One more time! Making $ with Zorro?

I was recently asked via PM: As a longer-term Zorro user than most have I made money with Zorro? The questioner said s/he has lost money over the last year or so...

In fact, I recently had my 3-year Zanniversary . I think this question is better answered here in the Forum as perhaps others would also be interested; and perhaps other longer-term users will chime in with their experiences as well.

The short answer to the question is no - I now have less money than I started with.

But of course it's more complex than that. Early on Zorro had made money and I was of course very happy. But then came my biggest loss when the SNB did its thing and Z5 died.

Even that is not as simple as it at first appears. My understanding is that other Zorro users fared differently... My recollection is that jcl said he made money off the SNB event...

From reading lots of Forum posts, trading results vary from user to user due to when a strategy is started, which broker, MT4 vs. API, using more or less aggressive settings, errors in setup, etc. - the list of possible differences seems to be pretty long. And is a big reason why results comparisons on the forum have usually ended up going nowhere...

Trading results also normally do not match Test results. To paraphrase what I think I've learned from jcl so far, the pretty, smoothly increasing equity curve that one works with Zorro to produce is really still only 1 of the many possible outcomes. Even assuming real trading matches the Test statistics, you could still have bad luck and start with immediately full maximum drawdown, which is much worse than 3% UI. And of course reality may not match Test - again paraphrasing what I think I've learned, worst case your strategy could have expired before you even start trading it.

In any case, since SNB I have been looking/hoping for a return to glory, but it has not (yet?) happened. It's particularly worrying that my running of the Z12 flagship strategy was essentially flat over post-SNB 2015, and has been a pretty big loser so far in 2016 - results that more or less match the questioner's. Of course, as noted above, these results are still within the range of the Z12 Test results; but the longer it goes on the more I wonder if it can/will ever recover, much less make money; and the mind fills this (hopefully temporary) down time with questions: Did something fundamental change after the SNB? Are the version upgrade cycles preventing equity trading from working properly? What else might be going on?! Am I really that unlucky? I don't know...

I am not a Forex expert. I was and still am a learning neophyte, and still in the Proof of Concept stage. I consider Zorro to be an excellent tool for what it does, and uniquely well documented, including concepts. But whether algorithmic trading can consistently make money in today's market, and whether a strategy developed in Zorro can consistently make money in line with its Test/predicted results are separate questions that I'm still waiting to see verified.

So, this is a long way of answering that my jury is still out. I haven't lost everything I invested, and was prepared for some loss as part of the learning cost. But to the questioner and others, I would definitely say proceed with extreme caution. There is no magic, simple get rich quick bullet here. There is lots of hard work and learning. And almost certainly losses. Even if you can develop a truly, regularly profitable strategy, Zorro is even then not set-and-forget. Good luck, happy trading, & hope you're smarter than me!

I share your points. I have not been ablo to let my z system run for more than 100 trades without touching or modifying parameters. This is my first error. Every time a new version comes out, I wonder if it make sense to start the new version or continue with old one. I have seen huge performance changes over different version. Something that I could not fully understand. I would like to have:
1) the opportunity to fully re-train z systems from scratch using my data and my broker costs.
2) the opportunity to retrain and test z systems on different STARTDATE & ENDDATE periods.

In order to handle this choice correctly.

My two cents:

I do not trade the Z systems because I can not see the source code. However, based on the information in the manual, it isn't too difficult to create systems based on the same principles as the z systems.

I trade a number of my own systems, some based on similar principles to the z systems, some based on my own machine learning research, and some based on relatively simple market models. I am making money fairly consistently for both myself and investors using Zorro as both my primary research environment and my execution engine.

If my experience is anything to go by, Zorro is an excellent tool for algo trading. But at the end of the day, it is just a tool. It takes a lot of research, persistence and a number of losses to work out what "works". And while I appreciate what the developers are trying to do with their z systems, I would never trade something that I hadn't developed myself.

Thanks @DdlV for sharing your experience so openly.
Sorry if I am touching in a sensitive point. But what "Z5 died" exactly means? Was it a problem with the Z5 or with your specific configuration?

Can someone share why the code of Z systems are not open?

@traderbot33: Z5 depended on the cap the SNB was maintaining. The SNB completely removed the cap, so the Z5 strategy expired. There is more info about this in the manual, Forum, & on the web.

traderbot33's question reminds of one other important point that should have been made:

Whatever one might think of Zorro, ZORRO'S NOT THE WHOLE STORY!

(caps for emphasis, not shouting )

Zorro, as well as you trading manually, are dependent on the broker, and on what's behind the broker. If the broker, for whatever reason, doesn't execute your trade, you're stuck. Worst case stuck in a losing trade(s) with no way out.

And worst worst case LOSING MORE THAN YOUR INVESTED CAPITAL!

Zorro can not only not prevent this, it can't take this possibility into account when Testing (since it assumes execution always happens within certain parameters).

Yet another reason to Beware!

Yep.

This thing of "if the broker, for whatever reason, doesn't execute the trade" is one of main worries with algo trading. It is seems to be something totally out of our control.

I'm still a total beginner with algo trading, so I am not in the position to invest real money, but your experience make me think if it would be not a good idea instead of driving the efforts/money to only one asset to distribute them to several assets. However I don't know how dificult would be to manage all strategies and assets at the same time.

There are a lot of reasons already mentioned for this, I just have another simple one: it depends on Zorro's version, sometimes. Througout the last 2 years I followed Zorro from 1.2x to the actual one and normally a version change happened while I tested sone strategies, so I'm regularly suprised about the results in different versions. Confusingly this happens not always, some strategies show the same results. Maybe these ones are the most stable?
An actual example changing from 1.46.3 to 1.47.7 (the only change: "History = .bar;" in 1.47):



It's only a simple model based script without Parameters, Factors, Rules or anything similar. Maybe those differences happen because of new features while their defaults does not match a version's environment that does not provide them at all, or maybe one result is only another one from the Monte Carlo simulation, I don't know. One can see both curves in the example are somehow parallel but the first one seems to promise something that the second one does not reveal. Strange, but in some cases it can be really interessting to run a strategy on different Zorro versions.

There can be many causes of different test results, but normally not the Zorro version. At least not without good reason. Compare the account parameters that you're using - that would be my first suspect. If you can't find the reason, it might be even a bug in the beta - can you then upload that script?

Yes - this is the reason for the difference between both versions (at least concerning this script). That means from now on we have to set the flag EXTRADATA always to receive results as accurate as in former versions?

No, the difference should be marginal. If it is not, there could be a large randomness in the strategy.

Only for interest: could you please explain a bit what has changed?
It might be that my script has a large randomness but I only use explicitely priceClose() and try to bring it together in mind with 'only high and low are stored'. Does this mean that the close is calculated somehow (e.g. as an average) of high and low?

Sort of. If EXTRADATA is set, the open or close price of a bar is the open or close price of its first or last tick. Otherwise it's the center price of its first or last tick.

Since the uncertainty range of a tick close price is bigger than the difference of center and close, a backtest without EXTRADATA is not less accurate. The remark under "News" is therefore a bit misleading. But you should set that flag when you use low resolution historical data, such as M1 data for one-minute bars, or D1 data for 1-day bars.

Thanks!

It is a very old thread, but I think some guys here seem to be able to use machine learning algos profitably but on the other hand do miss basic things about how the Z strategies work...

The Z strategies clearly state that there will be updates to the trained parameter sets in new zorro versions. This in fact means that the parametrization of the strategies are adapted to the current market behavior. Although WFO will basically help keeping the previous history record, it is not guaranteed to stay the same as WFOs are expanded at each training due to its' fixed segmentation logic. This is much more true for the latest WFO segment which provides parameter for actual trading.

As a logical consequence, the backtests will change naturally due to the parameters which are expanded in test time and optimized for the CURRENT markets and not the markets in 2015 or 2010. You may want to read about WFO, forward testing and market inefficiencies to understand the reasons.