Steve,
I have a lot of respect for your work, and your helpfulness on other boards. I don't share your view re Merlin though.
The vendors hiked the price of the system after the first decent run, then altered the parameters after the first bad run.
Fo a system with almost 30 parameters there's a real risk of curve fitting. Altering the system after each bad run doesn't ease the worry. The system may have broken a new high, but after an awful period that is almost (IMHO) untradeable for a black box system.
If you factor in the cost of the system, spread and slippage - to a realistic account someone would blindly commit to this system - I doubt the equity curve has delivered many happy customers - and they're surely not all "wrong"?
It's all about opinions, eh
UTB
Many good points and well-expressed Blades. I understand your points on their price and changes. Price I guess is each to his own, I made my own choices on mine. As to program changes I have more of an opinion.
Your point on changes and parameters, time will have to tell. While I know nothing of their internal program logic, I have some 2000 hours work in on mine. So, while I can't have an informed opinion about any changes in merlins code, I can provide a great deal of perspective on mine.
The discussion on curve-fit and/or over-optimized is very widespread on the net and I'm honestly not sure how to properly differentiate 'improvement' from curve-fit. Except maybe when they are done within a string of winning signals that did not have a blowout and were performing as per all the backtest history.
I can tell you for a fact though, that in my own work, I now have 6 months real-time trades at my yahoo group and then C2, which show 60% as of Friday. The equity curve has been steadily upward, including losers, and I have made dozens of changes 'during' that time, as well as during the 7 years I've been doing this.
The conundrum is these were not changes to fix blowouts, they were improvements while I was racking up winners in real-time.
I truly understand and appreciate your views plus the civilised way you express them. I guess my main point would be (not speaking about merlin), I don't think all code changes are to fix blowouts nor are they curve-fitting.
One for instance is during the past few months I was supplied a hand data mined 10 year history of the equity put call ratio. If you've seen my posts elsewhere, I had been unhappy with the increasing distortion in the original pcr because they have included index options in recent years. These didn't even exist when Marty Zweig invented the ratio and are 'not' a contrarian indicator, like small retail options players.
So, for 2-3 years I had been blending in cpce data and starting to use it as another cross-check to counter the increasing wildness of cpc. When I got the whole 10 years cpce, I was able to completely shift over. yet I had no disaster I was fixing. I could have gone as long as I want using cpce to say 'no not yet' to the original cpc based decision code but why have two steps when one will do.
I don't think what I described above was curve-fitting or over-optimizing. I was fixing a perceived growing problem in cpc even though I was still racking up profitable trades.
Sorry for the long coffeeless ramble, I guess my point is as follows: I can't judge merlins changes because I don't know what they are to put them thru the context(s) I've laid out here.
Based on my own experience I don't think all changes are for disasters or are they curve fitting. Some are changes in a system already working and just code improvements. Another for instance is lately I've added code regarding the spx climactic volume indicator and also trinq/cpce as short term sell signals for my piggyback trades. This did not replace or change anything already there. It did yield some better sells which still would occur if I took the new code back out, just not as well.
regards, Steve
