Software that finds trading systems

[Pat494]

Stick to your eye-balling TA. I underestimated it first look many years ago but having been around the block with algos, neural nets etc. I am back where I started, having wasted time and money.

 

[malaguti]

Stick to your eye-balling TA. I underestimated it first look many years ago but having been around the block with algos, neural nets etc. I am back where I started, having wasted time and money.

exactly what I said on the first page..however he seems adamant that he wants to waste that time. A real shame even with the benefit of hindsight, you just cant help someone

 

[Liquid validity]

Don't take this the wrong way, I'm not having a pop.

When people fail at something, why dismiss it when there is clear evidence it can work?
If I fail, I will say I failed, I wasn't good enough.

 

[wackypete2]

Don't take this the wrong way, I'm not having a pop.

When people fail at something, why dismiss it when there is clear evidence it can work?
If I fail, I will say I failed, I wasn't good enough.

You're on the wrong forum with that mindset

Peter

 

[Pat494]

I don't think he probably needs to have. It nearly always boils down to the same thing. You are looking for an edge, you want a system that produces backtested results. You are looking for something then that will be based on historical performance of indicators..
Later on you will then be asking, why doesnt that system work, what are the right parameters for that indicator to work..possibly go down the route of optimisation, still with even worse results would be the guess.
You look for something else..etc etc ad infinitum

THEN, you will realise its not an indicator based system you are looking for after all, as indicators are all lagging and you want to be looking at the price action and there aint no system that can do that for you..just eyeballing and screen time
Thats why PB doesn't need to have seen stratasearch or whatever its called to know you could be barking up the wrong tree.
Try a canned strategy, from a free trial of metastock or omnitrader (100s of canned strategies) that way then you can get some of what I've just said out of your system for nothing. PB is simply sparing you the expense as am I.

Its just a journey many have done before, including myself, thats all. Until the penny drops and you realise actually you could have saved yourself a few years.
But I honestly wish you the very best of luck

So you did.

Great minds and all that ?
Recently I nearly set out on a Bayesian mission but have thought better of it.

 

[Liquid validity]

You're on the wrong forum with that mindset

Peter

I'm inclined to agree.

 

[Liquid validity]

So you did.

Great minds and all that ?
Recently I nearly set out on a Bayesian mission but have thought better of it.

Pat and Malaguti, just to clarify, I completely agree on the points you both raise
as in how not to do it.
That is the wrong approach, as you have emphasized.
Its also the one most people take, so that may explain it...
The number of indicators that have any value can be counted on one hand with
fingers to spare, and even then, their value does not lie in their traditional application.

 

[Liquid validity]

 

[the hare]

Recently I nearly set out on a Bayesian mission but have thought better of it.

Been there, done that and got the t-shirt

Not altogether unproductive, but then again, not as productive as tossing coins.

 

[Jan Tabak]

What do you guys think of Connors & Alvarez indicators like 2-day RSI?
Or their 7-7 mean reversal strategy?
No edge?
It seems like fun to be able to test an idea like that.

And while everyone agrees that technical indicators are lagging, it doesn't mean they are worthless. When deciding to enter medium term trends (days-weeks) in commodities I find DI-/DI+ (7) and MACD very helpful (if not in contradiction with seasonality amongst others).

 

[Liquid validity]

And while everyone agrees that technical indicators are lagging, it doesn't mean they are worthless. When deciding to enter medium term trends (days-weeks) in commodities I find DI-/DI+ (7) and MACD very helpful (if not in contradiction with seasonality amongst others).

If used in discretionary trading with subjective interpretation, maybe,
although I personally think they are more often just a crutch.
In isolation, most indicators with no subjective interpretation are no better
than random, and quite often worse for a variety of reasons.

Its that lack of subjectivity that means a quant approach is vastly different
to a subjective discretionary approach, even if applied in a systematic manner.

 

[malaguti]

Pat and Malaguti, just to clarify, I completely agree on the points you both raise
as in how not to do it.
That is the wrong approach, as you have emphasized.
Its also the one most people take, so that may explain it...
The number of indicators that have any value can be counted on one hand with
fingers to spare, and even then, their value does not lie in their traditional application.

I understand..its very difficult to hear something and think "ok then I'll forget what Im doing and follow another path" That person will probably learn alot more doing it for himself..
The best I came to any indicator was one based on (dare I say it) solar activity I know i know uranus and all that, but jeez if there was just a shred more reliability fortunes really could be made.
I remember saying to a buddy, buy the FTSE at 10.45 3 days from that point. And lo and behold..just uncanny. But unreliable, and that was its problem. Anyway, I digress. Each to their own LV. no hard feelings

 

[flashflasher]

I have never used it but I got to know like few months back that we can’t expect the result what we want. It’s not accurate. Hunt and Peck method is a good method to use.

 

[Jan Tabak]

Thanks! Interesting read! Lots of systems developed in the bull market up till 2008 have failed over the last years.

I wonder though if he is right. Reversion to the mean is of course nonsense in a random system but then we know the markets are not completely random.

And what about Price Action Lab software itself?
Anyone any experience?

 

[D70]

Jan Tabak,

I am also travelling this route of self discovery. Sorry, I mean, trying to code up my trading. I dont believe the guys that say 'dont bother code it, it'll never make money' because there are umpteen very profitable billion dollar funds that only run automated systems. Who would of thought you could fly from here to another country in a giant metal dildo with wings. That's impossible, no?

I digress. The problem you now have is a serious lack of resources. Namely lack of coding knowledge. You can find a coder but I expect it would be most beneficial to learn to code yourself. That'll take a few years, assuming x many hours per week, courses followed, etc etc. It is a road you must tread or your end product will most likely be deficient. A few before you have already tread this path so your route should be quicker than theirs. You'll no doubt gain some valuable tech knowledge along the way and who knows, it may even highlight some very interesting results in itself.

Whether the 'system' makes money or not usually doesnt come down to the 'signals' but your risk management and portfolio risk management. Standard human response to most things is to think 'I need the perfect entry' then I'll make money but in trading you need entry, exit and risk (everyone reading this no doubt knows this anyway). Without all 3 taken to the nth degree of iteration. You could still not make money. People dont lose money generally because of where they chose to enter but because of position sizing and exit methodology and overall portfolio risk.

Hare, you should really right a thread on your coin toss experience given that you reference it so much.

Finally. It may well be a 'waste of time' but isnt life a waste of time?

 

[darktrade]

And what about Price Action Lab software itself?
Anyone any experience?

I received good support after I downloaded the demo. It appears this can be used by both automated and discretionary traders so one can switch between the two with the same software. I checked out the portfolio back-testing feature and it works. I think it takes some time to learn how to use properly and results will depend on experience, skill, etc.

 

[segutor]

Thanks! Interesting read! Lots of systems developed in the bull market up till 2008 have failed over the last years.

I wonder though if he is right. Reversion to the mean is of course nonsense in a random system but then we know the markets are not completely random.

And what about Price Action Lab software itself?
Anyone any experience?

Looks like a good resource for quants. I don't think the average Joe who can't do math and backtesting can use it. See in this post how he analyzes an pattern in GLD. I think the approach can be refined further.

 

[Jan Tabak]

Yes, I agree on both counts.

The system seems a bit daunting to me, but then I am only beginning in backtesting.

And it would be interesting to see if the double inside day is a reversal or a continuation pattern rather than a bearish pattern. I suspect it is a contunuation pattern but I lack the capability to test this.

 

[avano]

Yes, I agree on both counts.

The system seems a bit daunting to me, but then I am only beginning in backtesting.

And it would be interesting to see if the double inside day is a reversal or a continuation pattern rather than a bearish pattern. I suspect it is a contunuation pattern but I lack the capability to test this.

If a pattern is a bullish signal for example why would you care if it is reversal or continuation? It is reversal if it reverses and continuation if it continues. TA is tautology like in logic.

 

[Jan Tabak]

If a pattern is a bullish signal for example why would you care if it is reversal or continuation? It is reversal if it reverses and continuation if it continues. TA is tautology like in logic.

I suspect it is not a bullish signal but a continuation signal ergo it would only be bullish in case of a preceding bullish period.