Agree backtesting results must always be treated with extreme caution, but its fascinating trying to keep your real life trading in sync with your model from start of trading. You yourself Skate have pushed me and others to stick to their system signals. Qldfrog and others have noted the dilemmas starting a system from scratch (do you buy all the current strategy positions, or only take new positions?) and what to do if you run out of cash (take partial positions, or only whole)? what if you find an error - sell a strong stock to re-align with the system or stay in it?
It really is invaluable when systematic traders (and systematic discretionary such as Peter2) share their trades or even just comment generally on equity curve and performance. It helps give those struggling up the learning mountain an idea what they're aiming for, how much performance can vary between people on a similar strategy, or hint if they've overfitted parameters or aren't trading their system honestly.
At the end of the day the system is a tool. If Skate or anyone gives away their entire system, its not much different to a virtuso violinist lending their Stradivarius, Brian May lending you the Red Special, or Jack Nicklaus letting you use his clubs - or Geralt of Rivia giving you his sword (confession: Watched too much Netflix over Christmas....!). For the first 5 years or so you have to emulate and copy what you can of those that have succeeded, then eventually you might earn the right to start tweaking bits of the system or re-writing from scratch. Only then will you understand the subtle trade offs and gains and what you risk losing if start dismantling layers of protection for performance.
It’s not the vehicle
Peter’s 'ability', 'skill' & his 'mental toughness' decided the win, it's not when he got into the car (the BUY), not even when he exited the car (the SELL) but how he handled the car during the race. (the TRADE)
Skate.
I've stated before, backtest results mean "Jack" - but backtesting gives an indication between test results using different parameters for evaluation.
Hi Skate With regards to your backtest results for various systems why is it that you are only testing or posting the results over 2019, are you testing these systems in the development phase of longer periods?
@Roller_1 all my strategy development has been completed a year or so ago. Backtesting results from 20 years ago bear no resemblance to current results. I tend to use the last year (the last 365 days) for backtest reports because they are reflective of the recent turbulent times.
FYI
Currently I have only the last 10 years of data (the Silver Norgate Package) but in my development days I had the Platinum Package that included 20 years of data as well as (a) Delisted Equities & (b) Historical index constituents. I've been asked many times before in this thread to supply backtest results for certain years & I'm more than happy to comply.
Skate.
I agree, anything prior to 2005 i feel doesn't bear as much weight as data after to 05. Don't really have any proof but it would make sense things have changed since around then due to algos and everyday punters having access to online brokers etc. was that with Norgate or the old premium data?
how does the CAM strategy look from 2014-18?
@Roller_1 I've been using Norgate NDU for many years before it was released to the public. I had Norgate Premium Data before that.
# The CAM Strategy backtest is from 1/1/2014 to 30/12/2018 inclusive
View attachment 100431
Skate.
One of the ideas i have been toying with lately is a repository of backtest results on recent periods
Whole of 2019
01/7/19 to 1/1/20
01/7/18 to 1/1/19
This covers a few interesting period
I would happily share my system 1 and 2, zigzag mod system and current work on 123 system
Would you agree on sharing your or some of your results @Skate?
And should it not be better to create a new thread for that.
Any feedback on this idea
I know i woukd find it useful to have benchmarks
In AI, there are whole sets of predetermined images etc to be used by developers to test and rate their different software..be it face or gait recognition, etc etc
OK, will try to nail this in the next 2 days and create /populate that thread@qldfrog, sure I'll willing to share any of my backtest results if it help anyone, all you need to do is come up with a format that is consistent across the board for others to follow along & compare.
My suggestion
Lets make it a garden variety backtest. ($100k X 20 positions)
EXAMPLE
Portfolio size $100k
Positions 20
Is that okay with you?
Let me know when you have the thread set up.
Clarification
1. Did you mean a 6 months period ?
2. Or do you want the backtest report for a calendar year ?
3. Or do you want to backtest for a financial year ?
01/7/19 to 1/1/20 ??
01/7/18 to 1/1/19 ??
Skate.
OK, will try to nail this in the next 2 days and create /popu late that thread
I was interested in the 6 months for the following reasons
01/07/18 to dec 18 was a down period,
the 1/1/19 to 1/1/20 full calendar year was a great up momentum
but second half of that year was very up/down and gave a bit of weird signals
I like the ideal of separate 01/07/19 to 1/1/20 as to what a start in an irregular period would give
and the whole year would display the behaviour of a mature portfolio reaching that period
As you may remember one of my portfolio system has a slow ramp up... and so I do not believe the overlapping period is a waste of time?
You said:
Portfolio size $100k
Positions 20
fine with me :neither you or I use this setting but it is a clean neat one, let's use it
Lastly: what is our realm? all ord?(XAO?)
Many thanks and yes I try to create a thread this evening and populate it with my results..nothing as good as yours but benchmarks nevertheless
One of the ideas i have been toying with lately is a repository of backtest results on recent periods
@qldfrog you have come up with a good idea. In the meantime I might post some of my strategies backtests reports tomorrow using my standard backtest settings for readers of this thread. Skate.
Helping Others You might want to dump stuff here to help others
There's no get rich quick in trading. Trading does have the potential to be life changing.
But it is the vehicle (combined with a good driver/rider). duc
Newer traders should review @tech/a 's trade management closely.
1. The initial risk is fully accepted and if price goes down after entry he'll take the loss.
2. Price only has to sneak up a little and tech/a will reduce the initial risk significantly (approx by 1/2).
3. If price goes up a little more. The trailing stop is moved to break-even quickly.
4. Allow price to go much higher until the profit is above average, then protect it.
If you can trade with this discipline through lots of trades, you will be profitable. I would recommend following his trades on the charts each day until you understand why every decision is important.
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?
We use cookies and similar technologies for the following purposes:
Do you accept cookies and these technologies?