CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage. -- % 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.

Covid-19 highlighting potential problems of the VaR system?

Hello community,
I want to report results of my research and share my opinion about it.

My thesis is that the current market conditions are highlighting a potential problem in the VaR system.
Let’s start with the research and study I made on the current variable VaR system:

Date Investor’s VaR My VaR Ratio Darwinex Ratio
04/03/2020 5.14 3.19 1.61 1.62
05/03/2020 5.58 3.48 1.60 1.62
06/03/2020 5.58 3.48 1.60 1.62
07/03/2020 5.58 3.48 1.60 1.62
08/03/2020 5.58 3.48 1.60 1.62
09/03/2020 5.53 3.45 1.60 1.62
10/03/2020 5.72 3.45 1.66 1.62
11/03/2020 5.87 3.67 1.60 1.62
12/03/2020 5.91 3.7 1.60 1.62
13/03/2020 5.88 3.68 1.60 1.62
14/03/2020 5.88 3.68 1.60 1.62
15/03/2020 5.88 3.68 1.60 1.62
16/03/2020 6.02 3.77 1.60 1.62
17/03/2020 6.22 3.87 1.61 1.62
18/03/2020 6.57 4.07 1.61 1.62
19/03/2020 7.13 4.44 1.61 1.62
20/03/2020 7.29 4.55 1.60 1.62
21/03/2020 7.29 4.55 1.60 1.62
22/03/2020 7.29 4.55 1.60 1.62
23/03/2020 8.05 4.93 1.63 1.62
24/03/2020 8.73 5.33 1.64 1.62
25/03/2020 8.85 5.41 1.64 1.62
26/03/2020 9.37 5.78 1.62 1.62

From the results of March, you can appreciate that even if the underlying strategy’s VaR is changing, investors’ VaR is moving in order to keep the same ratio (in my case, 1.62).
This is a huge improvement, in my opinion. It erases the problem of having different risks in different periods, since the ratio seems to be quite stable over time, allowing investors’ to follow the same curve of the underlying, although with a different risk.

Darwinex deserves double thumbs up for this :+1: :+1:

This period of high volatility, though, is highlighting a potential serious problem of the system.
When the personal VaR increases over a certain point (my point is 6.17%), the ratio cannot be the same anymore, because investors’ VaR is capped at 10%.
So with a personal VaR going towards 10%, the ratio becomes smaller and this changes the behavior of the Darwin.
There are two potential outcomes:

  1. The trader keeps following the system. VaR increases during this period of high volatility, the ratio becomes smaller, this decreases the risk for investors, while keeping the same risk for the trader;

  2. The trader reduces the risk, this will reduce the VaR and the risk for both, trader and investor, but will keep the ratio constant. Results for investors is the same as outcome 1, they will follow the Darwin with a lower risk.

What when this crazy volatility will end? VaR will go down again. Investors will go back to the normal risk.
So, right now, with high volatility, there is no way to avoid a reduced risk for investors. When this period is over, VaR will go down and risk for investors will increase again, going back to normal, to the situation before this period.
The problem is that this leaves a lot of space for luck and not for trading skills.
What if the Darwin has a good period during the time VaR increases (less risk and less profits for investors) and then a bad period when VaR goes back to normal (normal risk and normal profits for investors)? The Darwin would lose money just because of the VaR, while the underlying would probably make money or break even.
Opposite, some Darwins can be lucky and have a bad period now that VaR is increasing (reducing the risk and profits for investors), then have a good period when the VaR goes back to normal (normal risk and profits for investors).

This is not only affecting my Darwin, but a quick look at top Darwins’ VaR, and we can immediately see what I’m talking about.

HFD

SYO

LVS

CIS

Right now, VaR is going up for most Darwins and ratio starts to decrease at a certain point, changing the normal proportion between the underlying and the Darwin.
Not only this, for Darwins with personal VaR above 10%, ratio starts to decrease immediately and dramatically. Fortunately, it is not my case.

My humble opinion is that this period proved that VaR calculated on 45 trading days is too short and can lead to tremendous changes like the one experiencing now, even if traders are keeping the same risk per trade.
This is a post just to show what I think, supported by some data. Then, obviously, it’s up to Darwinex to decide if it is the case to consider a longer period than 45 trading days for the VaR.

9 Likes

How many more trading days do you suggest? What are the drawbacks of a longer term calculation?

2 Likes

The problem is that when you have an abnormal parameter getting in the statistical calculation, the shorter the period, the worse the effect.
Right now, the calculation of the VaR is removing data of 45 trading days ago, when we had fairly low volatility, to make space for new trading days, in which we have huge volatility.
On a period of “only” 45 days, this has a strong impact.

It’s a bit like choosing the period for an indicator in your trading strategy.
I remember CavaliereVerde choosing 5 for the RSI.

Why is the RSI going down if the price is not going down? It is just because after 5 candles, the RSI calculation is leaving that huge green candlestick to make space to new candlesticks.

The opposite is happening right now with the VaR.
Exactly what happened to the RSI in the example, will happen to the VaR when market will go back to normal. When the 45 days will leave behind this huge volatility in the calculation, the effect on the VaR will be huge and we will have VaR dropping by a significant amount every single day.

I believe that Darwinex set 45 trading days to avoid to have a slow reaction by the VaR, that could damage investors in case a trader has a very poor risk management, increasing the risk in bad periods (martingales, for example).
It makes sense, but it damages good traders that are only trying to do their jobs.

I think Darwinex has a team that is more qualified than me to run some tests on it.
I would test anything from 60 to 100, if it was my call.

3 Likes

I am here since the beginning.
Before 2017 var was calculated on 15 days and it was even worse.

The point is that var and risk manager were designed to deal with gamblers and martingales, it has to be sensitive but not too much.

If you extend the lookback you reduce the penalty for good traders but you reduce the effectiveness dealing with gamblers and short term overtrading.

I agree on extending the lookback but monitoring only solid darwins is not enough.
Lookback is always a compromise between significance and sensitivity.
In general I would extend the lookback of many algorithms: Performance, Loss Aversion, Correlation…

4 Likes

I understand and I agree, that is also why I thought of something in a range 60-100, instead of a radical change.

The thing is that you also reduce the penalty for good investors, who carefully chose good Darwins, reducing the effectiveness on bad investors, who chose gamblers.

I agree that Darwinex has to do something to protect investors from gamblers, but, in my opinion, it also has to do something to allow investors to follow good traders with the best conditions, without being penalized too much only because Darwinex has to protect bad investors.

1 Like
1 Like

Right, I guess your OP table could show a 5th column like actual risk of trades that day so we can also see how quickly or how slowly the VAR calculations are reacting to the actual underlying changes.

I’m curious about what is the argument for a trader not adjusting the underlying strategy to the current volatility (reducing risk and keeping the underlying VAR stable)… IMO it’s a traders responsibility to do this. If she doesn’t do this then it seems like the underlying strategy is also suffering from a similar “luck-prone” syndrome during volatility (when vol first spikes, trader could have a few lucky trades and zoom or a few unlucky trades and poof) and so too does the Darwin follow this (EDIT: to some degree) on the first days the volatility spikes up until the VAR calculations catch up to the larger volatility.

I agree with that statement in an average situation. However, with the current volatility some traders are widening the SL and TP, otherwise many SL that usually work well on “regular conditions” would be hit. In the other hand, TP are also being wider, so we can see bigger gains too.

Adjusting to a totally new and sudden situation is not easy, maybe the solution would be stop trading for a while, but it is also a taught decision to make.

Personally, I trade here on Darwinex with a system that opens a trade every time it detects that there is a movement of “x” pips in favor of the trend.
Risk per trade is the same, but, obviously, we have many movements in the order of “x” in this period, since the market moves 50 pips even in a couple of minutes recently.
I trade GBP/USD on this system and it went through periods like Brexit referendum in June 2016, flash crash in October 2016, UK elections in December 2019. All periods with high volatility that gave me more trades than usual.
Maximum Drawdown reached has been -6.30% (with the underlying).
I had a tough period at the beginning of this period of high volatility. Maximum drawdown has been -3.68%.
I know what I’m doing, I think Darwinex should trust me and other traders, giving us the right to decide for our own strategy, for the sake of our investors.
I appreciate Darwinex trying to protect investors from hazardous behavior by traders, but this extreme case is proving that you also penalize traders that are simply following their trading strategies and systems and are making money for investors.
If this period of strong volatility continues, my VaR will easily go to 20% or more, unless I change something in my system. I believe it is the same for some other traders. HFD is already paying the price of it if you compare his Darwin with the underlying.
In my opinion, here the deal is to recognize that risk management is an important part of a trading strategy and the trader should have more freedom, a wider range to move within.

Like it or not, now, I’m forced to lower my risk, either with the VaR increasing or reducing the risk on the underlying. I’m in a situation in which if I have a bad period, I’m lucky to trade with a lower risk, if I’m in a good period, I miss a good part of the profit.
I don’t even know what to hope for… I wouldn’t be happy to have a great period while I have to cut my market exposure because of the VaR.

1 Like

I’ve mentioned before about the pros and cons of manual and automatic trading regarding flexibility and adapting to new market scenarios.

Regarding your trading set up, is it possible (ans easy to set up) a limit of the amount of trades per day/week/month regardless the amount of signals? Without changing anything else, so basically your system would remain same. That way you could control the VAR

I nmy opinion you are overestimanting the interventions of the risk manager on your darwin IRY .
If you look to Risk Adjustment chart there are very few orange dots so you are worrying about nothing.
BTW the performance of the darwin will be always slightly worse than the underlying strategy
You can trade 10 times on week 1 and 2 times on week 2 and there is no penalty.

1 Like

The thing is that I don’t feel comfortable doing that.
Imagine that you are playing a game in which you win $10, 6 times out of 10. You lose $10, 4 times out of 10.
You play against a guy (the market), who challenges you twice a day. Now, suddenly this guy challenges you 20 times per day, would you say “no, I have to put a limit to this game, I want to play maximum 10 times per day?”.

This is the reason I don’t feel comfortable. IRY is at +12.45% during March. Best month since the adventure on Darwinex and second best month since 2013.
Why should I put a limit to my trades if the system is detecting a higher number of opportunities?

I know that I cannot do something about it and Darwinex is not going to change this tomorrow, so I’m forced anyway to proceed through the path of lowering the risk.
The intention of sharing this in this thread is to give food for thought to Darwinex for the future. Because this can happen again and I think it can be handled better by working on the VaR system.

No, it’s not about the Risk Adjustment. I know that the Risk Manager will not intervene. It is the ratio that will go down, from 1.62.
After my VaR goes above 6.17%, investors VaR will go to 10% and will not increase anymore to keep the same ratio of 1.62.
This means that my VaR will continue to go up, while investors’ VaR will be stable at 10%, that will slowly bring the ratio down from 1.62 to 1.
After my underlying goes above 10% VaR, the ratio will be even smaller than 1.

So, if I have a profit of 1%, today my investors have +1.62%.
If I have a profit of 1% tomorrow, trading with the same risk and with the same system, my investors will get 1.60% tomorrow, then 1.55% in 2 days, then 1.40% in a week, then 1% in a month, then 0.90% in a month and a half… and so on like this.
This applies also to losses, of course.

My concern is not about the Risk Manager, I believe none of my trades will be adjusted. It is that the Darwin will soon follow a different behavior because the ratio will not be the same and luck will be soon a factor for it.
Good period during high ratio and bad during low ratio = lucky
Bad period during high ratio and good during low ratio = unlucky

Stable ratio like in the previous months = room for skills and not luck

To be clear, this has nothing to do with the underlying, I’m talking about the Darwin for investors.

2 Likes

Ok, what about trading half of size amount per trade? Your risk per trade would be half, and you’d take all the trading opportunities without changing anything. Maybe you can do that when trades per day are higher than “x” times…otherwise (if you don’t do anything) your VAR will fluctuate at least Darwinex make a tweak…these are just ideas

1 Like

Yes, it is the path I decided to take.
Still, now I’m trading with half risk. If I win, it means I’m missing half of the profit… I don’t know if I have to hope to win next trades or lose…

3 Likes

Anyway, thanks everyone for intervening.
I think no one was expecting a situation like this, neither Darwinex nor the traders.
I hope that data, facts and opinions shared in this thread, will give everyone (Darwinex, traders and investors) a cue for improving in the future.

Thanks and best of luck for your trading :slight_smile:

2 Likes

To clarify I’m also talking about reducing the lot size when the SL and TP are enlarged, thus keeping the VAR per trade more stable and in turn, as long as frequency remains the same, the monthly VAR too. It’s fairly easy to implement in to an algo but a little trickier for manual traders maybe.

I see your point, a fluctuating trade frequency also needs to be absorbed. A caveat of pinning everything to a monthly VAR instead of something like a VAR per trade.

But also:
good period during a high vol / high risk time before the VAR / ratio had time to catch up = lucky.
bad period during high vol/risk time before the VAR / ratio catches up = unlucky.
And these times would be negatively effected by a longer term VAR calc. Although these are mitigated by a cap they still exist to some degree.

2 Likes

Yes, I agree with that, on fact I mentioned it to FedericoSellitti as a solution to his VAR fluctuations, but I don’t know how difficult is making those changes to automatic strategies.

Agree, but having a VaR calculated on a longer period or introducing a larger range like 5-20%, would keep the ratio stable for longer or for good.
With a variable VaR with the superior limit at 20%, I would not have any problem. I could reach a personal VaR of 12.35% and still keep the same ratio between my account and my Darwin.
Also, even if my VaR went above 12.35%, I would certainly have more time to act. My VaR now went already from 3.87% on the 17th, to 5.78% today.
I have till 6.17% before the ratio goes down and I strongly believe that halving my risk will not stop the VaR going above 6.17% during next week. So what am I supposed to do? Risk 1/4 of the normal risk management of the system? I might as well stop trading if I have to see my profit cut by 1/4 :slight_smile:

I trade a semi-automated system, so it’s a bunch of rules that I apply manually, so it’s not hard to change it.
It is hard psychologically to have to lower your risk like this and see that on another broker you are making 10%, here investors are making 5% because of what is happening to the VaR.
What should I say to my investor? You got half of the profit because of the VaR?
I might also be lucky and have a bad period after my decision to halve the risk, so investors would lose less than usual, but I prefer to keep luck at the minimum in this business.
If I lose, I lose. Losses are part of trading.

This is my opinion.

You missed the time we had a 20% VAR limit…
I used to like it more (I don’t like limitations at all, but this is other story).
Maybe after these convulse times you can make a poll to see if more people in the community will support that, but I can anticipate that it would be a waste of time.
Anyway, and regardless the limit of the VAR, the VAR fluctuation is only on you, you can’t blame Darwinex or the VAR…it is directly linked to your trading behavior…more trades, then reduce the lot size…, less trades, do the opposite…less volatility and less pips at risk per trade, then increase lot size or trade more, and so on…it relays on your trading always.
You can’t pretend to take only the good side, there are two sides (that’s why I prefer not limit VAR at all, so you as trader decide what to do).

1 Like