hey Jow Forums
are there any indicators that help show the "velocity" of the drop to help see how close it is to potentially being over? I have ADX which helps but its just a bit too laggy.
I'm working on a custom strategy rn, its doing alright but you can see the circled trades would have failed and I'm trying to avoid them if possible.
Other urls found in this thread:
investopedia.com
twitter.com
Cmyk does
fucking bless you, user
Almost all indicators are laggy and can't stand on their own feet
Not asking for your stategy but what's it based on? Volume?
yea, looking for confluence, I unironically believe there is a set of indicators out there that can get you 80% with a 2 p/f
I can backtest and get those numbers, problem is getting them before. Trying to narrow it down to just taking profit based on volatility.
99% momentum, IMO its the only thing that matters and is seemingly consistent. I've looked at adding volume into the mix but it usually just makes things more complicated.
If I had the time/skill, I truly believe that if you find a way to take the sum of the top 20 indicators, it would work amazingly well, as lots of people are trading off them, they become self fulfilling.
Add a bit of price action and very basic TA and I think you're golden. Again, just IMO
Yes. Rate of change indicator. You can't honestly be retarded enough to not have been able to find that before asking biz... can you?
I'm not a fan of that one, I guess I'm asking for more out of the box indicators.
I used ROC in the past, it produces the same signals, ADX is/has been the best so far.
when you do backtesting, you will want to avoid basic mistakes which are fitting your parameters to past data to get the nicer looking results. these are traditionally all doomed strategies cause they dont posses inheritent edge, pretty much any strategy can be fitted to give positive results on a specific time frame in specific asset.
Okay user. here's an outside-the-box one for you. Percentage buys vs sells in the order history (usually only the last 5 minutes is available). This can be good for short term but you can even extend it long term if you collect the data yourself. Funnily enough this isn't a conventional technical indicator so you're going to need to know how to program it.
Use it wisely, young one.
The only things I change in the strategy are lengths, strengths and stops.
You could theoretically have these set before back-testing to see them work really good, so its a bit of a luck thing. i.e. a length of 17 and stop of 1% produce 80% correct and 2 p/f when backtesting last month. IF you were to go back in time and set it to 17 and 1% you could collect the what your backtest gives you. Of course that's a bit of a cop out but there's nothing stopping you for having those settings before backtesting. Trying to see why those particular settings worked/didn't work and if there's signs to show what will/won't. So you can run the best variables on the current month. I've been able to hit around 60% with a 1.5 p/f
I totally get what you're saying though, just trying to see what works and what doesn't.
if you switch those three parameters you will have a SUPER fitted strategy. I would be super careful to forward trade that with any meaningful amount of money
did anyone's TA predict what happened today?
thanks, this is actually a decent idea. Programming shouldn't be an issue... however, the order book has always been tricky for me.
Orders have vanished instantly and are almost instant reactions to this or that, at least in my experience. However, having the data there able to study would be really interesting, I'll see what I can do, thx user.
Fixing stops especially is very dangerous if it looks like changing them have huge impact on results. A trend following strategies dont necessarily need stops. Divergence strategies do require stops in general
Exactly, I haven't used real money but I've been working on this for months. Every month I've been able to hit at least 60% with a 1+ p/f. This past week was brutal though as it was slow af, hit 55%.
But yea, I'm just saying, there's no reason why you can't have those variables before starting the month, if you could gauge at all which to set, even a little I think it would help. If I fail I fail though, at least I learn something.
It's been like building a jenga tower, was able to solve the stop problem pretty well though, taking a custom algorithm and measuring volatility of the previous month will give me a good number for the next month. At least the past 4 months lmao, not the best numbers to study with but eh.
Unironically my Dad's. It was 95% price action though so it'd be hard to get a bot to find this.
I'm collecting a ton of orderbook data myself for later backtesting with arbitrage and the way I've gone about that is to get the weighted average of the top 3 buy orders and the lowest 3 sell orders. It's weighted with volume. I find it helps with that, especially if someone places a tiny order that's at the top of the book.
However, you needn't bother with the orderbook at all mate. It's order HISTORY you're after, it's a piece of piss. Literally just go through the recent executed orders and count the volume of buys and volume of sells in the last X minutes.
oh for sure, thing is, I'm not looking for anything crazy right now. Using the one I have, I could backtest and get 100% profit in a week, but that is unrealistic.
It would be disingenuous to think that you could find the perfect numbers every month. Yet, if you could just find a few, every bit would help.
I'm looking at basically scalping at this point. The stops are mostly winners that stop out, preventing huge gains but securing small ones. However, if you're able to find a stop % target for the current or upcoming month, it does wonders when its correct, which I've been able to get pretty damn close. (of course not looking for perfection or unrealistic accuracy)
Yea, I'm gonna be on that today. If nothing else, it would be interesting to see the order book data overlayed on the main chart.
nice
>see how close it is to potentially being over
no it will randomly go several time farther than your estimator tells you. That's just how markets work.
ADX does a pretty decent job, just looking for confluence desu, something I could have missed etc.
