We receive regular feedback pointing out that the outcomes of the Darwinia contest may not always foster our goal of democratising the markets.
For example, in response to last week’s post, Steven, mentioned:
“have been looking at the Darwins and the ranking. I spotted some traders that traded one of his/her strategies to 98% drawdown, however their respective D-score/ranking were not penalised. I wish one day the ranking algorithm will take into account for fairer outcome. After all I am sure your company do not encourage reckless risk management.
Clearly, we share Steven’s concern – who likes reckless risk management – but as you’ll see, when you decide to do something one way, you have to stick with your decision, even if parts of it hurt.
This post will touch upon several aspects, including the technical details on the ranking, trader incentives, and how to solve this going forward. We sincerely look for input on this one – perhaps with your feedback we’ll find a better way!
Overrule the algorithms?
Darwinia is ranked weighting:
- Grades for the DARWIN based on the strategy,
- The Level as determined by those grades
- DARWIN performance for the natural month and
- A measure of activity (to make sure that high level DARWINs win without trading activity don’t contend whilst on holidays).
We’ve covered the mechanics of Darwinia levels elsewhere – the relevant point here is the fact that the DARWIN’s underlying strategy does NOT enter the equation. Which implicitly means that neither do its Drawdown, VaR level, etc.
Arguably relevant criteria like “he’s trading the strategy with very high VaR”, or “the trader traded his strategy to 98% drawdown” aren’t even evaluated. On account of this “shortcoming”, sometimes strategies with “reckless risk management” make the end of month cut for the 20 strategies that are awarded Darwinia capital for 6 months.
Shouldn’t we overrule the algos, draw the joker wildcard and introduce rules so that DARWINs with underlying strategy VaRs higher than X or Y be disqualified from the contest etc. ?
Tempting! For starters, it may well save us some investing losses. But would that really help the underlying goal, in the long run?
We’ve thought this through, and think there’s a case for laissez faire (at least for now). If it proves too costly we may have to reverse course, but it’s important to share the pros of laissez faire.
Arguments for laissez-faire
The first is transparency. Not tinkering with the rules.
If we’re the DARWIN exchange listing DARWINs – which are genuinely different from their underlying strategy – and control risk so that DARWINs run on 20% VaR, why introduce the underlying strategy into the investor equation? Doesn’t this send conflicting messages? The last thing we want is anyone accusing us of a non-transparent process.
The second is not judging people’s incentives. There’s enough gambling going on in the markets, and we’re clearly not here to encourage more. Then again, having spoken to hundreds of traders in the last few years, we realised there’s a (possibly small) subset of “rational gamblers”. Their argument is: I only manage e.g. 1000 USD, and I won’t be able to manage a meaningful amount for the foreseeable future. So I trade several promising strategies with deliberately outrageous risk. It’s my personal way to build a lottery ticket cheaper than anything offered in the gambling market.
Not everyone will agree, but provided the underlying strategies work, there’s logic to it. Indeed (some) DARWINs based on high risk strategies get good grades for long-ish periods of time – compare DARWIN1, 2 and 3 with their underlying strategies. This begs two options. Option A is: the rating algorithms work, and the DARWIN based on the reckless gamble, with “investment grade” VaR, are investable for a while – in which case who are we to hide them from investors? Option B is: the rating algorithms need improvement, and we better get our act together to improve.
We welcome both. Simply “hiding” flaws in the algorithms or high risk in the closet would hurt the transparency of traders’ evolution, and paradoxically reduce everyone’s credibility.
The third is trust in market forces. Underlying strategies will always be visible to investors. Once hundreds of millions in Assets under Management (AuM) back DARWINs, presumably few investors will choose high risk traders, reckless risk however mitigated by the DARWIN, will put them off. Presumably even risk loving DARWIN providers will lower their VaR then… but we’re not there yet – our 0.5 MM capital allocation / month isn’t enough of an incentive for them to lower risk.
And last, but not least, there’s evolution. As AuM grow, Darwinex will both groom better traders in house (focussing on improving pays in the long run) and attract better traders who don’t know / can’t currently be bothered to join Darwinex. Nothing better to crowd out (rational or irrational) punters making the Darwinia cut every month, than higher density of good traders providing investable DARWINs.
When that happens, nothing would please us more than knowing that evolution worked its way through the ranking, without tinkering. Until then, the reality is that 90% of independent traders lose their shirts owing to nothing more than excessive risk taking. The only way to fix a problem is to acknowledge it – and that’s where we are right now.
Clearly, we couldn’t discuss this with users in length… so what’s your thought? Do you agree? Any argument we’re missing?