Backtesting Renko Bars

[jeremymgp]

Hello,

I believe there may be a flaw in backtesting Renko bars which typically yields unusually good results.

The problem may stem from the the close at end of bar function. When a strategy uses closes at end of bar with time based bars, it actually closes at the open of the next bar. This is more problematic in Renko charts as the open and close are not time-based and I believe are actually plotted at the same time.

Say the Renko bar height is 10 ticks.
If a strategy closes at the end of an up bar and the next bar is also an up bar, the close of the first bar will be the same as the open of the next bar. That's OK.

If a strategy closes at the end of an up bar and the next bar is a down bar, the close of the first bar will be 10 ticks above the open of the next bar.
However, the close of the first bar will also be 20 ticks above the close of the next bar.

With Renko bars - the close and open are plotted at the same time, so when the backtest is said to exit on the close of one bar but actually exits on the open of the next bar, as the open and close occur simultaneously it should actually exit on the close close of that next bar also. This only applies when the two bars are the opposite direction to each other.

Renko backtest results then typically have 50% of trades with the exit 1 Renko bar's height in favor of the trader.

Thanks,
Jeremy

[NinjaTrader_Bertrand]

jeremymgp, thanks for the post - in NinjaTrader 6.5 those are not supported by us (as it's a custom add on bars type) - for NinjaTrader 7 Renko Bars will be added per default and the backtesting on them would be comparable to other regular bars types.

[RJay]

Originally Posted by jeremymgp

Hello,

I believe there may be a flaw in backtesting Renko bars which typically yields unusually good results.

The problem may stem from the the close at end of bar function. When a strategy uses closes at end of bar with time based bars, it actually closes at the open of the next bar. This is more problematic in Renko charts as the open and close are not time-based and I believe are actually plotted at the same time.

Say the Renko bar height is 10 ticks.
If a strategy closes at the end of an up bar and the next bar is also an up bar, the close of the first bar will be the same as the open of the next bar. That's OK.

If a strategy closes at the end of an up bar and the next bar is a down bar, the close of the first bar will be 10 ticks above the open of the next bar.
However, the close of the first bar will also be 20 ticks above the close of the next bar.

With Renko bars - the close and open are plotted at the same time, so when the backtest is said to exit on the close of one bar but actually exits on the open of the next bar, as the open and close occur simultaneously it should actually exit on the close close of that next bar also. This only applies when the two bars are the opposite direction to each other.

Renko backtest results then typically have 50% of trades with the exit 1 Renko bar's height in favor of the trader.

Thanks,
Jeremy

Jeremy,

Do not backtest strategies with Renko Bars.

It is not a valid test. Renko bars only show open and close for each bar.

They do not show the high and low for each bar.

On larger bar sizes, Renko bars will break the bank if you get my drift!!!!

Just my opinion,

RJay

[jeremymgp]

NT7 sounds tasty, thanks and good luck!,

Jeremy