MT4 backtest different in 1 month and 1 year | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

MT4 synch objects to different charts (same pair, different tf) 11 replies

MT4 - Why lines appear at different prices in different TFs? 0 replies

Visual backtest result different from non-visual backtest 0 replies

Backtest different from Live test 2 replies

Can Visual Backtest mode render closing value of a different pair? 0 replies

MT4 backtest different in 1 month and 1 year

  • #1
  • First Post: Jul 5, 2020 11:54am Jul 5, 2020 11:54am
Hello.

I'm doing some backtesting with an EA (SMA cross, pretty simple), and it gives me different results if I change dates from 1 year to last month, i.e. I check my results tab and last month I get 10 trades with only one loss, but yearly last month I got 4 losses out of 10 and this gives me a negative result, just analyzing this particularly example. No trades are carried for more than 24h, so at the end of the month or end of the day I should be getting the same trades and same results, right? What am I missing here?

XM is my broker, I've downloaded historical data, I get 90% of modeling quality.

Cheers.
  • #2
  • Jul 5, 2020 12:04pm Jul 5, 2020 12:04pm
Quoting Art5
Hello. I'm doing some backtesting with an EA (SMA cross, pretty simple), and it gives me different results if I change dates from 1 year to last month, i.e. I check my results tab and last month I get 10 trades with only one loss, but yearly last month I got 4 losses out of 10 and this gives me a negative result, just analyzing this particularly example. No trades are carried for more than 24h, so at the end of the month or end of the day I should be getting the same trades and same results, right? What am I missing here? XM is my broker, I've downloaded...
If an EA uses an indicator that loops back n bars, it should start trading only after n bars.
No guts, no glory
  • #3
  • Jul 5, 2020 12:15pm Jul 5, 2020 12:15pm
Quoting abokwaik
{quote} If an EA uses an indicator that loops back n bars, it should start trading only after n bars.
Only indicator it uses is a SMA, doesn't use that concept.

As I said, if I backtest June/2019 to now, it should give the same results when June/20 came then when I'm backtesting only June/20. At least this is what I think.
  • #4
  • Edited 12:58pm Jul 5, 2020 12:47pm | Edited 12:58pm
Quoting Art5
Hello. I'm doing some backtesting with an EA (SMA cross, pretty simple), and it gives me different results if I change dates from 1 year to last month, i.e. I check my results tab and last month I get 10 trades with only one loss, but yearly last month I got 4 losses out of 10 and this gives me a negative result, just analyzing this particularly example. No trades are carried for more than 24h, so at the end of the month or end of the day I should be getting the same trades and same results, right? What am I missing here? XM is my broker, I've downloaded...
SMA cross might have happened before the month you are testing. In this case if you back test the whole year the cross will be taken by the EA (and might be a good trade), but when you change to one month, that cross will not happen...
Trading is the hardest way to make easy money...
  • #5
  • Jul 5, 2020 1:44pm Jul 5, 2020 1:44pm
Quoting Art5
{quote} Only indicator it uses is a SMA, doesn't use that concept. As I said, if I backtest June/2019 to now, it should give the same results when June/20 came then when I'm backtesting only June/20. At least this is what I think.
SMA(50) needs 50 bars at least to give correct reading.
No guts, no glory
  • #6
  • Jul 5, 2020 2:03pm Jul 5, 2020 2:03pm
Quoting MathTrader7
{quote} SMA cross might have happened before the month you are testing. In this case if you back test the whole year the cross will be taken by the EA (and might be a good trade), but when you change to one month, that cross will not happen...
I think I wasn’t clear Enough. The bt was made on a 15min tf, but in a span of 1 month and in 1 year period.
  • #7
  • Jul 5, 2020 2:05pm Jul 5, 2020 2:05pm
Quoting abokwaik
{quote} SMA(50) needs 50 bars at least to give correct reading.
Yeah! But in a 15min tf taking a month or a year period of bt should be the same, as a 15 min bt will use 20 past bars to create the sma, right?
  • #8
  • Last Post: Jul 5, 2020 3:19pm Jul 5, 2020 3:19pm
If I understand correctly you are getting different results over a year to over a month back testing. I find this happens all the time. The market dynamics change continuously and it is not that simple to think an EA will work in the future when it has been tested in the past with the same settings. Also you may have the problem of curve fitting.
0 traders viewing now