Why Heiken Ashi and Point and Figure bars, Renko, FlexRenko, Point original  bars are not supported.

A experienced GSB user made this comment.

Renko bars back-test differently than they forward test, even if you select the options to display wick and show real open. I have tested them extensively, and they just aren’t appropriate for GSB. Point Original  bars – in MC at least, don’t know about TS’ implementation – also behave differently in forward vs. back testing. In MC, Kase bars are the best alternative.



Recommended.

Kase


Not Recommended

Renko

FlexRenko

Point Original

Line Break

 

 

Totally Unusable

Point & Figure

Heikin-Ashi

*Kagi

*Reversal


*The user commented later

"Recent revelations have been regarding the significance of slippage. Despite my earlier statements to the contrary, one can indeed build a successful strategy based on MC’s Kagi and Reversal bars, if one allows for slippage of (bar points * tick value * 2) + commission. Likewise I’ve realized that slippage on bar types like Point and Kase needs to be at least 4 points. It is essential to compare back testing and forward testing over the same period to see how much slippage one can expect for a particular bar type. For too long I was assume too low a slippage. And it may not even be “slippage” per se we’re witnessing, but actually inherent behavior of the bar type as it forms new bars."

* Dec 2022,  Much later the user said "’Unfortunately I have given up on Renko and Range/Point bars, because I cannot get forward testing with them to match back testing.



There are differences between IQFeed, Rithmic, and CQG data, but not enough to warrant much concern. CQG seems to have the best quality data, but only 30 days of history. Rithmic now blocks you from downloading much history, whereas IQFeed reliably provides 180 days of tick data history. So I stick with IQFeed.

I’m still focused on determining the best bar types, and their associated quirks, such as “slippage” (for lack of a better term) due to how the bars are formed. For example, Kagi bars can be useful if you take into account that slippage for them is usually = ((2 * point/tick size of bar) + 3 ticks for bid/ask difference and true slippage) * tick value. So an 8 point YM Kagi bar would need to expect minimum slippage of ((2 * 8) + 3) * $5 = $95.

I used to think that Point and Point Original bars were useful, but there are significant differences, not attributable to slippage, between backtest and forward test results. Trades are taken at different times. In terms of simple price-based bars, Kase bars seem to be the only trustworthy ones forward vs. backward testing




Another user made another important comment about Tradestation data (Which i consider to be good quality data)

The tick data we have purchased from Kibot matches the Rithmic data exactly. The real issue and eyeopener is that the TS tick data is missing large ranges of ticks. Other sources of tick data from Rithmic and IQfeed seem to agree with each other.


My personal opinion is Tradestation tick data is also too short to build systems from. It can be done but there is a significant disadvantage having only 6 months or so of tick data.