown of 449 R sits inside those three years. From 2018 on the same rules made about 621 R a year. So three years and thousands of trades would not have told that book apart from a dead one, and 200 trades or three months can land entirely in either regime. Only 46 percent of its days were green, which is also why a short signal history says so little.
On point 3 I would add a question for the seller. The gap between a simple bar-based backtest of my rules and the same rules run as real EA code on real ticks was a factor of three to five per trade, so I would want to know which of the two the screenshot shows.
Fair point @Timon-pascal Krueger and a great example.
I'd put it this way: ~200 trades / ~3 months is the floor below which I don't even start looking, not a level where I'd start trusting anything.
I'd put it this way: ~200 trades / ~3 months is the floor below which I don't even start looking, not a level where I'd start trusting anything.
But I don't think the answer can be "wait three years" either. If you run EAs as a business, you can't wait that long, and by then the market or the EA itself has often changed. So for me the question isn't how long the live record is, but whether it fits what the backtest says to expect.
A long backtest across different market phases, on real ticks, tells you how deep the drawdowns and how long the flat stretches usually get. A short live record can't prove the edge, but it can show whether live is behaving within that range or already outside it.
A long backtest across different market phases, on real ticks, tells you how deep the drawdowns and how long the flat stretches usually get. A short live record can't prove the edge, but it can show whether live is behaving within that range or already outside it.
Then: start small, add size only while live matches the expectation, and set the stop rule before you start, e.g. live drawdown well beyond the backtest's worst, or a flat period longer than any in the backtest. That will sometimes cut a good EA too early. Your 2015–2017 stretch is exactly that case, and I'd accept it as the cheaper mistake.
And your backtest question is a really good addition to point 3.
In the MT5 tester that's the modelling mode: "Every tick based on real ticks" vs. "1 minute OHLC" or "Open prices only". If a seller doesn't say which one the screenshot uses, that's worth asking.
In the MT5 tester that's the modelling mode: "Every tick based on real ticks" vs. "1 minute OHLC" or "Open prices only". If a seller doesn't say which one the screenshot uses, that's worth asking.
Thank you.
Good list. Two things I would add, both from testing EAs myself:
8. Run the demo on dates the seller did not pick
The sales page always shows the best window. I download the demo and test it on a period the seller never mentions, ideally a quiet month and one with a big news shock (rate decisions, NFP weeks). If it only works on the dates in the screenshots, that tells you a lot. I also run it with the real spread of my broker, not a fixed low spread.
9. How it sizes the lot, and the minimum it needs
Check whether the lot is fixed, a % of balance, or increases after losses. Then check what happens on your own account size: many results are shown on 10k+ accounts, and on a small account the EA either cannot go below 0.01 lot (so the real risk per trade is much higher than advertised) or it skips trades. I work out the money at risk per trade at the minimum lot before I look at the return.
To your point 2, I would also look at the worst floating drawdown on a single day, not just the max drawdown overall, because that is what breaks prop firm daily limits.
Amul Ravikumar #:
Good list. Two things I would add, both from testing EAs myself:
8. Run the demo on dates the seller did not pick
The sales page always shows the best window. I download the demo and test it on a period the seller never mentions, ideally a quiet month and one with a big news shock (rate decisions, NFP weeks). If it only works on the dates in the screenshots, that tells you a lot. I also run it with the real spread of my broker, not a fixed low spread.
9. How it sizes the lot, and the minimum it needs
Check whether the lot is fixed, a % of balance, or increases after losses. Then check what happens on your own account size: many results are shown on 10k+ accounts, and on a small account the EA either cannot go below 0.01 lot (so the real risk per trade is much higher than advertised) or it skips trades. I work out the money at risk per trade at the minimum lot before I look at the return.
To your point 2, I would also look at the worst floating drawdown on a single day, not just the max drawdown overall, because that is what breaks prop firm daily limits.
Thanks Amul,
both are good additions.
Point 9 is easy to miss. On a small account the minimum lot can quietly turn a "1 % risk" EA into several percent per trade, and the sales page won't tell you.
Working out the money at risk at 0.01 before looking at the return is a simple rule, I like that.
It also works the other way round when you read a signal. On a very small account, $50 a month looks like 20 %. But if the EA trades the same fixed lot on 10k, the same $50 is 0.5 %.
So I look at the lot size and the profit in money, not only the growth in percent, and I check whether the drawdown in percent is just as large.
And yes on the worst single day. That's exactly what got me in the challenge I mentioned under point 2: the overall drawdown looked fine, the worst day didn't.
Testing on dates the seller didn't pick is a good habit too, especially with your own broker's spread instead of a fixed one.
Best regards,
Chris
Communicating with the developer and understanding the mentality of the developer can be a super plus point
Andrew Pun Magar #:
Communicating with the developer and understanding the mentality of the developer can be a super plus point
Communicating with the developer and understanding the mentality of the developer can be a super plus point
Hi Andrew,
agreed, that's a big one. It goes well with Sergey's question above: "When can I lose money with your EA, and why?" How a developer answers that, and how fast, tells you a lot about the mentality. Someone who talks openly about the bad phases has usually thought about the risk.
Best regards,
Chris :)
This is the checklist I wish every buyer used. It's also the standard I try to build to as a developer.
Two additions:
8. Does the logic match the risk? Ask the developer how the EA exits losing trades. If the answer is "it waits," that's your grid/martingale warning from point 2. A hard stop loss on every trade means the equity curve can't hide anything.
9. Is the settings page honest? Check whether the developer lists the recommended broker, spread limit, and risk per trade, and whether results were tested at those exact settings. Backtests on the default setfile with ideal spreads tell you little.
On point 3, I'd also look at the same period on a different broker's data. A real edge survives that. A fitted one often doesn't.
I build XAUUSD EAs, so these are things I check on my own products first: fixed risk per trade, daily loss limit, spread and news filters, and live results on Myfxbook. If I can't show it, I don't claim it.
Which of the 7 do you find sellers fail most often? My guess is #5.
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register