Normalisation of DARWIN VAR is one of the cornerstones of Darwinex Risk Management System. Darwinex tech team did great work. However, I think that there might be some changes that can be done to better control the risk.
First, I think the VARs of underlying strategies are under-estimated. A simple filtering finds that the number of DARWINs with Experience Score of 10 is 608. Among those DARWINs, 409 of them have drawdown of more than 20% during last 2 years. That is 67.3%. Do not forget that some DARWINs were closed because of high drawdowns. As many members know, DARWINs at 10% VAR have 5% chance of losing more than 10%, which means 10% or above loss occurs once in 20 months. But the probability of the loss extending over 20% for a DARWIN should be very low in theory. The probable reason is that the VAR of the underlying strategy is under-estimated, which lead to a higher multiplier when the DARWIN copies the trade. The textbook formula for VAR estimation has some flaws (I remeber @OldSchoolPT mentioned the flaws some time ago). Not all trades are born equal. For example, in a slow market, a trader may find a good (but not great) setup. The trader only wants to trade a small amount to try to make a small gain because the probability of success is not very high. But in a volatile market, the trader may get a great setup and the stop loss is not very wide, then he decides to place a very large trade. Obviously, the latter trade could possibly cause much more loss than the former one. They should not be given equal weights in VAR estimation (suppose the durations of the two trades are similar). A possible solution is that we rank all positions (in the last 45 days window) with D-leverage from high to low, only top 50% (50% is just a guess, simulations needed to decide the most suitable percentage) positions are used in VAR estimation of the underlying strategy. In this way, the VAR will also update much faster if the underlying strategy starts to use higher D-leverage.
Second, the highest D-leverage allowed for a DARWIN can be halved from 15 to 7.5. The markets have changed a lot this year. A sudden headline or a tweet from Trump could easily send price 100 pips higher or lower in one minute. It happens very often these days. If the maximum D-leverage is 15, then only about 67 pips loss (one or two trades) corresponds to 10% loss. A losing streak is common for almost every trader. DARWIN could easily encounter 15% or above drawdown.