Backtesting using Renko bars close price

[burtoninlondon]

Hi,

I have a strategy using Renko bars which processes the logic at the close of each bar. The strategy logic uses the close of the current and previous bars to generate the trade signal.

I understand that backtesting in NinjaTrader using Renko bars is not advised due repainting of the bar. If I am not bothered by how the bar forms, just the close price and whether it is red or green will that still have issues for backtesting?

Thanks,

Neil

[QuantKey_Bruce]

Yes, unless you take extra steps e.g. 1-tick series for back-testing and place all of your orders there - a key problem is that the Renko OHLC bars do not reflect the prices that actually traded during that bar's time period.

[NinjaTrader_Emily]

Hello Neil,

Thank you for your post.

You could still see potential differences between real-time Renko bar values (including the close price and whether it is red or green) and the results from a backtest if you do not use a workaround to aim for achieving more accurate results. You could consider testing using the Playback Connection rather than the Strategy Analyzer, or use a Renko bar type with either the High Order Fill Resolution of 1 tick or submitting orders to a single tick data series as mentioned in the section at the bottom of this page:

• https://ninjatrader.com/support/help...ime_vs_bac.htm

For a more in-depth description of the differences between backtest vs. real-time results, please see the following post from a colleague:

• https://forum.ninjatrader.com/forum/...ive#post773377

Please let us know if we may be of further assistance.

[burtoninlondon]

Yes, unless you take extra steps e.g. 1-tick series for back-testing and place all of your orders there - a key problem is that the Renko OHLC bars do not reflect the prices that actually traded during that bar's time period.

Thanks QuantKey_Bruce, I can understand the high & low's not being different during the bars' time period to the final bar but would the open & close prices not be the same?

[QuantKey_Bruce]

No, that is not correct. The opening price is not necessarily a price that even traded during the time period of the bar (because the Renko bricks are all forced to be the block size tall even if only one price traded during that time before the bar advanced) and the closing price is almost certainly not the last price that traded because the price has to trade past that for the block to close. So, if you were to graph just the OHLC of the prices that actually traded, many small Renko bars would actually be dojis.

[burtoninlondon]

QuantKey_Bruce - thanks for your reply, it has helped me better understand the issues.

I guess as an absolute minimum if you do need to use StrategyAnalyser to backtest a strategy using Renko bars then you'd need to factor in a high degree of slippage if using the close prices to 'fudge' a more real world price.

[QuantKey_Bruce]

Well, yes, and what's really more severe is many strategy designers plan on entering with a limit order, and those are almost certainly going to give false results when tested on Renko bars unless really serious measures are used to get accuracy in the tests. Almost every simple and easy test of limit order strategies on Renkos is false.