Introducing DWC, the hedging DARWIN

18 June 2017
The Market Owl

Introducing DWC, the hedging DARWIN

DWC, the hedging DARWIN is DarwinexLabs first creation. It’s already listed at the Exchange.

This post lays out:

  1. The rationale for a hedging DARWIN?
  2. Lessons learnt in the development process
  3. How DWC revenues will be shared with the community

As you’ll gather, actively leveraging community data is a major strategic milestone for Darwinex. We’d be extremely grateful for your thoughts on the decisions made!

Why create a hedging DARWIN?

Because there are periods when markets become abnormally “tough” and sweep most traders.

DWC leverages community behaviour indicators to protect investors in those periods. It offers 2 use cases for DARWIN managers and investors:

    1. Hedging “deep” and “tough” tail market moves where DARWIN portfolio diversification breaks down – similarly uncorrelated DARWINs end up correlated
    2. Market sentiment based on real time exposure by the Darwinex trader collective – without compromising individual trader IP

NB: standalone investment is NOT a use-case. DWC is NOT designed for passive returns.

Back since launching the brokerage offering (May 2014), we’ve observed what appear to be recurring “market cycles” that last anywhere between 4-6 months.

These cycles typically involve:

  1. A 3-4 month “benign” cycle where DARWIN portfolios perform well
  2. A 1-2 month “sweep” cycle where portfolios struggle – without obvious options to diversify the sweeps away

All of which makes a hedging strategy a valuable addition to standard DARWIN portfolios…

What explains market sweeps?

Analysis of community behaviour suggests that the Darwinex trader collective suffers loss aversion – as highlighted by La (Loss Aversion attribute) grades below 5. It also highlights market conditions that “sweep” (literally) low La rated strategies trading longer than 1 day timeframes.

Loss aversion is a well documented cognitive bias. This gives us comfort in extrapolating the behaviour of our community to the global community. i.e. we’re onto something systematic – and therefore “predictable”. Loss aversion as a standalone reality might be a major driving factor behind the BOOM in the forex industry in recent years, and is certainly an explanation of CFDs in their “cash for dealers” impopular description.

(As a matter of fact, it probably explains why retail traders are offered top of book spreads well below the spreads available to institutional participants!)

How does the DWC hedging DARWIN work?

DWC trades whenever normally “symmetric” community sentiment (everyone randomly on either side of a currency) turns overly “asymmetric” in one or several pairs. In line with its hedging nature, DWC then trades the opposite way as a liquidity taker – e.g. DWC trades are routed to the market just like every other ordinary DARWIN – Darwinex does NOT deal against customers. DWC doesn’t enjoy a particularly high success rate, BUT it does make significantly more on winning trades than it loses on losing ones.

Designing DWC boiled down to efficiency: replicate (the inverse of) community positioning with the least transaction costs – remember, we’re not dealing, but paying the market to hedge community exposure. Whilst we won’t give away the key secret sauce , both backtest and live testing suggest we’ve reached a good compromise – possibly because DWC only works for FX pairs, where all the community “votes” on a few select currencies.

DWC has traded live funds since shortly before the GBP flash-crash (which it anticipated and profited from :-). Both backtest and live behaviour have met ingoing requirements, so with the announcement of DarwinexLabs, it’s time for Exchange listing!

 What have we learnt?

DWC has been a rewarding research project, with the following core takeaways:

  1. DWC’s La (Loss Aversion) grade is 9.2 – which makes DWC yin to the community’s loss averse yang
  2. DWC sports a Cp (Capacity) grade of 10 – we believe it could well reach 500+ million AuM, e.g. there’ll be plenty of supply
  3. Luckily (or perhaps not?) it nailed both of the latest extreme market sweeps (SNB CHF move and GBP Flash-Crash)
  4. It’s a valuable leading indicator – watch out for DWC holding few trades as this could anticipate a swing in the market
  5. It’s a procyclical strategy – which explains its poor timing (Os & Cs grades)

Whilst we won’t disclose DWC´s live trades, its equity curve is valuable assistance in spotting “benign” vs. “tough” market conditions. Because it struggles when the community profits, it could be used as an indicator when:

  1. Approaching the end of it’s 3-4 down cycle could be a harbinger of “toughening” to come (winter is coming 🙂
  2. Correlating negatively with any DARWIN (starting by one’s own) might be worth some consideration!

 Who owns DWC?

That’s the million USD question!

For “Regular” DARWINs Darwinex collects a 20% success fee which is then then paid out to its provider. Given DWC’s provider is DarwinexLabs (i.e. Darwinex’s Quant team), should Darwinex pocket 20% success fee? Surely this is a conflict of interest given DWC builds on community data? Should DarwinexLabs waive a success fee?

As you can see, this is but one of several interesting questions.

The first decision made is for DWC to collect 20% success fee – just like any other DARWIN. Waiving the success fee would amount to “dumping” on regular providers.

The second, and perhaps more controversial one, is for DarwinexLabs NOT to retain any share in success fees collected for its DARWINs (including DWC and follow-up strategies). These will be collected, then shared with the DARWIN provider community, on merit.

We are internally debating on the best allocation key, and will explain the criteria in upcoming blog posts. What we do know is that we will share fees on a meritocratic basis (e.g. the better the trader, the higher the share).

Reaching both conclusions has required a fair amount of soul searching – so much so it’s quite relevant to share the main sticking points. As a matter of fact, it’s imperative, for we might revise our decision as we gather more information.

 What happens going forward?

Community unequivocally belongs to traders. Period.

Having said that

  1. Darwinex actively invests in growing and supporting the community,
  2. DarwinexLabs requires highly qualified (=expensive) and motivated personnel. One thing is to access the community data, quite another is to derive and optimize investable strategies from it. Broker-dealers generate outsize returns b-booking off the back of said information. Whilst Darwinex will never conflict with its users, arguably it merits some claim on any profits generated.
  3. The community benefits from the strategies as it is – anyone is free to leverage the strategies for an 80% (100% minus the 20% success fee) share in the upside.

We have decided NOT to make community datasets publicly available, as this could compromise trader IP: any community member could leverage community data in competition with DarwinexLabs.

For as long as we can’t remedy this, we’ll waive any share in the success fee… and we want your take on this, for this is a key strategic decision for Darwinex going forward!