How useful is this info/data?

[xddc]

hey guys,

long time lurker, fairly new poster....
So here's the story...i've been working on my own matlab code and have done several things, including a backtester and a scanner with the past 10yrs of eod data for each stk in asx200.

With the scanner, at each eod i can get it to generate a report showing which stks have been triggered by a particular indicator...right now i have MACD, 8 open/8 close, closing price crossing 21avg and their variants. It also tallies up the number of times a particular stk has been triggered (ie. the number of indicators which say it has triggered a buy sig) for the next day's trades and also the number of times a trigger had previously occurred (but would not have closed out since).

My question is, how useful would this info/data be to you guys? From what i gather, this is the sort of thing that people pay good money for subscriptions to? i'd be happy to share it if we can explore its usefulness/effectiveness.

Theres a few sheets in there but lets start with the 'BUY-REC' sheet...The 'code' column is obviously the code, 'XJ50' is whether the stk is in the XJ50, 'new recs' is the number of indicators that says a buy signal has been triggered for tomorrow. 'In port' is the number of indicators that says a previous buy signal had been triggered (and presumably the position opened and had not yet been closed). 'Total' is the number of indicators which say to go long in that particular stk, ie. buy signals previously triggered + currently triggered.
To the right, an X is marked for the indicator which says a new buy signal has been triggered for tomorrow whereas it would otherwise be null, or have the entry price and date for which it was previously triggered.

The other sheets may also require some elaborating and i will be happy to do so if anybody is interested.

cheers

 

[xddc]

in case anyone doesnt have excel...

 

[Wysiwyg]

If it is any consolation, CWN has come up on one of my strategies a few times lately. Too many candidates and not enough money though. :

My question is, how useful would this info/data be to you guys? From what i gather, this is the sort of thing that people pay good money for subscriptions to? i'd be happy to share it if we can explore its usefulness/effectiveness.

I don't pay to get a signal generation.

 

[PrudentInvestor]

hey guys,

My question is, how useful would this info/data be to you guys? From what i gather, this is the sort of thing that people pay good money for subscriptions to? i'd be happy to share it if we can explore its usefulness/effectiveness.

In my opinion there is so much data available today that mere numbers (or signals, if you will) are not worth much. But if you could somehow indicate the explanatory power of your info/data, then you'd probably have to be selling it instead of giving it away for free

 

[xddc]

yes, i could probably pretty it up and maybe even go so far as to market it later on down the track when its more complete, however thats not my goal.
I suppose the perspective im taking is to subject the results to peer review as opposed to giving it away for free.
How often can we say that about blackboxes/newsletters/etc etc that market their picks? bearing in mind they are certainly not free.

On a broader note, it would be good if i can get suggestions on how to make the data more useful (presentation, content, whatever) if that will garner more interest

Heres yesterday's after close. Again, X marks trigger for this morning otherwise its price and entry from previous trigger.

 

[PrudentInvestor]

I suppose the perspective im taking is to subject the results to peer review as opposed to giving it away for free.

[...]

On a broader note, it would be good if i can get suggestions on how to make the data more useful (presentation, content, whatever) if that will garner more interest

You want the forum members to review your stuff so that you can begin to sell it? Come on, buddy.

Seriously, in my opinion there are so many decent "signal providers" these days that it is hard to make a difference. What the big guns are spending money on is statistically proven models with genuine predictive capability. This is the substance I would be aiming to add if I were you. Easier said than done.

Just my two cents.

 

[brty]

xddc,

an X is marked for the indicator which says a new buy signal has been triggered for tomorrow

What makes these "buy signals" a buy signal?? All the different signals generated seem to show that the price is now higher than at some previous time. My question is what makes it a "buy signal", what evidence do you have, after allowing for slippage and commission, that any of these signals are in fact a good place to buy??

Further do you have any evidence that a combination of these signals would lead to better performance than one signal or perhaps no signals and a good set of darts and a monkey??

If you don't have any answers to the above questions, then why would the signals be of interest to anyone, including yourself??

brty

 

[xddc]

You want the forum members to review your stuff so that you can begin to sell it?

did i not clearly state that thats not my goal?

yes, i could probably pretty it up and maybe even go so far as to market it later on down the track when its more complete, however thats not my goal.

you said im giving it away for free, im saying thats not the point.
Yes, the plan is to include a way to assess performance both via statistics and simulating a portfolio. Further down the track ill try to even make it more quant oriented as opposed to tech indicators. As i said its still in early stages but the possibilities seem endless with matlab.

brty:
What makes these 'buy signals' a 'buy' signal depends on the indicator, as you look for the macd to cross its 9ema for eg. Why are there only 3 indicators and simple ones at that? because theyre the easiest to implement and code...ill have more coming when i have time. Theres no way i can model for your or everybodys commission, but for myself they are signals to go in at market at 0.1% comm for anywhere between 10mins to 3-4 days for the top5 picks (unless its an obvious false positive like the day before xd) at about $2k each.

evidence? see the top5s urself...
all opened at 23rd feb
cwn open 7.89 last 7.85
fkp open .685 last .74
ncm open 32.9 stopped out at -2.5%
ppt open 34.83 last 3579
and sun the biggest loser stopping out ~5% after the gap.
(as of ~1.30pm today)

i dont have any evidence that a combination of the signals would lead to a better performance blah blah......but that is exactly the point....to tabulate the indicators and see, how they perform individually and/or in unison. So you're saying nobody would be interested unless i go find the answers myself and only then come back? Is that simply reflective of the take and no give attitude nowadays? you want evidence? im not trying to sell anything...its an open invitation to investigate its usefulness or lack of.

 

[Trembling Hand]

xddc you have asked is this info is useful. What Brty is getting at is NO, not unless you already know that it offers some sort of edge over a significant period and under what circumstances resulted in that success.

Buy signals off simple indicators are out there in the millions. Whats not is the data/stats to back up why they are profitable. I can see as a programmer one would be interested in generating signals but as a trader signals alone are USELESS.

 

[brty]

xddc,

I think you have misinterpreted my question, or I did not explain it well enough.

You have chosen eg.

a 'buy' signal depends on the indicator, as you look for the macd to cross its 9ema

OK, the indicator crosses, so what makes it a "buy signal"?? Did you read somewhere that it should be?? Do you have evidence that it is etc?? Why choose these signals and not some others like a 15 day EMA or a 3 day EMA.

So you're saying nobody would be interested unless i go find the answers myself and only then come back?

You were the person who proposed this, what evidence do you have that it is any good?? What made you choose these indicators and the numbers you use within them??

Anyone can come up with a formula to follow, but what is the point of even suggesting it unless it means something. I'm asking you what these series of indicators mean to you, why do you think they will mean something to someone else??

brty

TH I suppose I spoiled your fun in another thread and this is payback. (insert smiley thingy here)

 

[xddc]

if you read my previous post, i chose the indicators because they were easy to implement using pretty standard parameters. i dont have evidence....why should i, i didnt come up with them! and in case you missed the point, that is the purpose of this exercise....to test whether they work and to tabulate the results, in effect forming the evidence.

When i start devising strategies and/or optimising parameters then ill care what makes it a 'buy signal' or not, or whether to use the 15 day EMA or a 3 day EMA instead.

Perhaps you've misunderstood, i'm not seeing what this particular series of indicators mean to other people, im seeing if having the tabulated results (on an expanding library of indicators) on a daily basis is useful to people.

clearly, as of EOD 1st march catching two x 7% runners against three stopped at 2.5% is pretty useless.

 

[Trembling Hand]

xddc you want to do so stop asking is it helpful. In your opinion its is, what do you want from us? just get on with.

But when you have 100s of signals every day some being triggered by one indicator, other being triggered by different ones, come back and show us how useful that is. May I suggest a sample size of about 100 real trades or 500 back-tested.

 

[Richard Dale]

Don't forget survivorship bias too. Historical testing on a subset of data needs that subset to be consistent with your universe of stocks present at that time.

eg. Consider One.Tel, HIH Insurance, Poseidon, ABC Learning, Sons of Gwalia etc.