Dump it Here

[Skate]

New Trading Idea
With all trading ideas, we are trying to tilt the odds of success in our favour by buying into trends whilst minimizing our losses. Buying good companies at the "right time" & getting off quickly "at the right time" when it all goes wrong is the simplicity of the system.

"The Ducati Daily Blue Bar strategy"
This strategy simply buys the first "blue bar" & sells the first "red bar" it simple & surprisingly effective. I've used @peter2 idea of colour coding the trend with "Blue Bars" so it's simple to follow along by looking at the charts. The "Blue" bars meets Duc's two conditions & the "Red" bars are when Duc's conditions "are not met". I've profited from Duc's advice where others might not have realised the opportunity that was posted. "The Ducati Daily Blue Bar Strategy" is clear to understand. The first "blue bar" is the signal bar (the white arrow) & we buy on the "next blue bar" at the open (the white square). The position is held till the first "red bar" & sell that position on the open on the next "red bar".

Making easy money
Trading the "Blue" bars & selling on the "Red" bar is a recipe for success.

Using a modified (RSI)
Let's turn or focus onto "Skate's modified RSI" indicator & explain how it works in relation to the "HappyCat Strategy"

The "HappyCat Strategy" is a pure Momentum Strategy
Momentum shifts the share price, "that's a given". Our decisions when to enter & when to exit a move really determines if we are to be profitable in the long run. I've calculated the momentum over a period from bar to bar as the entry signal. The exit signal is the reverse of the entry condition.

Momentum is either on or off ("Skate's modified RSI" indicator is green or red)
Momentum trading is simply trading the trend using the "momentum effect". The strategy works on the principle that momentum is either on or it's off there is no middle ground. Additional setting & parameters assist in the fine-tuning. The secret of momentum trading is the ability to filter out the false signals. The standard (RSI) indicator identifies potentially overbought & oversold levels that may reverse direction. However, the default lookback period of (14/15) doesn't work well. "Skate's modified RSI" is the stock standard (RSI) with a modified lookback period, nothing elaborate at all. The colour coding of the modified indicator identifies when the momentum is on & off. We buy when it's on & use the "off" period as a layer of our exit strategy.



Looking at the standard (RSI) line
The standard RSI is (the bottom line) that I found unhelpful as far as I'm concerned. The sensitivity of signals are found wanting using the (14/15) lookback period.

Skate.

 

[Skate]

I'm completely unsure about your sell signal on this one, especially with the price of gold increasing and hitting ATH's.

@Fury the "HappyCat Strategy" indicates that momentum has stalled after entering the position two weeks ago. If the strategy gets it wrong & sells too early "who cares" not me - we can always buy back in.

I'm completely unsure about your sell signal on this one, especially with the price of gold increasing and hitting ATH's.

The Buy & Sell pressure chart
They say a picture "paints a thousand words" & the picture doesn't hold the same view as you. What "you or I think" when it comes to trading is irrelevant. If you aspire to be a systematic trader "trade the signals" as they come along - you have one job to do & it's our responsibility not to stuff it up.



When you have one job
"just do it" correctly



Skate.

 

[Skate]

I was unimpressed with the recent COVID-19 flash crash (I had to give some of my profits back)
The "HappyCat Strategy" buys momentum. Using momentum to enter a trend is a handy tool but as you can see from the backtest below (version two) exits when momentum turns off where (version one) relies on a combination of confirmations that the momentum has really stopped before it exits.

The exit really determines if we are to be profitable in the long run
There are no free lunches when it comes to trading, it all about swings & roundabouts. Trading is a tradeoff.

Sharing is caring
I'm very curious how other strategies performed during this period.

I would be very interested if others would post a backtest of their strategy during the recent COVID-19 period
The backtest period is from: "1st January 2020 to 30th June 2020" (All ordinaries Index)

These are the "HappyCat" backtest results from the COVID-19 period - 1st January 2020 to 30th June 2020



Skate.

 

[Skate]

Moreover (for full disclosure)
The "HappyCat v2 Strategy" was the winner in the previous backtest "shoot off" so it's only fair that I balance the argument to which is the "better strategy" (v1 or v2). I've included a backtest for the last 365 day or exactly one year for comparison purposes.

Backtest report
It was not fair that I only post a backtest of the last half of "2019/2020" financial year or the "first half" of the calendar year. Cherry-picking a time period is rarely helpful.

# Backtest period - the last 365 days (1 year backdated from today)
The backtest report is for the previous 365 days






# Portfolio Equity curve




Sometimes too much information is too much

Truly appreciate the information overload you're providing.

Skate.

 

[Sir Burr]

Why do traders lose at trading?
You have a system. It trades 'X'. It is backtested, drawdown does not exceed 23% across a market history of 10 years. Looks good. You forward test it, drawdown does not exceed 17.5%. Looks good. You trade it live. You immediately go into drawdown over a period of 3 weeks by let's say 39%. Is your system broken? What do you actually do at this point? If you cannot answer this question, you have no business in the markets yet.

I'm interested in the answers.

A top shelf question Duc. I'd be sweating like a snowman in heat!

Every time my system exits with the index filter then re-enters, it immediately goes into drawdown as I sell off and crawls back. Happening the past few months yet again and never nice but don't deviate.

39% would be nasty. In that particular case, I'd be going back another 10 years (20 total) and check results. I'd also be looking at the recent market period and if there was something similar in the past 20 years to compare before considering it broken.

BTW if you had said it was already backtested 20 years, then what? Again looking at the past for a similar market. If most recent was actually worse, then consider continuing. If not, then maybe adjust with fresh new backtests

 

[Newt]

Hi Skate,

I started to write this post Fri night then held off, but in keeping with the sharing shown so often here, have written up some backtests from my newest system, cooked up pre-Easter in the heat of Covid. For over a year I'd meant to do a serious dis-assembly and rebuild of CAM in line with my own experiences and preferences. During this time finally took the trouble to track entry and exits on the price pane, calculate and display % profit for closed trades. Seeing backtests from yourself and others led to me targetting certain goals, including:

- (hopefully) improved average win
- reduced average loss (previoiusly always over 15%)
- (hopefully) reduced DD even in very difficult periods
- take more weekly trades, hopefully with smoother equity curve, faster entries into new strong market periods

Many ideas for entry, exit, trailing stop, index filtering were tried to create new systems in addition to CAM. A couple of ideas showed more promise than I expected and warranted more attention. The backtests below are from a strategy incorporating principles of your variable trailing stops (tigthening off ROC and index filter), trend filter ideas of my own, entry signals inspired by Panda posts many months ago. After many months of stress-testing parameters decided to trade this from May.

Have heard Mike Bellafiore say in "Chat with Trader" podcasts that people underestimate just how hard it is to get "edge", and that even the best traders may have now greater than 50% winning trades. He went on to say a lot of "psychology" challenges stem from having to cope with poor performance and edge.

Anyhow, last few months the stronger expected edge versus previous systems has held for this beast. I'm sharing these tests with and without ticker ZNO included, to show how just one stock can lead your backtesting and optimising astray if not careful. Actually that was another goal here - to reduce reliance on long trends in a small number of tickers (something I'd designed my previous strategy to do from the ground up), but ironically a strong trend in ZNO still affected outcomes.

8/8/19 - 8/8/20 (without, then with ZNO - All ords and historical constituents):







1/1/19 - 31/12/19 (without, then with ZNO - All ords and historical constituents):

 

[qldfrog]

Hi Skate,

I started to write this post Fri night then held off, but in keeping with the sharing shown so often here, have written up some backtests from my newest system, cooked up pre-Easter in the heat of Covid. For over a year I'd meant to do a serious dis-assembly and rebuild of CAM in line with my own experiences and preferences. During this time finally took the trouble to track entry and exits on the price pane, calculate and display % profit for closed trades. Seeing backtests from yourself and others led to me targetting certain goals, including:

- (hopefully) improved average win
- reduced average loss (previoiusly always over 15%)
- (hopefully) reduced DD even in very difficult periods
- take more weekly trades, hopefully with smoother equity curve, faster entries into new strong market periods

Many ideas for entry, exit, trailing stop, index filtering were tried to create new systems in addition to CAM. A couple of ideas showed more promise than I expected and warranted more attention. The backtests below are from a strategy incorporating principles of your variable trailing stops (tigthening off ROC and index filter), trend filter ideas of my own, entry signals inspired by Panda posts many months ago. After many months of stress-testing parameters decided to trade this from May.

Have heard Mike Bellafiore say in "Chat with Trader" podcasts that people underestimate just how hard it is to get "edge", and that even the best traders may have now greater than 50% winning trades. He went on to say a lot of "psychology" challenges stem from having to cope with poor performance and edge.

Anyhow, last few months the stronger expected edge versus previous systems has held for this beast. I'm sharing these tests with and without ticker ZNO included, to show how just one stock can lead your backtesting and optimising astray if not careful. Actually that was another goal here - to reduce reliance on long trends in a small number of tickers (something I'd designed my previous strategy to do from the ground up), but ironically a strong trend in ZNO still affected outcomes.

8/8/19 - 8/8/20 (without, then with ZNO - All ords and historical constituents):

View attachment 107135


View attachment 107136


1/1/19 - 31/12/19 (without, then with ZNO - All ords and historical constituents):

View attachment 107139

View attachment 107138

Seems great.well done

 

[Skate]

A couple of ideas showed more promise than I expected and warranted more attention. The backtests below are from a strategy incorporating principles of your variable trailing stops (tigthening off ROC and index filter), trend filter ideas of my own, entry signals inspired by Panda posts many months ago. After many months of stress-testing parameters decided to trade this from May.

@Newt what a great post
Thank you for the backtest captures they are impressive, well done. No wonder you decided to trade the strategy from May. Your post is well constructed, inspirational & littered with gems. I'm glad the "Dump it here" thread inspired you to keep looking for improvements.

There is an old saying that I live by
"Good is not good when better is expected".

You should be proud of your efforts, I am.

Skate.

 

[qldfrog]

Is Mr Skate in Victoria?
The current stream of data will take me weeks to analyse.
For those of us in Qld with good pleasant weather, and some freedom left, it is hard to find the time to analyse this flow, and work on it.
I believe we have just found one advantage of martial laws and dictatorship: system building LOL
Please keep the data flow, as long as Joe Blow keep the site post safe and searchable, this is Gold

 

[ducati916]

A top shelf question Duc. I'd be sweating like a snowman in heat!

Every time my system exits with the index filter then re-enters, it immediately goes into drawdown as I sell off and crawls back. Happening the past few months yet again and never nice but don't deviate.

39% would be nasty. In that particular case, I'd be going back another 10 years (20 total) and check results. I'd also be looking at the recent market period and if there was something similar in the past 20 years to compare before considering it broken.

BTW if you had said it was already backtested 20 years, then what? Again looking at the past for a similar market. If most recent was actually worse, then consider continuing. If not, then maybe adjust with fresh new backtests

Your 'edge', whatever that may be, needs to be constantly monitored. It may be momo, value an indicator or combination of indicators, whatever: it needs to be monitored. Styles (tactics) gain and lose traction in markets over time. First order of business: is your edge still an edge.

The most successful systems, mechanical or otherwise, find an edge in a market truism: something that is consistent through the decades and cannot go out of fashion or be subjected to manipulation over any great period of time. Almost anything can be manipulated in the short term. The shorter the time frame, the greater the manipulation can be (that is almost a market truism).

Assuming that the above is true, generally speaking a tactic is to take a sub-set of that edge and refine it to instruments and time frames that you build your engagement with the market around. It is often the time frame element that changes in some way. Take 1 variable (since it has been discussed quite extensively on this thread) Vol.



There are 5 quite distinct time frames: (a) 2009/2013, (b) 2013/2017, (c) 2017/2020 and the current period. Now I don't backtest systems, but if I did (ignoring the current period for the moment) I would break down each period individually as a test and then the aggregate for the entire period. The point being that they will (or should be) different. Different gains/losses. This is a market truism: vol. fluctuates. Long back tests could hide that fact. That is just one example. In any given system there will be multiple factors and variables that lurk unseen or unnoticed in time frames. As a further example of what I look at, as this will be an issue if Biden takes the WH.



The point being: s**t can change and if you are not cognisant of these variables, your system may well be through the instruments you are trading.

Outliers: this catches traders out time and again. So we have an established trend. Boom. COVID happens. Market sells off. Clearly your system isn't broken, it was just never designed for something like that. The question then becomes: is the trend broken (shut down the system as many did) or is the trend intact and you double down? Well the answer is outside the scope of this post, but look at Market Bottoms and Trading the Bounce, which are really about trying to decide whether the outlier is fleeting or trend destroying.

The takeaway: know your edge, check your edge constantly against market conditions and make sure it is still relevant.

jog on
duc

 

[Skate]

Is Mr Skate in Victoria?

Nope, I live in NSW.

Skate.

 

[qldfrog]

Nope, I live in NSW.

Skate.

Missing travel maybe..i do but actually too busy right now to spend the time i need on my systems
Thanks for your time.this is the only priceless item in life

 

[Newt]

Fantastic post, many thanks Duc. Skate has hinted more us would benefit from listening and then rolling our sleeves up. We probably all pick up on different parts of your message(s) depending on trading experience, outlier "lessons" learned, and personal mindset. We're fortunate to have such helpful information density, but sometimes it pretty hard trying to drink from the fire hose!

During the few weeks of intensive (more than F/T) effort in trying to be creative, tweaking existing systems, and backtesting noted that others here have quiet different approaches to bactktesting process, timeframes and outcomes sought. I can now see more clearly that many years ago what I thought were filters to improve "edge" and reliability were actually filters that after many iterations were overfitting the strategy to a few strongly performing stocks over the period in question.

I still don't have the skill to reliably incorporate Vix in my strategies, BUT definitely reading ever bit of info on how Duc, Skate and others use it, watching the graph through different market periods, and yes there have been good examples over last 2 years. Information from Norgate market internals (Cum new highs and lows) was one of a number of indirect "indicies" I've watched for many years and have now (hopefully) successfully incorporated within my market index filter. Probably no surprise the rapid market gyrations and VIX spikes co-incide with my longer weekly system going "defensive", even though the fundamental driver is currently non-Vix. For same reason try to follow with great interest macro Duc posts on interest rates, gold, silver.

What you say Duc about longer timeframes makes total sense - Mike B traders would of course be working in minutes/hrs/day timeframes. Fortunately amateur ASF traders till seem to be able to find edge reasonably reliably in the weekly charts, less so in daily - let's hope the algos don't erode that too soon in years to come.

Re backtest periods and changing edge, definitely agree there. Looking at this with more grey hair than previously, believe its wrong to keep tweaking a strategy on the expectation of smooth returns across the years. Skate has previously aluded to challenging test period (e.g. 2008, 2011) and Feb/March 2020 is a new one to add to that list. Something gleaned from watching others in the strategy kitchen here is the importance of "getting the hell in there" quickly and systematically in the good times, but quickly going defensive in the tough times.

I'd be interested to compare notes with others on a few other elements now Duc has stimulated the thought processes. As try as I might, for systematic weekly ASX trend following, it seems difficult to stray from the points below. I'm not worried, about this, but strongly suspect it reflects the way I create code, optimise etc. In science, the most exciting breakthroughs tend to come from very unexpected new experimental data that doesn't fit the existing model - so in that mindset curious where others diverge/disagree. So, the points in question:

* Early entry into breakouts is best, BUT performance often stronger as "Trend check" filters accelerate (trade off between being first in versus taking more reliable trend signal)
* Market index filter - when turned "down" or "off" - best value seems to be tightening stops - much less so blocking new trades (very strong stocks can still occur, particularly at the end of tough market periods and you don't want to miss climbing on board) - that is, use market index and individual stock momentum/strength together. In extreme downturns, do you sell everything or wait for individual stocks to turn down? (thought I'd never agree to doing this but the Coronacrash seriously challenges that mindset!).
* Position sizing - should this be fixed (e.g. fixed fractional) or vary according to market conditions - personally I'm very conservative on this because of the over-fitting risk in backtests (and Howard Bandy books) but interested if others increase total capital or number of positions in very strongly trending markets. (I don't - just allow same number of positions to have larger sizing as the account capital grows (or shrinks)).
* Edge is very hard to come by and refine, but there seem to be an amazing number of market opportunities (stocks) each week in the ASX during most periods. When you play with numerous strategies over different market regime periods, you start to realise that your strategies are often only taking a fraction of the available entries depending on starting date. Part of this will be luck of entering at a strong/weak market period (as per Duc above) but Monte-Carlo work usually shows how many other trade paths could have occured and definitely worth monitoring
* Following from the last point, the risk of losing edge from publically sharing trade entry/exit, strategy etc here for longer timeframes is probably negligible - there are just so many ways to skin the cat and the markets are always there.

Agggh - this is bound to be getting too specific now to the point of less value for others. Enough waffle for now.

 

[Skate]

* Following from the last point, the risk of losing edge from publically sharing trade entry/exit, strategy etc here for longer timeframes is probably negligible - there are just so many ways to skin the cat and the markets are always there.

@Newt another amazing post & your last point is on message. I'm an oversharer but that's just me & in the scheme of things it makes no difference. There is a ton of money in the markets & I only want a very small slice.

Skate.

 

[CNHTractor]

Fantastic post, many thanks Duc. Skate has hinted more us would benefit from listening and then rolling our sleeves up. We probably all pick up on different parts of your message(s) depending on trading experience, outlier "lessons" learned, and personal mindset. We're fortunate to have such helpful information density, but sometimes it pretty hard trying to drink from the fire hose!

@Newt , fully agree with your fantastic post.

I have read along with @Skate (& traded along the Action Strategy) particularly since the Covid19 crash. This period has opened my eyes to Index filters. I have implemented the Index filter as a "caution" to only turn off on another indicator, hopefully to be able to catch this stocks which take off notwithstanding the general market conditions.

 

[Skate]

@Newt I'm still interested
Would you please post a backtest capture of your new strategy with the "Portfolio Equity & Profit Table" during the recent COVID-19 period (without removing any securities) - the backtest period is from: "1st January 2020 to 30th June 2020" (All ordinaries Index)

These are the "HappyCat" backtest results from the COVID-19 period - 1st January 2020 to 30th June 2020
The Backtest results include the "Portfolio Equity & Profit Table"

I've worked long & hard on the "HappyCat Strategy"
I've found using momentum as a sentiment filter works a treat & is robust enough to withstand another black swan event (FYI, a pure momentum exit strategy works the best)



Skate.

 

[Newt]

Happy to help Skate. I have a little (0.5%) slippage dialed in both ways normally - probably doesn't matter much zoomed in this close - should that be removed?

 

[Skate]

Happy to help Skate. I have a little (0.5%) slippage dialed in both ways normally - probably doesn't matter much zoomed in this close - should that be removed?

@Newt just post as the strategy is designed. It hard for me to accept the difference (ZNO) made to the backtest results. Outliers are a requirement for a trend trading strategy to be profitable as the small wins & losses tend to cancel themselves out.

Skate.

 

[Skate]

@Newt , fully agree with your fantastic post. I have read along with @Skate (& traded along the Action Strategy) particularly since the Covid19 crash. This period has opened my eyes to Index filters. I have implemented the Index filter as a "caution" to only turn off on another indicator, hopefully to be able to catch this stocks which take off notwithstanding the general market conditions.

@CNHTractor I'm amazed to learn that you are also trading along with the "Action Strategy". I have a few experienced traders doing the same & personally I was chuffed to learn that they were - it wasn't expected.

Why join in?
I would have expected a few new members wanting to get their feet wet but not seasoned traders. Honestly, I would have found it difficult "if not impossible" to trade along with an unknown strategy, created by "someone" I didn't know, nor would I have ever recommended others to do so either.

Skate.

 

[Warr87]

@Newt another amazing post & your last point is on message. I'm an oversharer but that's just me & in the scheme of things it makes no difference. There is a ton of money in the markets & I only want a very small slice.

Skate.

@Newt
Losing the edge from shared entry/exits would be more of a worry if people actually followed through the system to a Tee. the psychological component of trading is much more likely to erode an edge IMO then technicals.

most people will fiddle or over-ride the system. they wont trust it and think they know best and alter it when it suits. i think people forget that the statistical results of the backtest are only valid for the system you trade. as soon as you start changing the system you have invalidated your backtest results and its now an untested system. (fixing a coding error or similar is different to fiddling with entry/exits/filters. and hopefully those are sorted in paper trading before live trading.)

that's not to discount overcrowding where more participants utilizing a common (or highly correlated) system will detract from the potential gains. but as @Skate says there is a ton of money in the markets. at our hobby scale it is unlikely to impact us too much.

also an important thing I like to remember with edge is the metaphor of an edge ... that is, the sharper an edge the quicker it will blunt. full quant strategies that exploit a razor edge will find it is short lived (think arb and HFT strats). My own systems are very basic and not exploiting nuances, but rather just riding price movement. its not a sharp edge, but it stays sharp enough throughout numerous markets to be profitable on a longer timeframe.