Australian (ASX) Stock Market Forum

XAO backtest benchmarks for 2018-2020

Joined
8 June 2008
Posts
12,600
Reactions
18,398
Dear all system builders, I toyed with the idea of having a repository of backtest or even paper/real trading results for various recent periods, to use as a rating/grading tool when developing systems of your own; what to beat, what to aim for or reaching for the sky;
Ideally if backtesting, please ensure you are do not have future leaks..however great the results are :)
And we all know that backtesting will nearly always be different from real trading action so no point discussing this here
The rules:
Portfolio size $100k
Positions 20
Realm: All Ordinaries (XAO)
While my systems are weekly, dayly or monthly system results are welcome

The periods we will test are recent enough to mitigate the survivor bias and chosen to represent interesting market trends
01/07/18 to 01/01/19 : a 6 month down period,
01/07/19 to 1/1/20: a 6 month unstable period,
1/1/19 to 1/1/20 a full calendar year with great growth momentum overall
 
upload_2020-2-17_18-32-32.png
excel details attached
Hope it helps

systems 1 has been traded for a year next week with real world trading matching these back-tests but for a few mistakes from the operator,
system 2 started in August 2019
thanks to @Skate for these 2 systems initial ideas

123 system is under development and based on the 123 or ABC(D) indicator as discussed by @peter2 Many thanks

Still trying to see where my zigzag is wrong..too good to be true so paper trading to detect future leaks if any;
Thanks to @rnr !!!
Looking forward to work on the super-trend next (@Skate @peter2)

I welcome all similar entries :) @Warr87?
 

Attachments

  • upload_2020-2-17_18-25-37.png
    upload_2020-2-17_18-25-37.png
    29.5 KB · Views: 28
  • benchmark qldfrog.xlsx
    12.8 KB · Views: 10
Might be worth specifying standard commissions per trade and slippage for people to apply if sharing qldfrog. For example, I'm trading with CMC with max $11 commissions, and apply 1.5% slippage for buy and sell. Its surprising how even these mild but hopefully realistic handicaps can change a smooth equity curve to more realistic obstacle course :)

I guess while on this line of thought, delays should be specified. Will vary depending on strategy and time constraints, but I use 1 day delay on buy and sell. If compounding returns (matters quite a bit after a few years), how many positions are targetted.

In Amibroker AFL:

//Slippage modelling:
BuyPrice = Open*1.015;
SellPrice = Close*0.985;

//Commisions for brokerage:
SetOption("CommissionMode",2);
SetOption("CommissionAmount",11);

//Trade delays:
SetTradeDelays( 1, 1, 1, 1 );

//Position Sizing and compounding returns:
Positions = 15; //(or whatever)
SetOption( "InitialEquity", 100000 );
SetOption( "MaxOpenPositions", Positions );
SetPositionSize( 100/Positions, spsPercentOfEquity );
 
Indeed, my results were performed using a weekly system, 1 day delay, no slippage and $10 commission per buy/sell
 
3 Actual trading strategies

1. The CAM Strategy with StaleStop
2. The HYBRID Strategy
3. The PANDA Strategy

Backtest Settings - 6 month down period
Start Date: 1st July 2018
End Date: 1st January 2019
Period: 6 months
Portfolio Capital: $100,000
Positions in the Portfolio: 20
Fixed Position Sizing: $5,000 (No re-balancing)

1. The CAM Strategy with StaleStop

2. CAM StaleStop Capture.PNG




2. The HYBRID Strategy

3. HYBRID Capture.PNG




3. The PANDA Strategy

4. PANDA Capture.PNG

Skate.
 
3 Actual trading strategies

1. The CAM Strategy with StaleStop
2. The HYBRID Strategy
3. The PANDA Strategy

Backtest Settings - 6 month unstable period
Start Date: 1st July 2019
End Date: 1st January 2020
Period: 6 months
Portfolio Capital: $100,000
Positions in the Portfolio: 20
Fixed Position Sizing: $5,000 (No re-balancing)

1. The CAM Strategy with StaleStop

2. CAM StaleStop Capture.PNG




2. The HYBRID Strategy

3. HYBRID Capture.PNG




3. The PANDA Strategy

4. PANDA Capture.PNG

Skate.
 
3 Actual trading strategies

1. The CAM Strategy with StaleStop
2. The HYBRID Strategy
3. The PANDA Strategy

Backtest Settings - a full calendar year with great growth momentum overall
Start Date: 1st January 2019
End Date: 1st January 2020
Period: 12 months
Portfolio Capital: $100,000
Positions in the Portfolio: 20
Fixed Position Sizing: $5,000 (No re-balancing)

1. The CAM Strategy with StaleStop

2. CAM StaleStop Capture.PNG




2. The HYBRID Strategy

3. HYBRID Capture.PNG




3. The PANDA Strategy

4. PANDA Capture.PNG

A full calendar year with great growth momentum overall

The results above explains why I made the post below - (a small Portfolio size of $100k & $5,000 bets) multiply that 3 & 10 fold with re-balancing profits gives an indication what can be achieved. Bear in mind, when I win, I win big & when I lose I lose big, it's all relative.
Bull Markets
Getting into a trend is easier than most traders think especially in a Bull Market (last year was rip snorter). Timing the exit is another matter, that takes finesse. Any trader who is not rolling in it at the moment should be re-evaluating their strategy to understand why they missed a great trading opportunity. Sure, I'll be first to admit trading was bumpy at times in the last 12 months but a stalestop eliminates this stress. Winners don't seem to carry the same stress as losing trades. Weekly re-balancing of my position-sizing is another of my personal favourites by putting every dollar to work (soldiers should be out there fighting the good fight not laying idle in the barracks)
Update - of my Hybrid Strategy
The updated Equity Curve of my Hybrid Strategy is for the last 12 months. The chart confirms trading with the trend still works.

Equity Capture.PNG

Skate.
 
Might be worth specifying standard commissions per trade and slippage for people to apply if sharing qldfrog. For example, I'm trading with CMC with max $11 commissions, and apply 1.5% slippage for buy and sell. Its surprising how even these mild but hopefully realistic handicaps can change a smooth equity curve to more realistic obstacle course

Sorry to deviate a bit but why don't you participate in the Opening auction to mitigate your slippage Newt? I just set a LMT order the night before (to sell) 5-10% below the closing price, then you get the opening price, i don't have any slippage issues.
 
@Roller_1 and if it helps @Newt
I do the same: ask abuy higher than previous close and a sell lower than previous close.
In most cases, it get acted on at the Monday open and I do not have any slippage
BUT
Bell Direct sometimes refuses my buy order as they are too high above the estimated open price and so might "tweak" the market.I asked the reason and it is a manually operator controlled action,
If I am deamed too high my order is refused and cancelled: I have had time when I wanted to by at X$ a share (around 2/3% above previous close) my order is refused at 9:30 Sydney time as too high and it actually opens around my X$ value 30 min later and I chase the price manually at X$+ to pass my order

so not an universal recipe,
I also have the case where I need to sell before buying as BD requires the money on the account before allowing the purchase ;
so once fully invested if I want to sell/buy lots I need to first sell then buy and so only the sells are slippage free, I have slippage on the buys

In conclusion, in the real world i have some slippage..sometimes so the slight divergence between backtests and real results
 
Last edited:
View attachment 100439
excel details attached
Hope it helps

systems 1 has been traded for a year next week with real world trading matching these back-tests but for a few mistakes from the operator,
system 2 started in August 2019
thanks to @Skate for these 2 systems initial ideas

123 system is under development and based on the 123 or ABC(D) indicator as discussed by @peter2 Many thanks

Still trying to see where my zigzag is wrong..too good to be true so paper trading to detect future leaks if any;
Thanks to @rnr !!!
Looking forward to work on the super-trend next (@Skate @peter2)

I welcome all similar entries :) @Warr87?

You've called me out so I guess I have too now ;), haha.

I will try and post something on the weekend.
 
@Skate much thanks, your data is much appreciated
giving targets which are achievable and emulating to work further and refine the systems further
 
You've called me out so I guess I have too now ;), haha.

I will try and post something on the weekend.
That was the idea but only if you want to,
:) one of my issue is the isolation: is that 20% gain on this period a great result or pathetic? how can I know if I have no one to compare to? I can not expect stellar results but where should I put my targets
Hopefully these "benchmarks" covering various relatively recent periods will help not only me but others in the same situation
 
You are right. It is hard to tell what would be considered good or pathetic, particularly when you start out designing a system.
 
Post it anyway .. Be a rebel !!!! .... I'm feeling like this forum needs a shakeup in MAE/MFE

that’s all the encouragement I need...I’m here to shake things up baby!!! Not sure why but I had George Constanza’s voice going on in my head when I was typing that out...hahaha. I’ll post my results tomorrow morning.
 
Sorry to deviate a bit but why don't you participate in the Opening auction to mitigate your slippage Newt? I just set a LMT order the night before (to sell) 5-10% below the closing price, then you get the opening price, i don't have any slippage issues.

No probs roller. I put in Limit orders night before trading and generally get close to the Open too, but for aggressive stocks moving quickly you don't always. I guess in answer to your question, 2 reasons:

1. Model the small number of trades where you don't get the Open - not a huge issue for weekly trend trading but your slippage is always >$0 if you average it over the year, so better to have something in your model than be too optimistic

2. Assist with an idea of what price to offer for next day's trading - so frustrating when you miss out, but frustrates me even more if I offer 5% or greater premium and price later recovers


Slippage modelling is admittedly much more important for short term trading (e.g. mean reversion).
I would challenge you to including a bit of slippage in your backtests - even 1% can make a surprising change to your final equity curve (not just lower - generally much more realistic "lumpiness").
 
That's a shame, I'm living trading a daily swing system but it uses a 20% position size so only max of 5 positions. Would like to post the results but doesn't meet your rules so I guess I'm out :(
Please post but highlight the daily factor, and whatever other specifics
So brokerage, fixed position size,etc
I am sure it will be of use.
looking forward to seeing this
 
Here are 3 backtests on the weekly trend system I'm currently trading:
All Ords universe, no slippage, $11 commissions, 20 positions, $100k starting, no compounding:

1/7/18 - 1/1/19
2020-02-18 22_02_51-Symbols.jpg
1/7/19 - 1/1/20

2020-02-18 22_03_21-Films & TV.jpg

1/1/19 - 1/1/20
2020-02-18 22_04_32-Films & TV.jpg
 
Top