Hi everyone! My live account is located under a different username, but no matter. First of all, I would like to thank @Quantessence for introducing me to Darwinex (the email you guys sent out in 2017 after migrating here). Moving on!
Before I start talking about Darwins, let us talk about diversification. The point of diversification is NOT to maximize the diversification factor shown in darwinex portfolio, but to actually DIVERSIFY your risk across assets so that when major event hits, only a portion of your money is affected. When I was first trying to allocate my money here on darwinex (thankfully only on demo) I initially got tunnel vision trying to actually maximize the shown diversification factor while forgetting to actually diversify. Then boom, brexit news actually shook a good ton of EA based Darwins that weren’t correlated, and that was a wake up call. Remember, while normally uncorrelated, many assets become correlated WHEN YOU WANT IT THE LEAST, meaning when shit hits the fan. So we also have to consciously, manually choose Darwins to the best of our ability so that their underlying assets and strategies are also uncorrelated.
Buy/Sell frequently or Hold?
Why does any trader/strategy ever open a trade? Because they think that the market is going to go their way. So if I were to be buying and selling darwins regularly (tyring to predict highs and lows of a darwin), I would have to be thinking that I KNOW BETTER THAN DARWIN PROVIDERS about the direction of their underlying future asset prices - and to be frankly, I am investing my money in Darwins because I DON’T KNOW BETTER. Of course there MUST be exit conditions, but chaos of the markets can send even golden strategies into drawdowns, and that does not mean the strategies suddenly stop being golden. Shortly speaking, I am going to be buying darwins and holding them, and more on the details later.
Loss averse Darwins
Loss aversion implies systematic behaviour of thinking that market is going to go your way when IT IS GOING IN THE OPPOSITE DIRECTION. I think such thinking MAY hold systematically true BUT ONLY in for short time periods. So any darwin with LA < 5, won’t be considered if average trade duration is > 10 hours. The shorter average trade duration for loss averse darwins, more forgiving I will be towards their loss aversion.
Now, I am a little biased (read: in love) towards component based anything. I am dividing my portfolio into 3 components. Let us assume that I have $30K to invest (My minimum per darwin rule is 3.33%, equating to $1000 in this scenario, meaning nice and round numbers). At first I wanted to set a hard rule that nobody gets more than 10% of my capital, since nobody is that special of a snowflake - but then I could not find enough good darwins so I had to break this rule. In the future, when I can, when there are enough excellent darwins, I will set hard limit of max 10% capital per darwin. My aim for diversification factor is 60%.
Component 1 - 1/3 of my money (33.3333%), $10,000
I like to call this component my cash cows, or best of the best. I am actually quite lucky that darwins in this component don’t clash with other parts of my portfolio - let me explain.
- 1.1 THA - $5000 (16…66%) - I assume everyone is familiar with this darwin. Return/Risk since inception is 12.50. It is insanely good. They trade a variety of currencies, but more importantly - THEY TRADE NEWS. It is not an automated strategy.
- 1.2 SYO - $5000 (16.66%) - again, I think everyone is familiar with SYO. They trade stocks and gold. They don’t trade that frequently. I am not sure of it is an automated portfolio, but there is nothing better to go here. SYO has almost 2 years of history, and an exceptional R/R since inception (4.53). Allocating this much to SYO is slightly out of my risk tolerance comfort zone. I would be much happier if SYO had longer track record, but eh.
Component 2 - 1/3 of my money (33.3333%), $10,000
This component contains automated strategies. I am actually quite satisfied with the darwins in this component.
- 2.1 LVS - $4000 (13.333%) - Very frequent, automated majors currency trading. Took a few hits with brexit news as can be read from their letters to investors, and probably might take more hits or recover with further brexit news - but they have solid track record. Performs well on markets with strong trends, and from what I observed, tries to enter markets early on in the trends.
- 2.2 FEG/GAF ($1500/$1500) $3000 (10%) - Same strategies with different parameters on different currency pairs. Not that frequent trading, and while they perform well on trending markets, they enter after initial crazy volatility. Should nicely balance LVS for fakeouts. Also by @finbou. Both have extensive track records.
- 2.3 ASY/SCQ ($2000/$1000) $3000 (10%) - ASY trades EUR/USD, and SCQ trades a good bunch of currencies. Both have sufficient EX and track record length, and historical returns. Both trade shortish durations.
Now, my component 2 trades currencies with automated strategies, and aren’t that correlated between themselves. Since automated strategies probably can’t adapt well to changes in market fundamentals, I tried not to assign more than a third of my capital to automated strategies (as I talked about in Diversification section above).
Component 3 - 1/3 of my money (33.3333%), $10,000
This component contains manual traders. Let me tell you, this was the hardest component to find darwins for. First half of component 3 contains darwins with LA > 5, and second half LA < 5.
- 3.1 CLA - $3000 (10%) - Read the interview. Mostly trades using statistical and mathematical methods. Trades currencies and semi frequently. Sufficiently extensive track record. R/R for last 2 years exceeds 2.33. Good return, despite current DD.
- 3.2 HFD - $1000 (3.33%) - Seems like a solid manual trader. High R/R score combined with high EX and 12 month+ track record.
- 3.3 DAQ - $1000 (3.33%) - Trades stocks, while not sufficient EX score, sufficient track period and high R/R.
- 3.4 ERQ - $3000 (10%) - Read the interview. Seems like a reliable Darwin. Low LA score historically, but trades last short period of time. Additionally, sufficiently proven track record of profitability.
- 3.5 HCP - $1000 (3.33%) - Low LA score, and trades lasted short amounts in the last 3 month. Again high R/R with sufficient track record length
- 3.6 MNW - $1000 (3.33%) - Low LA score again combined with < 10 hour trade length, and again sufficient R/R combined with EX and track length. I think this is manual trader, but I am not sure.
I wish nobody exhibited loss aversion, but could not find such darwins with sufficient track record length. Darwins I am keeping an eye on - SCS - if proves to be non-fake, DAQ + ERQ will be replaced with SCS ($4000). AZG, IFS are also worthy of keeping an eye on.
Right now, I am waiting on response to my request for API acceess, after which I will develop my own tool (of course, to be shared with the community here) to determine my exit conditions - but until then, I will exit a darwin at 20% drawdown, and that is it. When I leave a darwin, I consider what component it occupied in the portfolio, and find a suitable replacement for that COMPONENT among darwins, and reinvest. Of course, if I find a new darwin that is a better component than what I already hold - example, manual trader that is better than DAQ - then I will pull out of my existing component and replace it with the new darwin instead. Though I doubt I will be finding better darwins than what I already hold in short periods of time. This section of my post is very incomplete, and I will elaborate on it later when I actually get API access and finish writing my code. So, stay tuned if you have read this far and are interested?
How Did I choose my Current Darwins?
- For any darwin to be considered - Return 2Y > 20, RS > 5, Return/Drawdown ratio > 2
- IF LA < 5, Trade duration 3M < 10 hours and OS > 7.5 and CS > 7.5 ----> or LA > 5
For any darwin to be considered, it should beat at least 95% of random strategies per year on average. How do I calculate this, you might ask? Let X be the number of month in the darwins historical track record (I just count it manually), and let Y be the Return/Risk since inception value. You calculate Z = Y * 3.464 / squareroot(X). If Z > 1.645, then yes, it does defeat 95% random strategies on average per year since inception. If you want to set this percentage to something other than 95%, you can lookup z value for your chosen percentile. 2.33 is 99% FYI.
Example, at the time of writing this post, SYO has roughly 23 month history, and 4.51 R/R value. Z = 4.51 * 3.464 / sqrt(23) = 3.25. Since 3.25 is definitely bigger than 1.645, SYO meets this criteria. In fact, if we lookup percentile of z value 3.25 - NORM.S.DIST(3.25, TRUE) * 100 in excel - we get 99.94, meaning SYO beats 99.94% of random strategies on average per year, which is beyond excellent.
So I just went through all darwins that met these criteria, and resolving to invest no less than $1000 per darwin, selected the rest manually, giving special consideration to EX score and track record length.
I have much more to write about, and will get to that in the future. If you wish to offer criticism, please do! I sometimes get terrible tunnel vision, and perspective other than my own is most appreciated. Any questions and other comments are also most welcome.
Also, whoever in darwinex came up with the idea for pitting darwins against random strategies, and evaluating their performance percentile - namely PF IA and R/R since inception value - allow me to express my most sincere appreciation for your genius! Such a simple, brilliant, useful, descriptive and elegant metric that easily lends itself to interpretation! Kudos to you, sir/madam/group of people!
Edit: ERQ position size changed to $2000, freed $1000 remain uninvested.