Dr,
Considering his expertise in opthalmology, etc why is he wearing specs?
I suppose these academics are simply jumping on the "quants" bandwagon - their qualifications give them legitimacy, or so they think as do the gullible punters. Moreover, they believe they can apply their expertise across disciplines. That's why one hears of exciting new systems utilising the latest reasearch and developments in artificial intelligence, genetic algorithms, neurology, etc. It just doesn't work but newbies buy into it
This may seem an anti-scientific rant but it isn't, I can assure you - I believe scientific rationalism is the antidote to superstition and religious dogma.
Are you a "Doctor"? Of what - Medicine, Letters, Science, Philosophy, Divinity?
Grant.
Considering his expertise in opthalmology, etc why is he wearing specs?
I suppose these academics are simply jumping on the "quants" bandwagon - their qualifications give them legitimacy, or so they think as do the gullible punters. Moreover, they believe they can apply their expertise across disciplines. That's why one hears of exciting new systems utilising the latest reasearch and developments in artificial intelligence, genetic algorithms, neurology, etc. It just doesn't work but newbies buy into it
This may seem an anti-scientific rant but it isn't, I can assure you - I believe scientific rationalism is the antidote to superstition and religious dogma.
Are you a "Doctor"? Of what - Medicine, Letters, Science, Philosophy, Divinity?
Grant.
I know this is a bit late since NeuralDT has been locked out now after Breakout Futures sold 100 copies but...
I have to say that I found this thread accidentally and was struck by how completely negative the participants are. Like so many threads on T2W their are people who have no actual experience of particular topics just venting their petty prejudices to bolster their own egos. Well I can add some facts if anyone here is truly interested.
I have been auto-trading since 2000 and run several of my own systems (www.alchemetrics.org) and some purchased from other vendors. I highly respect Breakout Futures (look at Mikes Bryant's newsletter page to get some idea about the level of thought he puts into his systems). Of his systems, I own MinMax and NeuralDT. I did purchase another system called WCI (West Coast Index) but I was able to return it under his 30 day guarantee with no questions asked since I didn't like the way it traded.
With MinMax I never made any money but didn't lose much either. I can't remember the exact details. I think it cost about $500 to buy but after six months trading a $20,000 account I was down about $1,200 so I stopped. At that time I wasn't really into optimizing position sizes in a portfolio (I traded 1 contract of each system) so it may be that might have improved things. Indeed, Mike Bryant has just released an update to MinMax (some 5 years after I purchased it) so I may give it another go with variable position sizing this year.
Despite losing money, what did impress me was this continued support that Mike puts into his systems so much so that I bought WCI (returned for refund as stated above) and then NeuralDT. I'd used Neural Nets for my own systems before via Neuroshell Trader but with NeuralDT I liked the novel way Mike had built a simple net inside of TradeStation and used it simply to act as a confirming filter to an existing system. This confirming mode is the way which I've had most success with when using "real" neural nets. IIRC, NeuralDT cost me $2,000.
I started trading a $30,000 account for NeuralDT in Aug 2008. As of today I am still trading the NeuralDT portfolio of systems and the account stands at $86,000. I'm using a FF position sizing algorithm with various fractions devoted to each sub-system. The fractions range from 0.4% to 2.2%. There have been drawdown periods. The max equity reached $107K so the system is now in a drawdown period of around 20%. I haven't re-optimised/tweaked the actual system itself over that time, although Mike has done this once and I have applied his changes. What I have done, however, is re-optimised the FF risk sizes (without forward test) for the portfolio on a monthly basis. Given the equity run up, I consider the original $2,000 well spent.
Regards
I have to say that I found this thread accidentally and was struck by how completely negative the participants are. Like so many threads on T2W their are people who have no actual experience of particular topics just venting their petty prejudices to bolster their own egos. Well I can add some facts if anyone here is truly interested.
I have been auto-trading since 2000 and run several of my own systems (www.alchemetrics.org) and some purchased from other vendors. I highly respect Breakout Futures (look at Mikes Bryant's newsletter page to get some idea about the level of thought he puts into his systems). Of his systems, I own MinMax and NeuralDT. I did purchase another system called WCI (West Coast Index) but I was able to return it under his 30 day guarantee with no questions asked since I didn't like the way it traded.
With MinMax I never made any money but didn't lose much either. I can't remember the exact details. I think it cost about $500 to buy but after six months trading a $20,000 account I was down about $1,200 so I stopped. At that time I wasn't really into optimizing position sizes in a portfolio (I traded 1 contract of each system) so it may be that might have improved things. Indeed, Mike Bryant has just released an update to MinMax (some 5 years after I purchased it) so I may give it another go with variable position sizing this year.
Despite losing money, what did impress me was this continued support that Mike puts into his systems so much so that I bought WCI (returned for refund as stated above) and then NeuralDT. I'd used Neural Nets for my own systems before via Neuroshell Trader but with NeuralDT I liked the novel way Mike had built a simple net inside of TradeStation and used it simply to act as a confirming filter to an existing system. This confirming mode is the way which I've had most success with when using "real" neural nets. IIRC, NeuralDT cost me $2,000.
I started trading a $30,000 account for NeuralDT in Aug 2008. As of today I am still trading the NeuralDT portfolio of systems and the account stands at $86,000. I'm using a FF position sizing algorithm with various fractions devoted to each sub-system. The fractions range from 0.4% to 2.2%. There have been drawdown periods. The max equity reached $107K so the system is now in a drawdown period of around 20%. I haven't re-optimised/tweaked the actual system itself over that time, although Mike has done this once and I have applied his changes. What I have done, however, is re-optimised the FF risk sizes (without forward test) for the portfolio on a monthly basis. Given the equity run up, I consider the original $2,000 well spent.
Regards
vvierailija
Newbie
- Messages
- 4
- Likes
- 0
What instrument you are trading ?
Back again after 18 months..
I'm trading TF and EMD with NeuralDT.
Just thought you all might like an update as to the state of the NeuralDT systems. Last time I posted, the systems were about 20% in drawdown. This continued to about 35%, so back in September 2010 I decided to re-optimise all of the NDT systems from scratch. However, instead of using just the standard TradeStation optimisation function as described in Mike Bryant's manual, I've used a more sophisticated process called Walk Forward Optimisation.
The Walk Forward Optimiser (WFO) I used was from GRAIL Global Futures Trading System for TradeStation. I know the website looks rather “Mickey Mouse" but behind the garish colours the tools are actually rather good (so much so that they have been acquired by TradeStation and built in as part of their lastest version). I won't go into the full details of how WFO works but the idea is that one constantly re-optimises a system after fixed periods of time.
Since September 2010 I have been running the following parameters for NDT 60 EMD:
FloorFr = 2.2
TrailPr = 36
StopFr = 2
w1 = 0.54
w2 = -0.32
w3 = 1.4
w4 = -1.31
w5 = -0.99
w6 = -1.49
The optimisation process suggested that these parameters would produce an average gain per contract of $8900 per year. I ran this for nine months and achieved approximately $7000 per EMD contract over that period which suggests that the system is trading according to plan. The drawdown over that period was very modest.
In June this year (2011) it came to the time when the systems needed re-optimising and the following parameters for NDT 60 EMD are now current:
FloorFr = 0.9
TrailPr = 26
StopFr = 0.9
w1 = -1.65
w2 = 1.94
w3 = 1.43
w4 = -1.42
w5 = -1.93
w6 = -3.07
The new optimisation predicts an average yearly gain of $10,800 per year per contract. Please note that while performing many of these Walk Forward Optimisations I have rarely seen any average yearly gains greater than $12,000 per contract (at least while keeping drawdown within sensible limits). Hence I'm still of the opinion that the NDT 60 systems are viable and worth trading. Here is the TS equity curve for those parameters:
However, more interesting is the WFO optimization report:
Although slightly worse than the current TS report in terms of total gains, this actually shows what would have happened if we had traded the system whilst re-optimizing at each of the vertical lines. The periods between the green lines are all out-of-sample sections, so much more likely to be realistic than the standard TS backtest results.
I have also re-optimised NDT 60 TF with similar results. The whole optimisation process takes about two days but is a lot simpler than the method suggested in the original NDT manual.
Incidentally, the WFO optimisation process has shown that the NDT 1 min systems are not really robust enough to trade so I have given up on these and now only trade the 60 min versions. Here's a WFO equity plot that shows why:
I hope this has been useful for anyone out there who has also purchased the NDT systems in the past.
Best regards
I'm trading TF and EMD with NeuralDT.
Just thought you all might like an update as to the state of the NeuralDT systems. Last time I posted, the systems were about 20% in drawdown. This continued to about 35%, so back in September 2010 I decided to re-optimise all of the NDT systems from scratch. However, instead of using just the standard TradeStation optimisation function as described in Mike Bryant's manual, I've used a more sophisticated process called Walk Forward Optimisation.
The Walk Forward Optimiser (WFO) I used was from GRAIL Global Futures Trading System for TradeStation. I know the website looks rather “Mickey Mouse" but behind the garish colours the tools are actually rather good (so much so that they have been acquired by TradeStation and built in as part of their lastest version). I won't go into the full details of how WFO works but the idea is that one constantly re-optimises a system after fixed periods of time.
Since September 2010 I have been running the following parameters for NDT 60 EMD:
FloorFr = 2.2
TrailPr = 36
StopFr = 2
w1 = 0.54
w2 = -0.32
w3 = 1.4
w4 = -1.31
w5 = -0.99
w6 = -1.49
The optimisation process suggested that these parameters would produce an average gain per contract of $8900 per year. I ran this for nine months and achieved approximately $7000 per EMD contract over that period which suggests that the system is trading according to plan. The drawdown over that period was very modest.
In June this year (2011) it came to the time when the systems needed re-optimising and the following parameters for NDT 60 EMD are now current:
FloorFr = 0.9
TrailPr = 26
StopFr = 0.9
w1 = -1.65
w2 = 1.94
w3 = 1.43
w4 = -1.42
w5 = -1.93
w6 = -3.07
The new optimisation predicts an average yearly gain of $10,800 per year per contract. Please note that while performing many of these Walk Forward Optimisations I have rarely seen any average yearly gains greater than $12,000 per contract (at least while keeping drawdown within sensible limits). Hence I'm still of the opinion that the NDT 60 systems are viable and worth trading. Here is the TS equity curve for those parameters:
However, more interesting is the WFO optimization report:
Although slightly worse than the current TS report in terms of total gains, this actually shows what would have happened if we had traded the system whilst re-optimizing at each of the vertical lines. The periods between the green lines are all out-of-sample sections, so much more likely to be realistic than the standard TS backtest results.
I have also re-optimised NDT 60 TF with similar results. The whole optimisation process takes about two days but is a lot simpler than the method suggested in the original NDT manual.
Incidentally, the WFO optimisation process has shown that the NDT 1 min systems are not really robust enough to trade so I have given up on these and now only trade the 60 min versions. Here's a WFO equity plot that shows why:
I hope this has been useful for anyone out there who has also purchased the NDT systems in the past.
Best regards
GambitRemy
Newbie
- Messages
- 1
- Likes
- 0
Back again after 18 months..
I'm trading TF and EMD with NeuralDT.
Just thought you all might like an update as to the state of the NeuralDT systems. Last time I posted, the systems were about 20% in drawdown. This continued to about 35%, so back in September 2010 I decided to re-optimise all of the NDT systems from scratch. However, instead of using just the standard TradeStation optimisation function as described in Mike Bryant's manual, I've used a more sophisticated process called Walk Forward Optimisation.
The Walk Forward Optimiser (WFO) I used was from GRAIL Global Futures Trading System for TradeStation. I know the website looks rather “Mickey Mouse" but behind the garish colours the tools are actually rather good (so much so that they have been acquired by TradeStation and built in as part of their lastest version). I won't go into the full details of how WFO works but the idea is that one constantly re-optimises a system after fixed periods of time.
Since September 2010 I have been running the following parameters for NDT 60 EMD:
FloorFr = 2.2
TrailPr = 36
StopFr = 2
w1 = 0.54
w2 = -0.32
w3 = 1.4
w4 = -1.31
w5 = -0.99
w6 = -1.49
The optimisation process suggested that these parameters would produce an average gain per contract of $8900 per year. I ran this for nine months and achieved approximately $7000 per EMD contract over that period which suggests that the system is trading according to plan. The drawdown over that period was very modest.
In June this year (2011) it came to the time when the systems needed re-optimising and the following parameters for NDT 60 EMD are now current:
FloorFr = 0.9
TrailPr = 26
StopFr = 0.9
w1 = -1.65
w2 = 1.94
w3 = 1.43
w4 = -1.42
w5 = -1.93
w6 = -3.07
The new optimisation predicts an average yearly gain of $10,800 per year per contract. Please note that while performing many of these Walk Forward Optimisations I have rarely seen any average yearly gains greater than $12,000 per contract (at least while keeping drawdown within sensible limits). Hence I'm still of the opinion that the NDT 60 systems are viable and worth trading. Here is the TS equity curve for those parameters:
However, more interesting is the WFO optimization report:
Although slightly worse than the current TS report in terms of total gains, this actually shows what would have happened if we had traded the system whilst re-optimizing at each of the vertical lines. The periods between the green lines are all out-of-sample sections, so much more likely to be realistic than the standard TS backtest results.
I have also re-optimised NDT 60 TF with similar results. The whole optimisation process takes about two days but is a lot simpler than the method suggested in the original NDT manual.
Incidentally, the WFO optimisation process has shown that the NDT 1 min systems are not really robust enough to trade so I have given up on these and now only trade the 60 min versions. Here's a WFO equity plot that shows why:
I hope this has been useful for anyone out there who has also purchased the NDT systems in the past.
Best regards
thanks for the update bowerandy. how familiar are you with his Reliability trading system? I see that the Neural has similar results given what is posted on his site. Have you been happy with your experience with his systems performance? I assume so since the returns seem to be going very well. Also, it seems the Reiliability system needs some optimization, is this difficult for a novice to do? What, if anything, would hold you back from purchasing another system from him again? thanks!
Similar threads
