I'm using renko bars in my strategies, but I have 2 problems with testing EAs by strategy tester.
1. Even if I download good data where's no discrepancy between any timeframes and I have 90% modelling quality when I convert it to renko there're a lot of mismatched charts errors. Can I do something with that?
2. Strategy tester produce incorrect results when for example first bar is bearish and second is bullish. Conditions are met in latest bar, so EA opens short position, but it takes opening price from next bar, so in this case it takes better, higher price. Does anybody know script that covert standard charts to charts similar to renko, but when there's trend reverse it starts new bar from price that last bar was closed?
I'm definitely talking about renko, not constatnt range bars. I enclose two screenshots from MT4, test are made on the same period, 15min chart- no errors, offline renko chart with wicks- 21648 errors. I'm not sure what do you mean by offline generator?
1. I opened it on M5, because there's no option to use strategy tester on M2.
2. I got it from http://novacodetrader.com/s/ for free. It's true, renko doesn't have wicks. It's additional mode in this script that allow to show it.
3. Any idea how to avoid these errors?
If you want to test strategies on classic renko chart, do not forget
that, the value of Open can never be used for opening/closing
positions or for calculating indicators, because you get it when the
block closes only. I recommend to use sbs renko chart with wicks instead (google it
if you do not know how it looks).
I converted 10 pips renko block chart, by backtesting RenkoLiveChart
V3.2 with every tick mode, based on M1 OHLC data (then i converted sbs
blocks from them with a script). On these blocks even the original MACD
Sample and Moving Average experts (with a stoploss, and using the
current open value instead of the current close) could make huge profit
on the 8 years
backtest after an optimization. Then i converted the blocks from real
tick data (same broker like at first!) by making a 99% quality backtest of this renko expert, and i
could not optimize those original mt4 experts to make profit any more.
With other words you always need real tick data for generating renko block chart.
I edited the open datetimes of the blocks too, because normally there were more blocks which opened in the same minute. I think that's why the open chart after testing shown opened orders far away from current price. So i exported renko chart from the mt4 and i copied the date and the time columns from a normal M5 chart to this renko block chart csv file. Then the tester always opened position on the right price, but this time conversion caused difference in calculating the swap values.
To avoid the mismatched charts errors i have one idea: use real tick
data to generate blocks (always with wicks), then you can use "open
prices only" mode in the strategy tester. That way you won't get these
errors, plus the backtest and the optimization will be much faster too.
I recommend to try FSB strategy generator with this wicked sbs chart, because it can also give good strategy ideas.
I visit the link but they show
This mailing list is not currently active.
Can you give me the script? My mail address is email@example.com .
Hi, You can do a reliable backtest if you have tick data. I've written a blog post on the topic. This one deals with MedianRenko charts, but the same can be applied to the renko charts.
Here is the link: http://www.az-invest.eu/how-to-properly-backtest-rangebars-medianrenko-renko-and-pointo-using-tick-data