Dump it Here

[Skate]

The BBO Weekly Strategy
Start Date: 1st January 2020
Portfolio Capital: $300,000
Positions in the Portfolio: 20
Fixed Position Sizing: $15,000 (No re-balancing)

Weekly Update Format
1. The "Share Trade Tracker" Dashboard
2. Portfolio performance line chart
3. Open Summary
4. Pending buy & sell positions























Perfect scenario
This week hasn't been kind to traders but it's the perfect scenario for paper trading (let's see if the strategy can recover)

Skate.

 

[Newt]

Phillip Lasker gave the ABC News business report tonight, and put up the graph below. This week we had the dubious honour of experiencing the fastest Wall St correction (>10% fall) ever:

The link probably won't stay current for long, but full report for today (Friday) here:
https://www.abc.net.au/news/business/kohler-report/

 

[Skate]

Phillip Lasker gave the ABC News business report tonight, and put up the graph below. This week we had the dubious honour of experiencing the fastest Wall St correction (>10% fall) ever: The link probably won't stay current for long, but full report for today (Friday) here: https://www.abc.net.au/news/business/kohler-report/ View attachment 100876

Well what a week !! I've looked back at the markets over many decades trying to understand the natural patterns of falls but this time the pattern was "servere & quick", the quickness has caught most out, not only Weekly Trend Traders. My only advice is to put dollar amounts into percentages so the evaluation of this week results retains perspective.

@Newt the graph posted above "time taken to fall 10%" sums up the current situation nicely putting this week into perspective confirming this pattern (fall) is different to most others as this fall has happened so quickly.

WTF - what's going on?
That's an acronym for (Way To Funny) - The Pre-Market Trading (futures contracts) below has put a "smile on my dial" (go figure) - what are we to think? It's an important point that most may not be aware of: "The Dow was down nearly 1,086 points on Friday before closing down -357 points", nice late recovery. Again, what are we to make of this?



Dow in perspective (down 3,583 points this week)
The Dow (INDU) closed 357 points, or 1.4%, lower, on its seventh day in the red. At its worst, "the index was down nearly 1,086 points on Friday" The DOW dropped 3,583 points this week, including its worst one-day point drop in history on Thursday. On a percentage basis, it was its worst week since October 2008, as it fell 12.4%.

S&P 500 this week (down 11.5%)
The S&P 500, the broadest measure of the stock market, fell 0.8%. It dropped 11.5% this week, its worst weekly percentage drop since October 2008 as well. The selloff has shaved $3.2 trillion off the S&P this week. It has dropped $3.6 trillion from its February 2019 high.

Nasdaq (-10.5% down this week)
The Nasdaq Composite ended flat. The index fell 10.5% for the week.

Skate.

 

[kid hustlr]

I think the chart about fastest fall is interesting but there have been far worse weeks in the history of the mkt to my understanding.

Not denying it was a tough week.

Disclosure 1 my buy and hold etc portfolio got hammered.

Disclosure 2 I bought a parcel on Friday.

 

[Newt]

You're probably right about size of rapid drops kid hustler - I should have clarified that is time to fall 10% from a market peak.....

 

[jbocker]

Plan B
Going to a beach wedding today 3 PM at a Beach in Perth. If you live here and look out the window at the current storm lightening and rain you would hope the Bride and Groom have a Plan B.

I guess a lot of people are looking at their plan B's over the last week in this market.

I feel sorry for my nephews new bride a beautiful English Lass who decided to come to sunny Australia and has dreamt of a beach wedding on some of the best beaches in the world, probably would not have dreamed or risked a beach wedding in England.

Dreams. Reality. Plan B.
It is still beautiful.

...and sometimes despite the odds, Plan A still comes off. The wedding was done on the beach after all.Here is a shot of the beach from the reception later after the wedding

 

[aus_trader]

I think the chart about fastest fall is interesting but there have been far worse weeks in the history of the mkt to my understanding.

Not denying it was a tough week.

Disclosure 1 my buy and hold etc portfolio got hammered.

Disclosure 2 I bought a parcel on Friday.

Yes as Skate also mentioned, things have to be put into perspective in % terms. It's bad what happened last week, but not the worst week in history. The 1929 falls were so quick and sharp, which eventually lead to the great depression and were far worst in % terms. Most weren't around including myself to remember or to reflect on... which would have helped us to really look at the current falls with more experienced eyes.

 

[Newt]

Another chart.....

 

[Skate]

P2 Wkly/Dly portfolio: Update after the virus panic selloff week. We'll all remember this week and we'll be better for the experience. If we didn't know that the market can fall this hard, we do now. When the market falls this hard, diversification across few or many positions means nothing. Everything gets hit, hard. All portfolios got hit. Hopefully many of you will now know that real diversification is a much broader concept than how many stock positions we hold.
Now, what are you going to do? Hopefully the system traders will continue following there plans without any adjustments. One disastrous week shouldn't invalidate their systems.

Its a bit easier to take and work through knowing others are also coping and working through their trading plans. Thanks for these wise words Peter2.

@peter2 made a great post recently about diversification & why one disastrous week shouldn't invalidate a system traders strategy.

Let's look at 2 backtests over a different period of a few systems
There is a fine line when it comes to the period used for backtesting as the results can vary significantly. This last disastrous week can cause traders to reflect if their trading systems is still robust enough going forward. This weekend gives you the time to validate a system if you are concerned before the next open on Monday to confirm that the current parameters are still within an acceptable average range.

Examples below
To give an example I've posted a few strategies with backtest results for a short period (this financial year) compared to the same unaltered system averaged over the last 3 years. Some of the results are like chalk & cheese. The CAM Strategy blossomed compared to the other systems for this financial year & it's not hard to understand why, the CAM strategy enters on pullbacks & trend continuations where the other strategies all jump on uptrend as early as possible.

This "Financial year backtest" from (1/7/2019 to 28/2/2020)
@qldfrog settings used: $100K x 20 ($5k fixed PositionSize) Portfolio (Alphabetical Order)




3 year average period Backtest from (28/2/2017 to 28/2/2020)
@qldfrog settings used: $100K x 20 ($5k fixed PositionSize) Portfolio (Alphabetical Order)



Skate.

 

[Newt]

Many many thanks Skate for openly sharing these tests. Its incredibly helpful seeing a population of alternative strategies with subtle or large differences to those a trader has developed him/herself. For myself it certainly helps broaden perspective and confidence at all times, not just in the midst of the current panic selling. Thanks for Qldfrog too for stimulating so much useful discussion in recent weeks around backtesting benchmarking.

 

[peter2]

Hmm. . . I assume the backtest results are from one sample run rather than averages of many run throughs. If last weeks market fall produced the max DD (indicated by the one year results) then the results from the three year sampling should have similar max DDs since they include the period producing the max DD.

 

[Skate]

Hmm. . . I assume the backtest results are from one sample run rather than averages of many run throughs. If last weeks market fall produced the max DD (indicated by the one year results) then the results from the three year sampling should have similar max DDs since they include the period producing the max DD.

@peter2 by using (Hmm...) I'm at a loss what you are implying or what you are saying?

Your Statement
"If last weeks market fall produced the max DD (indicated by the one year results) then the results from the three year sampling should have similar max DDs since they include the period producing the max DD" - No this is not necessarily so, the mathematics used belies those assertions.

Clarification
When backtesting I don't use the "AmiBroker mtRandom function", meaning all my runs are fixed (single runs). The "AmiBroker mtRandom" function is used for Monte Carlo runs. Avoiding the use of the "mtRandom function" in a single backtest allows for consistency of results between strategies. Also it's important to comment that when a "Maximum or Minimum" is set for any metric doesn't translate to every backtest period used thereafter. Using this logic is a fallacy not a fact.

Keeping it in context
Also it should be noted that the maxDD of the last week is un-proportional in relation to maxDD for the previous 7 months. The 3 years of backtesting also include the results of this week & in the scheme of things didn't rate highly at all. A larger time frame for backtesting naturally smooths the maxDD because of the averaging, mathematics & entry timing used for the calculations. In the first run (8 months) the maxDD is concentrated in a few weeks of that period. FYI, the first backtest run is for this "Financial year" (8 months) not a "calendar year" (12 months) as you have indicated.

The look back period of Moving averages verses Exponential moving averages
Let me try to explain it to others. Selecting different look back periods using moving averages, the longer the period the smoother the average. Using the exponential moving averages formula for the same period gives more weight to the most recent period. It's the same with Backtesting - a longer Backtest period has a greater smoothing effect than a backtest of a short duration. One trading day doesn't break a system, similarly one week of trading is but a heartbeat in the scheme of things. Long term results of a strategy should be measured in years.

Summary
My previous post was to highlight "using different time frames" has a big bearing on the strategy results as indicated in the graphs above.

Skate.

 

[peter2]

Apologies if my post was unclear. "Hmm" indicates that there's something I don't understand or can't work out. ". . . " indicates that I've got to carefully think about how to express myself in order to get more information to resolve my conundrum.

A maximum drawdown (MaxDD) is the maximum observed loss (expressed as a percentage) from a peak to a trough of a portfolio, before a new peak is attained.

I understood that a MaxDD is a single point result that occurs sometime throughout the testing period. This value doesn't get modified in any way. It most certainly doesn't get smoothed. If the MaxDD is expressed as a percentage then yes it is proportional to the prior account equity high.

If there's a 14% DD in the results of the fin year (8 mths) back test, then I'd expect to see it reflected in the 3 year results unless a larger DD was present in the additional preceding time period.

 

[peter2]

Summary
My previous post was to highlight "using different time frames" has a big bearing on the strategy results as indicated in the graphs above.

I understand this. eg When starting the testing 8 months ago 20 new positions were required to populate the portfolio where as a test period started much earlier may only require one or two replacements at this time.

I'm making the assumption (that could be wrong) that last weeks market fall was so severe that ~95% of stocks were sold off significantly and all long only portfolios would be experiencing a 10-15% DD currently as shown by your fin year (8 mths) back testing results. I assume that portfolios started years earlier would also have been hit by the same % amount and I'm surprised that your results show that they haven't.

 

[Skate]

If there's a 14% DD in the results of the fin year (8 mths) back test, then I'd expect to see it reflected in the 3 year results unless a larger DD was present in the additional preceding time period.

I understand this. eg When starting the testing 8 months ago 20 new positions were required to populate the portfolio where as a test period started much earlier may only require one or two replacements at this time. I'm making the assumption (that could be wrong) that last weeks market fall was so severe that ~95% of stocks were sold off significantly and all long only portfolios would be experiencing a 10-15% DD currently as shown by your fin year (8 mths) back testing results. I assume that portfolios started years earlier would also have been hit by the same % amount and I'm surprised that your results show that they haven't.

@peter2 if all the positions were exactly the same in a backtest period your premise would be correct, different time frames result in different positions being taken at different times.

Backtest metric doesn't translate
The "Maximum or Minimum" metrics changes because of position selection, that's why the starting date of any strategy can be the decider if a strategy will be profitable or not, the Net Profit, AR% as well as maxDD changes & belies the expectations you have of "if it happened once it should be always reflected in all other results"

Skate.

 

[Saqeeb]

Time for me to post my weekly update on my MAP paper trading portfolio.
A wounderful timing to have started this portfolio as this will be a very good stress test for my system. As @Skate noted in his weekly reports "let's see if the strategy can recover". I can only wish that I had started my paper trading a week earlier, that way I would have been fully invested and my drawdown would have been a lot more than it is for the system to recover from.

BUYS:
No buys for this week

SELLS:
7 sells this week - PNI, BVS, VUK, BIN, EHL, NEU, GOR
This leaves my portfolio with just 4 open positions. Let's see what this week will bring!

 

[qldfrog]

. I can only wish that I had started my paper trading a week earlier, that way I would have been fully invested and my drawdown would have been a lot more than it is for the system to recover from

And you can really thank whatever God of your choice that you are dealing with paper only, I so wish i had not enough (real) open position to test my draw down

 

[rnr]

Hmm. . . I assume the backtest results are from one sample run rather than averages of many run throughs. If last weeks market fall produced the max DD (indicated by the one year results) then the results from the three year sampling should have similar max DDs since they include the period producing the max DD.

Hi @peter2,

Whilst I get where you are coming from in one sense, given the longer time frame, you need to allow for alternate trades being taken which may result in a earlier or later exit (resulting in a higher/lower profit or loss) and in the application of "position score" over a larger/smaller range of available stocks, if I understand the code applied, then open positions may exist which are totally different from those taken with a shorter time frame.
Hopefully the above comments make sense!

Cheers,
Rob

 

[Lone Wolf]

Hmm. . . I assume the backtest results are from one sample run rather than averages of many run throughs. If last weeks market fall produced the max DD (indicated by the one year results) then the results from the three year sampling should have similar max DDs since they include the period producing the max DD.

I believe the answer to your question might be in the position sizing settings.

@qldfrog settings used: $100K x 20 ($5k fixed PositionSize) Portfolio (Alphabetical Order)

If I understand correctly, the system will only ever invest a maximum of 100k in the form of 20 x 5K positions. No compounding.

The financial year backtest hit the recent drawdown period while still only having an account size of around 100K (fully invested). The 3 year test hit the recent drawdown period after doubling the account size (only 50% exposed)

Both tests show a similar loss in terms of dollars, but the % loss is different due to the size of the accounts.

Using fixed position size is misleading in terms of the real max drawdown you'll suffer assuming you intend to compound.

 

[Newt]

I believe the answer to your question might be in the position sizing settings.



If I understand correctly, the system will only ever invest a maximum of 100k in the form of 20 x 5K positions. No compounding.

The financial year backtest hit the recent drawdown period while still only having an account size of around 100K (fully invested). The 3 year test hit the recent drawdown period after doubling the account size (only 50% exposed)

Both tests show a similar loss in terms of dollars, but the % loss is different due to the size of the accounts.

Using fixed position size is misleading in terms of the real max drawdown you'll suffer assuming you intend to compound.

Hmmm, I rarely backtest in this way, but must admit to a mindset this will be "worse case", but as you rightly point out MaxDD will be too conservative....