Strategy Tester questions
I'm having real questions about the legitimacy of the ST. Setting aside that I can get different results simply by hitting the start button more than once, and that trades often are completely missing from the journal for days in the middle of a time frame, how about these two items [....]
You'll doubtless get lots of responses about this; it's a popular topic. A very brief answer to your questions is as follows. Lots of other possibilities which I'll leave it up to other people to add.
Regarding your first question, live and demo accounts can have differences in things like spread and swap rates. Depending on the nature of your strategy these can have anything from a tiny to a massive effect. It is entirely normal, realistic, and absolutely to be expected that you see different results in backtesting from using live versus demo accounts.
Regarding your second question, if the only variable is the lot size, then it is a bit odd that fewer trades are being taken if there is no explanation in the form of errors in the tester's journal (e.g. error #134). But you need to jump through a lot of hoops to make sure that the lot size is the only variable. For example, it could simply be because your lot size increases over the period of the test, and your strategy happens to generate fewer signals as time goes on.
BTW, it's well known that results can vary from one test to another, because the tester uses the most recent live spread for historic modelling. If the live spread changes between tests, and your strategy is spread-sensitive, then you'll get different results from consecutive runs. The degree of difference will depend on how sensitive your strategy is to variations in the spread. Also, the list of trades is written asynchronously; it's not guaranteed to be complete until the end of a test.
read https://www.mql5.com/en/forum/124778 about spread
Lots of other possibilities which I'll leave it up to other people to add. [...]
There is one other thing I'll add, because it isn't often mentioned this explicitly.
If either your live or demo accounts are Alpari UK... The Alpari demo account is the least realistic demo account I can think of. In fact, it's actively and deliberately unrealistic. The Alpari UK demo account has the spreads of their pro account, but without commission and without the various restrictions. In real life, in order to get these spreads from Alpari, you have to deposit at least $20,000; you have to trade at least 3.0 lots per order; and you get charged a commission on each order. See https://alpari.com/en/trading_accounts/account_comparison_table.html
(Digressing slightly, Alpari's popularity means that it's often assumed that all brokers do this, and that all demo accounts have unrealistic spreads etc. This isn't true. A lot of the general remarks made about the realism of demo vs live accounts are ill-informed scare stories.)
EDIT: By the way, this isn't a criticism of Alpari. It's not as though they hide the fact they're doing this. The link quoted above is perhaps the most clear and open account comparison I've seen from any broker.
There is one other thing I'll add, because it isn't often mentioned this explicitly.
If either your live or demo accounts are Alpari UK... The Alpari demo account is the least realistic demo account I can think of. In fact, it's actively and deliberately unrealistic. The Alpari UK demo account has the spreads of their pro account, but without commission and without the various restrictions. In real life, in order to get these spreads from Alpari, you have to deposit at least $20,000; you have to trade at least 3.0 lots per order; and you get charged a commission on each order. See https://www.mql5.com/go?link=https://alpari.com/en/trading_accounts/account_comparison_table.html
(Digressing slightly, Alpari's popularity means that it's often assumed that all brokers do this, and that all demo accounts have unrealistic spreads etc. This isn't true. A lot of the general remarks made about the realism of demo vs live accounts are ill-informed scare stories.)
jjc, could you say which broker have demo server as realistic as possible ?
jjc, could you say which broker have demo server as realistic as possible ?
In other words, the realism of a demo account partly depends on the nature of your strategy. And there are other issues as well. For example, if you are scalping, you will run into all sorts of issues about authenticity which are caused by the MT4 strategy tester and the way it models historic ticks and fills simulated orders, not to do with the differences between live and demo accounts (or even between live and demo prices/spreads).
I'm having real questions about the legitimacy of the ST. [...]
The limitations of the Tester are actually well-documented but unfortunately they are not well-known. I recommend reading and making sure you understand everything in the article: Testing Features and Limits in MetaTrader 4. Other limitations and features are documented in other Tester related articles: https://www.mql5.com/en/articles/mt4/tester.
Related -> https://www.mql5.com/en/forum/126108.
B"H
Hello,
A question regarding the ST:
Shouldn't the price curve be contiguous across weekends?
In other words: Shouldn't the price on market re-opening time on Sunday be exactly equal to that on market closure time on last weekend?
Thanks,
Simha
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
I'm having real questions about the legitimacy of the ST. Setting aside that I can get different results simply by hitting the start button more than once, and that trades often are completely missing from the journal for days in the middle of a time frame, how about these two items ...
First, why do I get very different results on the same pair and time frame simply by having a live rather than a demo account selected? What does the account have to do with backtesting data?
Second, when the lot size gets to about half of the MKTINFO max lot size (I don't know where it happens exactly), I suddenly get about a quarter of the trades that I have at a lesser lot size. There are no errors, no 131's, no not enough money, no errors. Just fewer trades. I'm trading at about a half of the MKTINFO max lot size. No where near zeroing out the account at any point. I don't get it. It just seems to stop trading if I put too much money into it.
I am at the point that I don't believe anything other than trading forward. I feel very little confidence in the tester. But I could be wrong, so I'm asking ...
Can anyone explain either of the above results?
Thanks.