CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage. 66 % of retail investor accounts lose money when trading CFDs with this provider. You should consider whether you understand how CFDs work and whether you can afford to take the high risk of losing your money.

What is wrong with my portfolio allocation philosophy?

I choose according to best D-score over 70 and good MAR over 2. I buy and hold.

And after 3 months the results are disappointing:

Is there a way darwins can game the system? They look great before but then they lose money?

History of portfolio rotation.

Current portfolio backtest:

https://www.darwinex.com/darwins-backtest?startDate=1483225200000&investment=10000&leverage=1&activeProducts=ASY.4.6,BUX.4.4,FSK.4.12,GAF.4.16,GPN.4.10,HFD.4.17,HLA.4.8,JMC.4.1,MNW.4.13,OVL.4.17,SCS.4.5,SKJ.4.22,STV.4.12,TER.4.15,TGB.4.17,VVC.4.8,YZF.4.16,ZXW.4.5&share=1

4 Likes

DScore is quite volatile and tipically floats from 55 to 80 so if you buy above 70 often you are buying after a good quarter and before a bad quarter.
A quarter is not significant, neither for a single darwin, neither for a wide portfolio.

That could be a very good portfolio but it needs more time to develop profit.

Why did you sell SCQ, a DD of 18% is enough for you to sell a darwin?
Is this your definition of “buy and hold”?
NTI has a max DD of 20% , also YZZ .

8 Likes

I sell darwin, when it attains largest drawdown in his history. Since we are talking about amateur/semi pro traders, it could mean some difficulties in their personal life, health issues etc. SCQ had worst DDs in its history.

Look at this image and you would understand:

Thanks for the tips!

O maybe amateur investors considering max DD on short trackrecords and with too high expectations buying high and selling low. :slight_smile:

I think for MAR you mean (annualized return)/( max DD)
If this rate is never high for long trackrecors it means that it isn’t a realistic expectation.

BTW no personal offense, your results for 2018Q4 are better than mine as investor, sometimes there is just no way to win.

3 Likes

hi

maybe the problem could be fees and turnover

I did a backtest as well from 2017 until now.

maybe the picture is not clear enough, so I will write some of the details:
The backtest has a return of 16.6%, P&L 2158 Eur, fees 498 final equity 11659
if you average the returns of the darwins : 62.35%,52.46%,33.45%,26.5%,13.59%,-5.69% the average is 30,44%
so getting a return of 16,6% means you are only receiving 54% of the total returns of the darwins

the P&L before fees, you may think should be 30,44% but is only 21,58 % (2158/10000), the reason this happens, is that the invested capital is not compounding fully, it is reduced by the fees, (one of the reasons investing in stocks beats other asset classes is that stocks compound faster due to reinvestment of dividends and this is equivalent to a “negative dividend”, it slowdowns compounding)

and then of course the fees themselves are 23% of the profits (498/2158)

and this portfolio has no turnover, buying in and out makes things even worse.

please find attached the back test report

https://www.darwinex.com/darwins-backtest?startDate=1483225200000&investment=10000&leverage=1&activeProducts=OVL.4.17,TMB.4.5,JMC.4.1,MNW.4.13,ZXW.4.5,NTI.4.12&share=1

1 Like

Hmm, from my understanding this shouldn’t be quite so. Fees are not deducted from invested money OR from open PL, they are deducted from balance, which CAN go into negative from what I recall (negative is covered whenever money is freed) -> Thus compounding should not be affected, as long as buy and hold is true.

Edit: I think you are forgetting about divergence when calculating what SHOULD be the return.
Look at THIS -> This serious deviation from darwin return is caused by divergence, not fees.

3 Likes

Hi @rickinvestor,

Another reason could be that not all darwins start at same point, my 2cent…

EDIT:

I am talking about this

1 Like

Here the explaination is much more simple.
A backtest of the survivors is always profitable.
You sell darwins in DD because you dont’ like it and you realize many losses, you shoud backtest the initial portfolio not the survivors.

3 Likes

hi yhlasx

yes thanks for correcting me ! I think you are right!

makes sense since the fees are paid from the balance and not by reducing the positions it should not impact compounding so divergence is a better explanation

2 Likes

hi naturalcapital

In the backtest all the returns start at zero

that is a good point:

I hope that when I run a backtest from 2017 and I filter on a given set of attributes it gives me the darwins that at that time met the requirements and those should include darwins that later on went bust , so there is no survivorship bias when doing backtests that is key of course

I must agree with @yhlasx, that seems to be about divergence.

If i am wrong correct me please, but as far as I know, you get the today’s ones.

3 Likes

I don’t know the details myself of how the backtest tool was build, but if the backtest filters are based on today’s values and not on values at the start of the backtest, then the backtest tool is useless and misleading…

maybe with the API it will be possible to perform backtests free of survivorship bias

I don’t understand how you mean you filter on the past data? Backtesting tool does not include filtering at all right now! You would have to separately filter somehow on the past data, but the backtests should be accurate.

here is how I select the filters in the backtest:

so I say select the filter "most investors " and “good scores” and I tick box the first 5 , I would expect that those were the ones that 1st of Jan 2017 had the “most investors” and “good scores” not the ones that meet that criteria today

1 Like

Filters are just a way to pick the darwins, aren’t executed/rotated every day.
The backtester is just combining the return chart of the darwins.

1 Like

Oh no, those filters show current data. Here is a haphazard filtering tool I made with which you can filter past data, but it is not easy to use.

2 Likes

thanks ! I will have a look at your tool :slight_smile: I will also focus my backtest efforts on the API history data