Perhaps a new stat or some weight to the RS stat regarding a steady pattern of d-leverage increasing specifically when draw-down is occurring. What do you think?
It is 3 years that I ask for more weight to RS.
They gave a lot of weight to LA to penalyze martingales but unfortunatelly LA has a very short lookback and goes up and down like a crazy.
RS is much more efficent.
Also here in the community LA is more popular than RS…
Today you are noticing the same things I noticed 3 years ago.
Don’t you think RS penalize too much slight Risk changes? In my opinion if RS would be ruled by absolute amounts of risk changes, it would be fine to give it more weight. But they penalize Risk changes percentage.
I.E: if a darwin move between 5% and 10%, the change is only a 5%, but when you consider it as a percentage it is 100% difference if increase or 50% if it decreases. While another darwin moves between 30% and 40%, and it is less penalized because even the gap is 10% (double than previous example), the relative move it is only 33% if it increases, or 25% if it decreases.
So, in the first example the RS attribute would be very low, while in the second it’d be medium or high.
Yes I do actually! I think that maybe a Darwins with changes in D-Leverage that seem more random and could be from a variety of reasons and do not correspond to any DD pattern should be penalized LESS and a pattern of D-Leverage increasing specifically during DD especially if there is a pattern of D-Leverage increasing even more as DD gets worse should be penalized (a sign of a martingale like risk management strategy) significantly more.
Or maybe a separate flag or warning when such a pattern is detected.
From 5 to 10 is not slight, it is a factor of 2.
For me a guy working from 30 to 40 ( for example ULI ) is fine as a guy working from 3 to 4 .
RS is not designed to penalyze high risk but random risk.
As investor I have nothing against high risk, as long as you dont’ blow the account.
Yes maybe instead of using LA to catch martingales, which also have a lot of false positives on excellent traders who simply operate with a wider SL for instance, we could bake in to RS or even a new stat to detect martingale patterns based on D-Leverage vs DD patterns.
Low LA is bad only if confirmed by low RS .
Clearly the final confirmation is analyzing raw data on Trading Journal.
Sometimes I feel you don’t read my comments, I gave you a complete explanation with two examples.
How is how risk trading from 5% to 10%?
Isn’t it higher risk from 30% to 40%?
I am not saying whether if I like high risk or not, I am just talking about the way RS works.
5 to 10 deserves a worse score than 30 to 40 and I agree with it.
RS in the only algo I would not change.
I have to totally disagree with your statement.
Yes that’s how it is now but I’m trying to think of a way to use an algorithm to find patterns of D-Leverage increasing during DD to alert investors somehow and maybe reduce a score and increase (or don’t decrease as much) scores for risk swings not corresponding to a DD pattern.
Yes, I agree with that. RS should vary more when it happened and less when a trader change the risk between reasonable parameters without running the darwin with high risk, like making one trade with 4% and next with 9%…both are very reasonable risks and don’t put the account in danger at all. While for example running ten trades at 3% to recover a DD is totally different.
What you say is true and is called statistical ruin and with a VAR above 30% it is very likely to blow because of a bad streak but this is a different concept than risk stability.
Yes you’re right, it is. So perhaps a new stat then to show a score or flag for a pattern of D-Leverage increasing specifically during DD?
If you want to penalyze behaviours like the recent on HFD in my opinion RS is already doing it.
RS 5 to 6 is suboptimal , the mistake is the low weight that allows DScore to be above 70 .
The same happens also with GFA .
My thought is whatever is being done now could possibly be improved with an algorithm to detect D-Leverage correlation with DD.
I dont’ think they will design new algos
The only new algo designed in 4 years is MC but they changed weights more times so a higher weight to RS is a more realistic expectation.
Let’s take an example here, IRY is a young darwin which its maximum Risk is 7.62% and the lowest is 3.00%.
In my opinion his risk is stable, under control and also is not high, however his RS attribute is only 5.4.
So, I think the RS parameters should have a couple of tweaks here and there, otherwise it is one more unuseful attribute.
Of course I am contrarian to increase the weight of RS at least the parameters to calculate it don’t change.
That is only because of the short trackrecord, RS already recovered from 2.5 to 5.5 in 3 months and will be above 8 in few months.
I don’t think so, if he keeps the 3.00%-7.62% oscilation, the RS will stay low.