Discussion volatility-trader-r20-LIVE

FX Traders,

Introducing the latest version of the 'Volatility Trader' (Rel. 2.0) running live on a $30K account. All details are pretty much wide open at this point. Currently batting 100% after a week or so -- of course that can change at any time.

As far as can be understood from the design work and testing, this automation could work out well. We've got nothing to hide and a lot to gain if it works, so we thought we would just open it up and let it run. Be advised that there is some potential for floating losses as this is not a scalping system. That being said, VTr2 does not hold losers open indefinitely to try to increase balance irrespective of account equity.

Any observations and comments are welcome. If VTr2 works as we intend, we will likely commercialize with hosted accounts, minimum $10K - $20K or so.

(Note to members of the existing pool: If we commercialize you will get preferential access as per the existing agreement.)

looking good!

will keep an eye on your progress..good luck!

Thank-you pavelr.

Our long GBPUSD positions closed today with over 100 pips each... We'll see next how it does on these new short side positions. Still 100% after a couple of weeks.

Just a quick update... With the GBPUSD cross hitting new low territory last night (during the euro session that is) the system closed out its short positions to maintain its 100% average. Will be interesting to see how long it can keep up this accuracy rate.

Note that VTr2 is expected to trade with decent accuracy (at this point!) but not 100% ongoing so if you are following our open trades please be careful as there are no guarantees!

thx, t.

2010 January 25 update:

Still going strong; still batting 100 so to speak. Interestingly we are getting differing signals at this point suggesting a potential for another drop in cable over the next few days. (The instance that is signaling short is 100% accurate in 6 month back-testing, so it will be interesting to see if it holds.) Also, due to the differing short-term and medium term signals, the system is currently hedging across chart instances.

Anyway, comments from experienced traders in the gallery are welcome. We have noticed that martingale set-ups and locked open trades are not appreciated by those wishing to evaluate systems (and rightly so, especially with respect to martingales...) so a reminder that our system activity viewing is not locked, nor does the automation utilize martingale techniques.

Feb 1/10 update:

Well, we knew it had to happen sooner or later. Interestingly we had a solid month of gains without a single loss, and then it only took one loss to wipe out half of the gains.

Obviously the 'Achilles heel' of our system is the SIZE of the losses. The winning trade rate is fantastic, but if a single loss takes out 2 weeks' earnings then either we need a near 100% success rate or we need to manage the loss size somehow.

Accordingly we are working on determining if some combination of new modeling with tighter stops and/or FX options (as insurance) and/or fully hedged underlying FX positions can be made to work together to solve this large loss size problem.

Will keep you posted, and in the meantime we will let the system keep doing its thing as so far we have not seen any problems with the higher volatility set-ups (the loser today was a one of the two relatively lower volatility set-ups of the four models currently running.)

I like that you have nothing to hide, and willing to discuus about it publicly - that's an excellent way to go commericial and is nothing like the current snake oil sellers with all the fancy websites but with zero proof about their performance.

Looks like you have potential there. May I know what happened with 'Volatility Trader Rel 1.0'? Did it stop working?

Thank-you wilkinson, there does seem to be some good potential.

Lately we have been working with signals that go long and short on cable simultaneously with very strong historical back testing results (some with 100% accuracy, both sides -- short and long, 6 months backtesting.) Again here though the trouble is the size of the stops. With that in mind the extreme potential losses and gains as well as averages are being computed for these set-ups using FX options to protect the positions.

In testing the main set-ups under review the results are encouraging, with only 2.3 winning trades on average required to recover from a loss. This is the configuration that does not lose in backtesting and trades long and short simultaneously. That being said, in order to make things less complex -- involving the use of puts and calls over FX positions that is -- the focus right now is on another trigger set that also goes long and short simultaneously, with both sides at profit factors of only around 2.5, but much more manageable loss sizes. This is where the effort is being spent right now.

Good question re VTr1 -- that was a version that did not support draw-down positions and ranging market adaptation, two features which improve the profitability greatly. That version did trade successfully in forward testing for a while, and then was not successful in live trading due to market conditions that did not match the conditions found during optimization and forward testing. (Hence the addition of 'Ranging Market Adaptation' in this version... If you've watched our trades you may have noticed that the point at which wins are locked in varies according to how the market is moving sideways or trending.)

Thanks for taking time to answer my questions!

If you get 100% accuracy in back testing, I think you've over-optimized the EA to fit the history. I believe you should try to lower the drawdown and the expectancy also, to not have big losers. When a client sees a big losing position, it has a negative psychological effect - he would be scared to continue trading the system. You should concentrate on steadier growth.

About what you wrote -- in the context of our overly large hard stops you are absolutely right about the curve fitting issue. So as you are suggesting, for the past few days our emphasis has been on coming up with a configuration that takes many small losses but even more larger wins.

That's exactly what is running now on another account -- a set-up that enters about 100 trades per month with 57% or so winning and those winning trades being about 57% on average larger than the losers.

That release of the Volatility Trader (r3), running with this new configuration, will be under the account here shortly. The account it's on is not new and has prior trades and withdrawals, so a custom start date will be used -- probably today's date as a matter of fact. There are also some other set-ups running on the same account with the new VT release so there will probably be more than 100 trades per month on average.

VTr3 main features:

a) Volatility Levels can be configured by parameter use to act together in 'clone sets' which trade the same set of parameters and establish master and drawdown positions identically except spaced by a set time lapse value expressed as order frequency.

b) Volatility Levels can be configured by parameter use to act together as 'hedge sets', going long and short simultaneously on the same or very similar signal received from the detection algorithm. Further, these hedge sets can be configured to run in 'strict hedging' mode where the trades are only taken simultaneously or in 'normal hedging' mode where the side not currently open may open on the signal -- provided the direction of entry hedges the other side currently open.

c) A host of system level parameters have been moved down to the Volatility Level, allowing fine tuned control of each of the twenty supported volatility levels within a single instance of the EA. This allows much greater complexity of configuration so what previously took multiple charts and EA instances all sharing the terminal's trading context can now be achieved with a single chart and EA.

d) Fine-tuning of live (running) internal data structures of the program's primary arrays can be done without access of the EA's parameter box, thus allowing control of the main functionality without a deinit/init cycle.

Well...

It's clear that the big stop set-up is a problem, but we already knew that.

The plan is to remove the most frequent / largest stop combination, that uses a VL40 & TD360 combination and works with an initial stop of 450 pips. While it has made some good wins in January the losses are simply too big when they hit.

In the meantime a new set-up that is designed to trade with SUSTAINABLE losses is being plugged in. As a result the chart activity will look significantly different going forward.

Quick update on our system.

We have been running the 'sustainable' losses set-up since February 5 and it has been working very well. The system is making good consistent gains and taking small losing trades here and there when it needs to. Overall the performance seems to be much better. We plan to go live with a signal service with the new set-up sometime in the next little while.

The Volatility Trader signal service will provide fully automated copying of our account activity into clients accounts. Currently in the planning stage, with anticipated roll-out in the next month or so.

Email [email protected] if there is any interest.

Another update...

VTr3 is proving very effective, now that we have resolved the problem of the overly large losses that we were taking a few weeks ago by replacing some of the weaker models with better ones that take smaller losses when necessary. The resulting set-up looks like it should be able to do 10% a month quite easily without much risk exposure.

Our trades are still open for viewing, analysis, and comments from the public. We are moving forward with the signal service, currently running a test client/server signal environment that is working quite well.

If you wish, see the 'volitility-trader-r30-LIVE-A' set-up to monitor our secondary account, which is running another set that has some similarities to the models under this account, but on a different broker.

T. Buitendyk