Back Testing Not Respecting Spreads

Viewing 8 posts - 1 through 8 (of 8 total)

Author

Posts
03/23/2021 at 12:09 PM #165021 quote

charleskyriakou
Participant

Junior

I have a back testing strategy for the Tesla (TSLA) market which returns the same gain results whether or not I specify a spread. E.g if I specify a spread of zero I get the same results as when I have a spread of 100. By the way, I have tick mode on. This hasn’t happened when I have put together strategies for Wall Street, where the specified spread is respected. Have anyone else found a similar situation or know why this could be?

03/23/2021 at 4:15 PM #165063 quote

GraHal
Participant

Master

charleskyriakou wrote:

I specify a spread of zero I get the same results as when I have a spread of 100.

Specify meaning entered in the backtest engine settings (not hard coded in)?

charleskyriakou thanked this post

03/23/2021 at 9:16 PM #165098 quote

charleskyriakou
Participant

Junior

I mean that I click on the Spread check and enter the number of points. Not in the code. Please see attachment. Thanks for taking an interest GraHa. Much appreciated!

GraHal thanked this post

2021-03-23_20-13-22.png

03/23/2021 at 10:42 PM #165116 quote

GraHal
Participant

Master

Attached image 1 is with spread = 0, image 2 is with spread = 100

There is a difference of 9827 – 9175 over 669 trades.

It would appear there is some inbuilt limit on spread as a % of price or some absolute limit on spread value?

Why would you use / enter a spread of 100 anyway?

Why not use what the spread actually is … 0.41 currently??

charleskyriakou thanked this post

Tesla.jpg Tesla-2.jpg

03/24/2021 at 12:29 PM #165153 quote

charleskyriakou
Participant

Junior

Interesting result. Thanks for that. Correct me if I’m wrong but (9827 – 9175)/ 669 trades means an implied spread limit in the system of 0.974. I used a spread of 100 just to see whether there would be a difference in the resulting gain. However, as you’ll know, Tesla has a very volatile spread which seems to vary between 40 and even over 100 depending on when one trades so in a back testing strategy I like to use more of an average spread rather than what the spread is at the time. For Wall Street I specify a spread of 3.2 which is excessive but introduces a margin of safety, and find that the back testing and actual results in my IG demo/live accounts against live data are pretty much in line as a result.

03/24/2021 at 1:52 PM #165170 quote

GraHal
Participant

Master

charleskyriakou wrote:

implied spread limit in the system of 0.974

The difference between above and 1 is maybe due to slippage?

I agree with all of the rest of your comment and I do the same re use worst case spread etc.

I tests I posted above were from on my CDF Platform … hence the spread I quoted of 0.41.

You must be testing on your SB Platform when you say spread 40 and even over 100? True price (currently $669) is multiplied by 100 to give point size = $1 on IG SB Platform (current Tesla price is $66,900).

I just bought 1 Tesla share on IG SB and straight off I am $87 down due to spread! 🙂

charleskyriakou thanked this post

03/24/2021 at 1:54 PM #165171 quote

GraHal
Participant

Master

GraHal wrote:

The difference between above and 1 is maybe due to slippage?

I deviated in previous post … so spread of 1 on CFD would be same as 100 on IG SB Platform.

I’ll do another test later and let you know!

charleskyriakou thanked this post

03/24/2021 at 2:04 PM #165172 quote

GraHal
Participant

Master

Attached image 1 is Tesla on SB with spread = 0

Image 2 is Tesla on SB with spread = 100

Oh for no spread, we all be billionaires! 🙂

So spread of 100 is taken into account, There were 2,545 trades with spread = 100 … you do the maths from values on rhs of equity curve.

SBTesla-2.jpg SBTesla.jpg
Author

Posts