/biz/ - Business & Finance

>>51036064
You’re driving while looking at the rear view mirror. You will only get so much out of it and there’s never 100% certainty.

>>51036085
i have gone so far off the deepend with this i think the chart if a reflection of God and quantum physics.

I'm not sure if the past is even relevant to the future anymore; at all, and not just on the chart. You can put anything on the chart; and if TA is a lie, then momentum is a lie, and physics itself breaks down.

An object at rest may accelerate at any time, an object in motion may stop at any time.

>>51036085
>You’re driving while looking at the rear view mirror.
this is a perfect analogy; and any solution to playing TA would also seemingly apply to vehicle navigation where your only camera is pointing backwards and you have no map, and you don't know where the intersections are.

You can measure where the intersections might be; on a freeway, every 2 miles, from looking backwards, but it would be nearly impossible to drive a car this way and stay on the road unless the road was always a straight line and you never had to turn.

The answers that I come up with no in my mind are equivalent to flooding the entire highway with ants, some will flow off the wrong exits, but the bulk should stay in the pipe.

>>51036064
do you know how to code or are you just using TV strategies reccomended by pajeets on youtube.

you just need a half decent entries and trailing stoploss/take profit for a profitable bot. with a well tweaked trailing tp/sl you can even make profit on a simple ma cross. protip: leverages and tp/sl is more important than win rate / entries

>>51036085
everybody is driving on the same highway. at least TA have a rear mirror and are trying to drive. hodlers are just curled up in the boot

>>51036064
You need to assume that the biggest market movers are going to follow your TA algorithm to begin with, because retail doesn't move sheeit. And that's the problem with TA, you need to convince a bunch of midwits to believe in some gay pattern or lines in order for it to work (exactly why it's important for twitter influencers to convince their retarded followers how they'll make it, but only if they follow the underwater jew, interest rising wedge on the 5th sabbath timeframe), and even one whale dumping a fuckton of btc on spot can make your meme lines useless. Assuming they use TA at all, they've most likely agreed to use the gayest most obscure meme lines to begin with. That's why I trade macro instead of micro since announcements nearly always have the expected price action. Sell even after a 10-20% on some of the most bullish news and your portfolio just skyrockets.

>(1) good news, unexpected = strong poompa (euphoria)
>(2) bad news, expected = poompa (relief, priced in)
>(3) good news, expected = week doompa (fomo, buy the rumor sell the news)
>(4) bad news, unexpected = doompa (panic)

Don't let the midwits fools you into TA like that ben cowen cuck because it's all psychological, not mathematical.

There is only one TA which still continues to work on a longer time frame, it's EMA but I won't divulge in to the configuration of it and timeframes because it's a closely kept trading secret. It will confirm whether we are in a bear or bull very soon with good certainty and will also indicate where to set a SL/TP.

>>51036225
I wrote a completely custom bot with it's own built in backtester, that not only can play signals but grid/market make (pure market making is a separate scam, grid seems legit).

i haven't tested hard take profit / stop losses outside of shittokens, and that didnt' play well from the sheer amount of rugs.

playing purely on indicators fails, and I have metrics to shit filter the 1,000,000 combinations of macd or whatever indicator i'm backtesting. i don't just backtest macd with 12,26,9; i backtest ALL combinations from 3-100.
1,1,1
1,1,2
1,1,3
etc

filter then for profit factor and winrate and profit and all time low/high on the indicator and pick the best one.
it fails miserably on everytime frame (default settings also fail miserably).

you can only go so far with the zoom out cope, the 1hour or the 4hour doesn't test well either and it's equivalent to an "always long" strategy at that point anyway. and there's not enough 4hour/daily candles on most of these to test zoomed out that far.

the only thing that I've ever had profitable is a pure grid, with no TA whatsoever. grids are equivalent to tp/sl i guess. but it has the problem of
>what happens when the coin 45 degree moons or dumps for 6 days straight
grid fails hard, but if you can't predict the reversals reliably there's no point in even comparing. it seems like you just have to account for an immediate dump at all times; which boils down to an ultrawide grid.

Op the fact you're back testing is a good step you're taking. Now I hope you're testing at least 3 different types of stratergies.

1 for bull markets
1 for bear markets
1 for oscillating markets where this not much action happening

Then for each of these 3, you need different ones for different time lines, a daily one, weekly, monthly, yearly etc

Don't back test a single strategy over a decade of data. In that decade you should be using multiple strategies depending on the market

>>51036267
>There is only one TA which still continues to work on a longer time frame, it's EMA
that's why I've gone hard on MACD, and it backtests well but never plays out in reality; maybe because re-running backtests and flipping settings too quickly; but it shouldn't really matter if the backtester is rolling over 2000 candles everytime on longer timeframes (also failed)

The only thing that backtests well is equivalent to holding the trade open for 500 bars, at which point; of course it backtests well, it only flipped once or twice during the entire 2000 candle backtest period and the test isn't really reliable.

With backtesting there's a massive problem of biasing towards just holding something or not holding something; because that's seemingly the most profitable.

The backtester will bias towards a 100,100,100 macd settings, because just holding the bag and not trading at all was the most profitable outcome; so I have to artificially punish it or straight filter values out, at which point it just biases towards the max again.

MACD is just MA cross, and I can use EMA with the MACD calculation; but it backtests worse than SMA (just adding more lag).

PSAR seems to be the most reasonable thing at the moment because it has no actual bearing on reality; it's just a timer. still backtests and performs like shit and you can't run psar on anything higher than the 1/5M or it's just equivalent to setting a -90% stop loss on every flip

Well, use very long timeframes weekly and monthly to be more accurate on buy ins. Using moving averages on large time frames also seem to be more accurate

But be aware that the weekly chart = months for the move to happen. And monthly chart takes years to happen. But if you could catch/dca under the weekly/monthly average, then you will have months and years of gains

>>51036329
>shittokens
there's your problem. shittokens are too random. they're just gambling. put $100 on one you either lose it or make 10k. TA only works on the big coins,

>>51036330
>Now I hope you're testing at least 3 different types of stratergies.
the idea of brute force backtesting indicator combinations (and re-running it every 5 minutes or whatever); should have had the effect of a "one size fits all" strategy.

I'm already experimenting with making a long/short/flat call myself initially and biasing the bot with that.. with the pure grid - no TA grid bot testing.
the only call you make with that is starting bag size; either start full, start empty; or 50/50. obviously if it's gonna dump start empty.

i have also experimented heavily with combinations of indicator timeframes (ie 1m MACD must be long *and* 5/15m MACD must be long); but that seemingly didn't work well either; if the 1m is making unprofitable calls to begin with then why use it? (and dropping it improved performance), but they were all unprofitable in the end; so your stacking shit with shit.

>>51036475
i'm playing and testing on CEX I only briefly experimented on shittokens with a PCS bot. so many rugs it wasn't worth it.

i've tried shitfiltering pairs for likely success too.
>if the bot performs like shit on the pair, why not brute force the pairs and turn the bot on the best backtests?

TA only works on big coins seems to be accurate; or specifically coins with large market caps / decent volume, but even then, on major coins, I have never come up with anything using TA that performs reliably; all of the calls are bad and it eats losses from not only the coin loosing value; but from making terrible calls.

I have also tried combinations of indicators; and a literal "best out of 20" thing where I would run 10 indicators on 5 timeframes and just go with longs > shorts = long, etc. (that just adds lag, it ultimitely fails too).

there seems to be a common theme here

>>51036439
>maybe because re-running backtests and flipping settings too
you're over fitting. stop fucking with the settings too much. no wonder it won't work in forward testing, because what you did is fine tune it so much to fit the chart that's in your test range.

https://www.ibm.com/cloud/learn/overfitting

>>51036442
i have not testing very long timeframes, daily; weekly or monthy, from lack of data and just questing it in general; testing those timeframes would take forever to actually play out and the losses are much greater (the equivalent stop loss would be massive on every trade).

how much of that is just cope i wonder? If it takes 3 years to play out, it's going to work great until it doesn't. on the lower timeframes it seems you can atleast find out quickly if it isn't going to work.

TA has failed so many times this is boiling down to a spiritual philosophical question.
>Does the past matter?
The chart is a reflection of human consciousness; and on longer timeframes you would think more data = good and you can actual measure human thought, which itself is a reflection of the universe or God if your religious.

But the question is; is there random?
The entire TA theory assumes fate is pre-determined and there is no random. If there is no predetermination though; then the past has no or little bearing on the future, and TA will never work

>>51036526
the over fitting concept was deep in this. I was aware of the overfitting problem, and the thought process is "let's test out MAXIMUM overfitting". Let's tune the indicator EXACTLY to the pair, and then to solve the overfitting problem, re-run the backtest/scan every 5s/15s/1m/5m/whatever. clearly it doesn't work.

the alternative is to just go with baseline default indicator settings; but that doesn't backtest well or playout in reality either.

TA is astrology for men.

>>51036064
Indicators built on average values are useless.

>>51036064
Where did you learn TA from? You may have the wrong ideas on how it works.

>>51036650
aroon and psar didn't test well either. and psar makes the most sense, it's literally just a timer; not now instead of predicting trends your trying to predict market periods and timing.

>>51036668
i read on it constantly but it's been a combination of self taught and being brutally destroyed testing bots using it. Initially it was a mess but being wrecked a thousand times is a powerful teacher on what exactly is going on. I understand the math behind the indicators at this point I think I have a decent understanding matched with significant real world performance testing.

I mean really there's only so many things you can go with.
>Moving averages (ma cross) variants
>ATH/ATL (aroon / support / resistance levels)
>Volatility (bbands, raw volatility)
>Timer (psar, not sure if there's others)
there aren't really 50 indicators they're all just minor variations in 5 different themes; i've pretty much dug in to every single one that talib provides (and finta); and some of the wonkier ones on traderview; but those ones are usually just tweaks on other indicators too

>>51036735
at a certain point with backtesting it doesn't even matter what the indicators are. all that matters is the calls, and their performance metrics.

a mouse could press a long/short button and generate the calls and the backtester would run it.

I would have to agree

My "random lol" trades where I just buy or sell at a whim without looking at anything are actually more successful for me than when I do my due diligence and use TA

>>51036735
Sounds like you are using too granular a time scale. Trading channels form on the 1-day/1-week/1-month chart.

Weekly/Monthly MACD works perfectly well in picking out tops and bottoms.

>>51036771
weekly / monthly timescales are not something i've tested (not really on option on a lot of coins). The current testing around alwayslong grids... doesn't feel right without a stop loss anyway, weekly MACD is as good as any.

I'm going to start testing in that area obviously; there's nothing left.

>>51036800
In crypto, there is too much noise in the intraday charts. Whales have too big an impact on price in the short term. The winning move in crypto is to buy when the weekly/monthly are oversold, hold for many months, then sell when weekly/monthly are overbought.

>>51036064
its taken me about 2.5 years of all day every daying it neet monk mode and i can do it now. theres still dud trades but you generate enough profit to cover stop loss losses. it was a hard won fight, if it was easy to do it nobody would be able to make money because everyone would. stay mad. use your anger, let it drive you.

>>51036583
That’s why at the end of the day crypto is a farce in a sense because there’s no underlying fundamentals like discounted cash flow. With stocks at least you could determine the price to current cash flow and put a multiplier on the company. Most residential rentals are 5-10 p/e so if you can find companies in the same range or quality then you could have passive pillars of cashflow.

I think technicals work on established markets like stocks because at the end of the day if you strip out growth (crypto) there’s still cash flow that the company is generating. But I only buy and hold, and then use the dividends to buy companies under their 200 day ma with great balance sheets

>>51036860
playing the lower timeframes is a greed aspect too.
i don't have experience outside of crypto; i have no idea how much impact a "whale" really has on a chart, or the relative market caps where that becomes a problem. I can feel a difference between a 1 million market cap coin a 1 billion market cap coin; but compared to stocks or forex i have no internal feeling i can generate on that.

crypto i think has skewed my impressions and biased it towards lower timeframes for some reason.

>>51036064
>The pursuit of the perfect TA strategy feels like a drug addiction now that I must quit.
If the perfect TA strategy actually existed then it would be an infinite free money printer and everyone would know about it by now. Why does everyone think they can watch a few youtube videos and start gaming the entire system

learn pinescript you need to be able to customize your setup and understand the usually surprisingly simple math of whats going on. its good that you're testing a bunch of stuff. top down view, consider the intersection of multiple time frames, counterplays, volume conf, etc. i did not read your thread but from what i skimmed it sounds like you're trying to use a single indicator on a single slice of time like a retard

spend a lot of time on wikipedia just doing a general survey of math concepts, casino concepts, etc

>>51036894
the greed aspect i think can be dealt with with grids with the goal of eating all the volatility. Though how that performs relative to a hold bag is always debatable. but then again; if TA actually does hold true on higher timeframes, there's indicators to help shitfilter that (ADX)

>>51036909
>like you're trying to use a single indicator on a single slice of time like a retard
i've done the opposite really; I've tested every combination (within reason, about 1 million indicator combinations per indicator), on every timeframe under the 4hour, with variable length backtesting distances (200 candles back, 1500 candles back, etc) (longer usually plays out better obviously).

i'm probably too risk averse with the backtester shitfiltering though; it's not reasonable to filter out 20% dips before an indicator flip on a daily chart

>>51036952
what kind of conditions are you looking for give me example

>>51036064
>Is TA a Lie?
It is predicated on a belief that the future state of the chart depends on the past state of the chart to the (variable) exclusion of all other factors. Which is ludicrous. Whirling lumps of rock in space deciding your destiny ludicrous. Crystals channeling your Chakra ludicrous. I am continually amazed that people continue to give credence to this nonsense. Check my dubs.

>>51036960
I take for example 1500 1 hour candles, and then backtest the MACD calls with settings
[3,3,3]
[3,3,4]
[3,3,5]
....1 million combinations later
[100,100,100]

and test
>trades
simple trades > 3 shitfilter
>average bars
simple ab > 5 shitfilter
>all time low average (in indicator call)
this is a big one, filtered out -10% in indicator call for a while; changing that to -30% for daily now. also use it for total grid size
>all time high average (in indicator call)
no filter, just collect metric for total grid size
>max drawdown
same shitfilter as ATL
>total profit
filter under 0%
>profit factor
filter under 1.75
>winrate
filter under 40%

these are all tweakables obviously.
after filtering i sort those 1 million results (what's left) by either profit factor or total profit (PF isn't reliable if there's only a couple trades) over the entire backtest

>>51037033
checked, if true endless time wasted
>>51037053
i take the #1 result and play that obviously, and then i rerun the entire thing every 10 minutes (doing it more often causes flips due to changing indicator settings way too frequently and kills the bot)

>>51037053
this is what i mean, first of all doing a "bot" style trade like this is much harder than reading it "organically" using multiple indicators at once. you're testing a single indicator with a bunch of settings, and thats good because you need to find the good tunings that are resonant, but your criteria needs to be much more complex. you need to be considering ema's not only at different lengths, but at different time frames, you need to be considering volume, you need to be doing ta across history, looking for divs, higher lowers, lower highs, etc, all at once, multi indicator package that correctly weighs all this shit in good proportion

the biggest thing is the need to explore and see whats out there, and then when you find something good take it apart and learn what its doing

you say in this thread its all variations of the same thing and thats exactly right

>>51037094
i've tested indicator combinations, but they have always tested worse, which is why I gave up on that and went straight 1 indicator 1 timeframe. (and adjust the backtest distance, 300 candles, 1500 candles, 3000 candles, etc)

given, this was all lower timeframes, how this plays out on the daily is totally unknown.
the idea of doing
macd + psar
aroon + cci
macd 1min + macd 5min + cci 15min + aroon 1hour
etc

seems to always just add lag and results in worse performing backtests and in reality.

there's an exponential number of those combinations though, and I haven't tested them all.
a rough estimate
>5 indicators
>5 timeframes
would result in
15,625 seperate indicator+indicator+timeframe combinations.

At which point if I brute force backtested every indicator (which generally takes 10s), would require
15625s ( 4.3 hours)

The last time I tried to do a multiindicator / timeframe test I just used default settings on the indicators, because of how computationally expensive this can become (and it didn't perform well), but given that was a long time ago and I have better metrics now.

>>51037214

>>51037214
higher / different timeframes also adds the problem of lacking candle data.

exchanges will give you the last 2000 candles, those can be 1min, 1hour, whatever, but your only getting the last 2000.

which means if you want to match actual backtest *time*; across timeframes, not candle length;

to match 1500 daily candles you would need the last 2.1 million 1 minute candles, which your not going to get. you have to cut candles yourself; and that doesn't give you the past data, you can only start where you start.

so far i've avoided this problem by simply not doing it.
I generally backtest max candle length the exchange provides everytime. Everytime I try to drop candle length to match (drop 5min candles from 1500 to 300 to match 1 min 1500 candles); 5min performance degrades. more candles = more accurate backtest.

ultimitely the end game for this is you have to track your own candles

TA negros with confirmation biases don't want to admit this, but TA is complete bullshit due to the litany of untrackable qualitative factors that heavily affect the price. You're better off holding, researching gems, or trying to trade along general market sentiments or at specific points of high volume (like stock market opens).

>>51037254
i'm testing daily / weekly as we speak now. I atleast have the evidence to say sub 1hour TA is a lie. higher timeframes will be a new experience

already out of 1 million combinations the bot can't find a good backtest with a 30% in-trade filter; have to change that to eating 50% of the dip...

but at that point; what your doing here is setting a 50% stop loss, are you really swing trading? guess it doesn't matter, it doesn't work on lower timeframes at all. a "i should hold this" or "i should not hold this" call is sufficient.

>>51036064
with only a few clicks of my finger i could end your agony, your doubt, your despair. But i wont. O how delicious is your suffering, o how sweet is my cruelty. Is this how jews feel?

> always wrong always late
> believing in free indicators

i leave you a breadcrump thou just in case you actually put all that work in testing and trying:

this is an industry constructed to milk you for every cent you have. Brokers sell your order flow to hedge funds so the hedge funds can counter trade you because retail traders are 90% wrong of the time. the only real 90% accurate indicator. and its not for free. this is btw also the reason why a octopus randomly touching differently colored fields outtrades retail.
In addition the exchanges directly trade against you. when you long they are automatically short as their function as market makers. they can settle this short now or keep it open if they think the market will go in their favor. Hence there is an incentive to ensure you trade the wrong direction.
Thirdly, they borrow money from investmant banks for low interest, so they can borrow it to you (where do you think the leverage comes from?) charging you ridicolous fees and interest. raping you on a daily basis. or why do you think you get 100x leverage by just giving an email address? Guess who has the risk here in case of default? the investment bank.
The influencers get paid if you use their affiliate link to open an account at an exchange or broker. they get paid every single time you lose. the broker shares his profits with the influencer. the profit is your lost money. Hence there is an incentive to make you lose and trade a lot of volume....

>>51037330
nevermind that was with default settings
this is the type of shit I see that never plays out in reality; this is what the bot found on ETH3L (ETF on kucoin)

last 504 daily candles:
Total Profit: 876%
Profit Factor: 2.48
Winrate: 66%
Drawdown: 0%
Average Bars: 67.2
Average Low: 34.9%
Average High: 131%

MACD settings: 23,31,41

those actually seem far more reasonable than some of the bullshit that comes out of the lower timeframes, shit like 93,95,98 being the most profitable

>>51037487

and cause i am feeling particularily good today, some more crumps:

TA works a) only to define entries and exits. its a tool to improve your profitability and risk management.

b) works on its own only as part of a very specific set of rules. an these rules you will. not. get. for. free. its literaly the edge that feeds succesful traders' families.

no professional prop trader = hedge fund managers - uses TA beyond a) or very rarely b). that should make you think.

exchanges offer you for free indicators and TA tools. if they would give you any edge they would not do that. in fact exchanges and brokers run an inversion scheme where they tell you via influencers always, ALWAYS the opposite of what would be the profitable thing to do.


t. a macro trader

>>51036064
TA is price action, not indicators lol, of course they all suck.

>>51037330
oh no no this nigger cattle is waking up

> shut him down

ever wondered why exchanges and brokers playing your human flaws and want you to trade and as small TFs as possible? hm, you nigger cattle, ever wondered?

>>51037214
i would like to congratulate you on the attempt of beating algos created by MIT graduates getting paid 400k a year working for quants. mucho respecto, anon. Participation medal ftw.

>>51037507
it's free at c0tt0nc4ndyta on youtube, has 10 courses there, creme of the crop.

>>51037757
inb4 retard doesn't check the courses and gets stuck on cotton's attempt at gamifying crypto and attracting audience with the only way they understand, greed.

>>51037740
>i would like to congratulate you on the attempt of beating algos created by MIT graduates getting paid 400k a year working for quants.
at this point, i have a hard time believing they know anything that I don't.

maybe they're throwing AI at it, maybe they're throwing more CPUs at it. i'm not saying i'm smarter than MIT but I can't really conceive of any magic strategy they can even generate.

how many ways can you fuck with ohlc data?
with unlimited cpus and money, maybe backtesting with raw trade data over an infinitely small timeframe.

it boils down to predicting when two elites get together and decide when to fuck with the price; not because of any organic movement, but even that on larger timeframes can be seen as an organic thing, when it plays out 50,000 times over the last 500 years.

if you're doing a million setting combinations on an indicator you're doing it wrong.
you're never going to have a 100% hit rate with entries / exits.. the trick is to get a somewhat decent entry rate, then use trailing tp/sl to make it profitable..

>>51038049
Why do you fucking idiots repeat stuff like this? Obviously no one is expecting to win 100% of the time.

>>51037967
yes my biggest question is what will happen when they start training AIs on it. i dont see how they wont kill it

"failed" as in what exactly? To make 1% on a trade or to not lose max 5%?

>>51038194
i don't know much about AI in general but from what I've glanced at it's a meme.

it's just database operations. they aren't forming fake brains with nuerons or creating new life or anything; there isn't anything "intelligent" about it, they're just brute forcing raw data instead of TA indicators.

>>51038284
>fake brains with nuerons
they're using genetic algorithms. i think if you set it to correlate good entries with the correct filtered indicator output they will be able to win almost always given the right risk profile. and it wont be out of the publics hands to get access to it and start sharing it

>>51038250
i track profits based on all 3
>profit vs 100% hodl bag
>profit vs 50% bag (useful grid metric maybe, don't use it much anymore)
>profit vs 0% bag, raw profit.

it it's failing it's always going to fail on all 3 over time, it's playing longs and shorts (not being in a long is a short when tracking against 50% or 100%)

i asked the gpt philosopher ai, how would you test if the sky was blue and it described hooking a sensor up to itself and reading the wavelength of the sky. you're right, it doesn't know what its saying, its all numeric associations and weights, but i'm not sure that will matter either

>>51036800
I consistently get wrecked if I try to trade on the daily or weekly charts following a death cross on the monthly. I won't even touch a crypto until I see the golden cross on the 1M.

>>51036329
>retard codes a trading bot
>tests it on super volatile shitcoins instead of BTC

you are the most retarded person i have ever seen on /biz/

https://www.youtube.com/watch?v=hLV09_FDnIQ

>>51036064
Tradersumo.com
Polytrends.io

>>51038570
but +- 50% swings = profit ! :(

as a result of this thread i've made the following changes:
>only test on major coins with big market cap
>don't test on ETFs
>only test 4hr+ timeframes
the reasoning for lower timeframes and shitcoins is greed clearly. running a grid instead of straight hold on signal will satisfy the greed portion

any future shittoken gambling will be done with pure grid and no TA

>>51038349
ok but when do you sell? do you hold on to a failed trade? do you always get filled at expected price ?

>>51038681
OP, if you understand the concept of trend, you win.

https://youtu.be/ot6Bv33MSQY

https://youtu.be/9BL-71JBcxI

>>51036329
You are going to go crazy looking at the numbers and the patterns in the numbers. Just ignore them. Don't go down this way. It's a gargatuan, ever-changing adaptable system, you're not going to figure out the patterns because they are just not in your scale. Your model couldn't possibly hold all the variables needed to simulate this.

>>51036064
lol I just went all in ohm. gl with your little math equations.

>>51038718
you buy and sell on indicator flip
>when is the best time to buy?
when the indicator first flips
>when is the best time to sell?
whenever the best time to buy a short is, on the indicator flip.
any kind of hardcoded TP/SL has no basis in reality, or atleast that's the thought.

i measure the ATH and ATL of each trade in the backtester and use that for setting grid percentages if I'm gridding it; but I don't preemptively sell when it exceeds the backtested ATH average or anything.

if your under the assumption that the indicator is accurate then there's no reason to sell before the indicator flips

>>51038879
ok, but try to adapt that then. maybe have a sliding stop loss at all times?

>>51039203
that's part of the reason for the grid; not only are you playing volatility; but (in a long) as the price goes up, your selling inventory and your risk is going down.

if you start at 100% inventory; and the the coin goes up 100% (with a 1% grid spread); at the top you are now at 0% inventory.
with MACD that is essentially the stop loss; the indicator will flip back before it goes back to it's starting position again.

if the market dumps 50% again, then you didn't just ride your entire bag to -50%; the bought bought back in, and the stop lossed on an indicator flip (dragging 50% of the bag down tops, instead of a 100% bag).

ultimitely one of the problems with any static TP/SL or even combinations of indicators; is that it creates a "neutral" concept for the bot; where it's neither long or short.

there should be one objectively correct answer at one one time; long, or short, there is no neutral.

But based krown would never lie to me, right?

No, it's massively misunderstood by people looking for a crystal ball.

>>51039451
it's more than that.
i think i can prove that TA calls are flat out inaccurate on every timeframe < 4hour; at which point they are worse than useless (they are worse than 50/50 gambles) and should never be used

You're a fucking idiot. You won't trade profitably until you follow macroeconomics.

TA is used to discover entry / exit positions. THAT'S IT.

You need to follow MACROECONOMICS to realize where the flow of money is going.

So fucking stupid lol

>>51039487
>You need to follow MACROECONOMICS to realize where the flow of money is going.
crypto is a ponzi scam though under the hood how am i supposed to do that?

>>51039521
>crypto is a ponzi scam
Then why are you trying to do TA on this ponzi shit, are you fucking stupid? just get out and save your money, buy stock or whatever.

>>51039544
because it's not supposed to matter
>All information about a thing is contained in the chart
there is nothing but ohlc data to look at.
i do my own pair shitfiltering obviously i'm not going to run on yield token dump coins or the ever growing list of shit i refuse to provide exit liquidity for (anything out of india, "gamefi" etc)

>>51039521
crypto is a ponzi scam that subsequently outperforms during low rates + QE.

Now that we're facing the opposite of those conditions, crypto is going to the dirt. I've been short since $40k/BTC because I know this. You haven't because you are too stupid to follow macroeconomics.

Trying to trade without following macro is like trying to jerk off without hands

>>51039306
>there should be one objectively correct answer at one one time; long, or short, there is no neutral.
There's your problem. With a strategy like that, you can only be profitable in huge timeframes like 1w. If you want to use lower timeframes then you absolutely need stop losses and take profits.

>>51039521
Weight RSI by volume, Look for divergence in peaks/valleys vs. price. If those divergences coincide with extensions off moving averages (exponential or simple depends on time frame) Then you have confirmation of a high probability move towards the closest untested supply/demand zone.

>>51039649
Listen to this man. Crypto is 100% dependent on cheap money. Venture capital doesn't fund shit coin projects without cheap money.

>>51036064
use quantum mechanics, thats how i make money every time

>>51040212
I am currently learning how to do this.

I'm reading a book called 'Proven Techniques to Remote View The Stock Market'

>>51040257
>>51040212
i've unironically considered generating calls off the GCP dot

>>51036064
Trading is PvP. Much better off going by general sentiment and emotion, it's not hard to tell when you get close to a top or bottom.

>>51036064
TA is trying to predict the future by looking at past price action
It is essentially an attempt at mass paychology

You might want to learn value inveating instead

>>51037036
I'd just smoke a joint take some acid and watch this shit

>>51036064
If you're into trading lookup Smart Money Concepts, best are ICT and RTM. Lookup them i youtube. ICT taught for free recently. Lookup ict 2022 mentorship.

If you're into investing, lookup value investing. Buy cheap, sell expensive. As said by benji graham.
Lookup economic cycles, they're very very important as said by billionaire investor ray dalio.

Do not combine TA into investing and fundamentals into trading.

Trading is all about playing with statistics.
If You can win 50 times and lose 50 times, Then you'll make it.

Don't ever day trade, not worth it.

>>51039306
what do you mean? the market change, so you need to change

you need to add something that change position size in % based on the move for example, to get the outsized returns but not risk too much

>>51040770
>shitty forex factory users made trading courses
>here bro, go waste 1 year watching all these shitty videos

dumb newfag

>>51040909
I assumed op had no knowledge or capability to do factor trading.

Without a solid plan, using TA is equivalent to gambling. Tail risk is too high.

Trading regular stocks is like donating money with each stop loss. Not fun.
Rather, you could lose money and have fun while shorting pharma stocks or buying OTM Options.

>>51041041
You are lost cause faggot. You just posted a meaningless paragraph in response to me exposing the faggots from Forex Factory making made-up methods of trading and retards spreading them around sending money their way by selling their shitty courses.
You are the dumbest mother fucker and an oxygen thief.

>>51039666
Devil Mark Checked !!
But, this unironically.

>>51036064
you completely misunderstand what its about.

trading using bots/TA/fundamentals/short selling is about using your time more efficiently.

HODLing has way too much cost-opportunity loss. you literally have to wait 4 years to at least to see any results for like 1/3 of all shitcoins.
2/3 never recover.

a good trader has more predictability in his outcome.
I mean, why the fuck should I hodl through a bear market when I can short and take profits if markets signals downturn??

also, TA, is obviously not 100% and even using a 50/50 trading strategy with money management, well, you are going to win in the end.

macd is a typical trend indicator (40%-50%) chance of being right.
the times that it is right, it can win more than money than it loses money,
a macd trade has unlimited upwards potential and a limited downwards loss potential since the trade will be closed the next day since the ma crossover will probably happen the day after if price goes down after you got in.

hodling is literally retard cope.
every time I see a thread that shits on trading, it shows 99% of the times that OP has completely misunderstood everything and is throwing out the baby with the bathwater.

about backtesting:
you are not supposed to optimize 1 million times and choose the value that gives you the best backtest.
you are supposed to pick a value that is in a good "zone".
if indicator value 9 gives huge profit but 10 and 8 give you losses, then your strategy is worth rat shit.
if indicator value 9 gives acceptable profit and 10 and 8 also give acceptable profit in the same ballpark, then we are are talking!

stop losses? some guys use only the exit rules of their strategy as stop loss.
some add an additional stop loss,
some do not use stop losses at all,
some use margin calls on their highly leveraged positions as stop losses while keeping 90% of capital off-exchange. thus making a 10% capital stop loss.

>>51036475
ignorant and false statement

>>51041072
> op asks how to make TA profitable
> me say, look this is one way
> literally free to learn.
> free to backtest
> mathematically profitable

Just because you don't like it or can't understand it doesn't mean it's bad.
Also, with that logic, everything is scam, every finance concepts are scam, in long run passive wins over any kind of brilliant active strategies.

What's your brilliant suggestion to use TA or in money making in general? What's your take in it? I'm listening, genius, say it.

your strategy should work on 70% of coins, if it works on btc and not on eth then its bullshit.

>>51037053

again.
you are supposed to optimize to get a view of the following:
are most among 100 indicator values profitable? OR
are only a few amont 100 indicator values profitable and the rest are losing.

this is the only question optimisation should answer.

if most are profitable, then your indicator is likely to adapt to changing market conditions.

if only a few values are profitable, you have an overfit pile of shit and you should not trade it

the people making their 60% winrate are literally just long on growth problem is they get a string of losses and either get wiped out or wussy out
the key is just to not give a shit

>>51041263
your definitely right there, i ended up putting much tighter rails on the MACD combinations, and i can easily factor instead of [0-100]
default: 12-26-9
test 0-25, 0-50, 0-25
presumably a large share of those MACD backtests should come back positive and i can cut off on that.

i do have an indicator stop loss essentially. if the backtest comes back and they ALL fail; out of 1 million combinations nothing passes the shitfilter; then it nopes out for 5 or however many minutes.

>>51036064

look at it this way:

virgin overfitted indicator:
optimizing values: 2 to 20 with step of 2
results:
2: -77% loss
4: -33% loss
6: 9000% profit
8: 69 % profit
10: 10% profit.
etc.

you are not supposed to chose 6 just because it has 9000% profit on the backtest. I would never trade above virgin indicator

CHAD good overall indicator:

2: 120% profit
4: 150% profit
6: 250% profit
8: 200% profit
10: 167% profit.

I would trade the chad indicator with random value between 4 and 10

>>51036064
>Is TA a Lie?
Yes
>>51036064
>I'm about to abandon the entire concept of TA as a waste of time
Smart move. Now go learn python and TensorFlow. You cannot predict market movements based on past data. It is useless. Where you can make predictions is based on market inefficient tangents such as geopolitics but being a poltard spastic won't win you any prizes

I need to emphasize again though:
if your shit works only on one coins, its not worth your time and money.

your strat should work on most coins that you throw at it.
also, getting 1 hour candles or even worse 15 m candles for more than 1 year is a pain in the ass.

just go for daily chart or the weekly chart. like that you get bull and bear markets.

>>51041533
im going to implement this for sure; it's not something i've been doing.

if the bot is going to flip on 1 million combinations of indicators there's a big difference between 10 out of 1 million being winning combo's and 500,000 out of 1 million being winning combo's; right now the bot doesn't care it's going with #1 and will only nope out if they all fail.

i've been backtesting some retarded settings
ie 997,1000,1000; retarded crap that isn't expected to do well, but if the bot flips to it... really it's to study behavior; and the bot will definitely flip to those retarded values because it's the only thing profitable (usually equating to a always long / always short preference maximum average bars, ie don't trade at all preference).

i've been blind to that and the percentage of actual winning indicator combinations compared to the entire available test range is important and would explain a lot of these failures.

>>51041533
You are an idiot who is no better than a gambler, I would go further and say you are actually not mentally competent to handle your own pocket money and need a guardian. If you could *with any odds* produce 120-200% profits you would own the entire planet in a matter of weeks. You don't. Please fuck off. You don't know the first thing about portfolio management or the mathematics involved because you are a drooling halfwit who buys animal coins and longs GME

>>51036064
The TA is most likely a background tool; you need more Richard Wyckoff accumulation and distribution methods. He takes the ideas from a TA although continued to write and develop today's modern stock trading strategies still used by today's market shareholders. First, learn to spell Wyckoff.

>>51041699
if I am an idiot then you are a piece of toilet paper stuck in a fat cunts ass

>>51041680
the lower the timeframe the less you should focus on values too close to each other.
on the weekly chart its literally enough to test 10 values most of the time from 1 to 10 because the candles contain so much information already.
anything less than 4h is mostly noise anyway for a trend trader.

if you are a mean reversion trader then lower time frames are much better but usually also much harder to program.

>>51036064
lagging indicators never work rip bozo.

use fib channels like a man and not meme indicators like a fat nigger.

>>51036225
hodlers are curled up in the boot of the car that will make it to the destination.

>>51042157
well the traders can enjoy eating at some nice restaurant during the trip, they can listen to music and even have women in the backseat that they have sex with in hotels instead of sleeping in the boot of the car, waiting to arrive at destination...

>>51041632
>You cannot predict market movements based on past data
fucking negative 500 IQ
nobody is trying to predict shit retard, the bots only purpose is to watch the fucking chart 24/7 and get in on a move that's already in progress, so the botter could sell out when there's a clear dump and buy back in when it's pumping again while you bag hold from 69k to 20k coping and seething for another bull run. stop fucking pretending to be smart when you're clearly a retarded boomer

>>51036064
most of is useless and drawin meme lines etc. quite useless in general. much like most of 'fundamental' analysis is totally wrong and useless even done by experienced
professionals they are horrible what they do

but there are stuff and indicators not known for wide public that are good and can offer clear edge

people are so concerned about buying the absolute 61.8% bottom that they will literally miss out on time is money upside but I guess they want to be "perfect" LMAO

>>51036267
89d ema. WOOOOOW BRO

>>51042340
>pumping again while you bag hold from 69k to 20k coping
this is why even starting empty with a grid buying all the way down 50% i can't justify a full "always long" strategy.
even if TA sucks you can't tell me there's no way you can identify a straight 45 degree dump down which is what it often does.

>>51042544
>>51042340
it's especially lulzy on the ETF's
should a bot have DCA'd into this or held?

all the hodlers who shit on traders are like obese people who talk shit about athletes

>>51042500
89EMA, ha okay amateur.

off-topic but some of you are alright.
watch out for the dollar pumping. its positioned for a potential breakout upward

>>51036064
You might as well read tea leaves for all the good TA does. At least with tea leaves you can have a drink to help cope with your losses.

>>51036064
In 2013 I became OBSESSED with indicators and TA. O back tested literally close to a hundred different setups on historical tock data I purchase and I ran through a special data provider for extra features.

I tried all sorts of stuff. The most successful was a stochastic model I came up with that was hugely profitable but almost impossible to manage psychologically because of big swings.

I came to the conclusion that TA is bullshit that newbies and suckers fall for. Wallstret LOVES fresh meat and TA provides just enough to get idiots like me to waste money on their casino.

jesus h christ you faggots suck, i thought i was in a serious board but damn yall retarded
no wonder you're losing money
ta works, and is accurate to the cent, if your double digits iq isn't allowing you to grasp it, cope and seethe.
t. max leverage degen (100x for btc/eth, 50x for other large caps), 91% accuracy in the last month. 0 indicators 0 volume mumbo jumbo. pure price action, levels and trends, you don't need anything extra.
I'll say it again, price action, levels and trends. there are free courses online, check c0tt0nc4ndyta on youtube, don't mind his retarded videos, go to playlists and find the 10 courses.

>>51036064
>He hasn't tested the moon phases indicator

>>51043930
you have less than $100k

>>51044302
yes, because fiat is worthless ya dingus, how did you figure this out?

>>51044302
100k to make a million FY is huge
even with a 50% winrate you can spread your capital it out every quarter and make some insane ROI which means less capital and time risk than conventional investing

>>51036583
TA stopped working because retail is less than 10% of the volume. TA analyses herd psychology but there is no more herd. Also there are algorithms that set up TA entries only to move against them and dump on anyone who took the bait. The result is the current market. Stop using TA and stop trading.

>>51045864
>Also there are algorithms that set up TA entries only to move against them and dump on anyone who took the bait.
this is claimed a lot but if I have a hard time pulling 1% APY momentum trading I have a hard time believing there is some super advanced sophisticated algorithm that not only is profitable but is capable of destroying other bots.

i guess anything is possible when your tossing around so much you can move the entire chart

>>51045993
coincidentally since i started this thread and switched to testing the 4hour chart the test bots are in the green; for whatever that means on a 6 hour test .

>>51036598
All of those oscillators and moving average crossovers, bollinger bands etc. are useless without the underlying context of the PA behind them. And that's something which a human can grasp intuitively and through a wide frame of reference, by integrating data from many sources and scanning for confluence. But it also requires confluence in a manner that is not intentionally biased i.e. using confluence between highly correlated indicators to draw a conclusion about the market, is an exercise in futility. It is possible to build bots which can simulate this kind of "thinking", but eventually they will lose calibration. They aren't "smart" enough to adapt and integrate new information sources in a way that is market predictive.

Personally I avoid nearly all mathematical indicators (oscillators MAs etc) as they are perpetually in a lag, and suffer easily from overfitting problem. They are useful for scalp/mean reversion type strategies but I do not trade that way. I use old fashioned chart scribbling, fibs, candlesticks, volume profiles and volatility. Then take a big picture view, look across all markets, assess all the potential risks, come up with a probability distribution for different possible outcomes, and position accordingly. It is very difficult for a bot to do that consistently over time. If you could make a bot capable of trading that way, you would be working at the largest banks/hedge funds making tens of millions. That's how difficult it is, and even they lose money from time to time.

>>51046987
what's your trading style? as a scalper almost all of these are honestly irrelevant, map out the levels and trends, see where we're at in the big picture, decide on an entry (most often it's retest of a broken local trend, which would get me to the global trend, that's the move, if it's on BTC, it's usually 1-4%, on 100x that's 100-400% in a few minutes, two hours tops.

I follow the exact same setup for longer trades too, pick a range where it's never gonna be touched again till it hits your target > enter > enjoy free money.

>>51036064
i think youre just fuciking up

my friend is geeked the fuck out on this gay ass ta shit im always bustin ghis balls but everytime he calls me hes not lying that hes made another several gs from yesterday

>>51045864
I agree with this generally, TA will fade in relevance the more a market becomes market makers bouncing a ball between one another. You can tell when its a market makers market, when they will just walk price around in uncannily clean sine waves and lines. All the zippity zap of crypto is from retail and leveraged retards treating crypto like the casino it is, but when they run out of money (liquidation), only the house is left. That is often when you will see "market maker markets".

>>51047171
I trade vol squeezes similar to how you describe, taking tight risk on a small range for large move. But generally wait for setups on big timeframe, so I do not trade often. Trend continuation, looking for signs of strength/weakness in the PA structure. Seeing how the patterns are laying out and fractalizing, interacting with psychological/technical levels. And if late trend, watching for signs of exhaustion, anticipating the warnings of what could become trend reversal structure.

I used the basic MA 50/100/200 as one of my tools when I ran a low timeframe scalping strategy, always easy targets or potential entries. But that was back in 2020 so idk if the same context exists for it to work the way it did then. Was basically countertrading the liquidation bots and order flow, calling bullshit on spoof orders and watching them manipulate the book for my benefit. Fun all except the insane amount of intensity and time required to really make it work. I had to give it up, it honestly exhausted my ass.

>>51036064
I've experimented with probably 20 million + different combinations at this point using different indicators in combinations, SL/TP, trailing, and a couple of other factors.

Even with monte carlo simulations and walk forward optimziation I am yet to find a robust strategy that works in live markets.

I think the issue is that fully automated systems arn't able to understand the context of the market and they rely on lagging indicators to determine things like trend. You can program for some of the low hanging fruit quite easily (avoid trading during high impact news events, brokerage rollover, holidays, etc) but getting a bot to understand the market structure is quite difficult. Not to mention macro trends.

That being said, I'm sure there are some winning algos out there, but the odds are probably something like 1 in a Billion if not higher. I imagine there are probably some hedge funds/prop shops that have entire server farms to "mine" for strategies.

my TA works and i posted it in my trade log but jannies nuked the thread and banned me. they dont want you to make it. meanwhile they leave up scam threads,

good morning

>>51045334
im turning $100 into $500k in approx 150 trades including losers

>>51037757
I thought cottongoy was a scammer?

>>51047654
>fully automated systems arn't able to understand the context of the market
>rely on lagging indicators to determine things like trend
>getting a bot to understand the market structure is quite difficult
>Not to mention macro trends
bingo

>>51036905

Because it can be profitable. The only question is how much are you willing to go after? There are way too many pip hungry traders. You have to understand if your TA strat is good for 5,000 pips or just 50 pips.

>>51036064
cont.

If you have to rely on raw computing power or on speed its probably a losing game for the retail trader.

>yield farming
Too much platform risk imo. Even then, you may be getting 20% a year but you also getting diluted by 20% a year with new emissions.

>grid
There are plently of EAs on mt4 that will paint a nice rosey backtest but at the end of the day they all carry significant tail risk and usually blow up after some time. Now on the other hand you could pick an asset with a directional bias (S&P 500 for example) and go long only and use options to cover your downside risk. This is essentially how some firms like Jane Street trade. Basically DCA with extra steps.
You could also do some interesting stuff with synthetics like the VIX (volatility drops over time) or a rebasing coin like AMPL but that has its own issues.

>>51047767
am I getting pip hungry? I know that I'm hungry for pips.

>>51036064
I think the best opportunity is probably in selling option premium. Basically IV (implied volatility) will generally be higher than the actual market volatility because institutions, corporations, industry giants have to hedge their positions. So you can profit on the spread. When you sell options you basically become the casino for WSB degenerates.

>>51036329
I tend to think of these things more like a gestalt, rule of thumb, or nice to know data point. You should still asses the environment and do all that good value investing shit that seems to consistently work, then take a look at the technical and. you know, don't buy if line is up and try buying when line is down. In other words, increase your odds a little bit. If you do a bunch of other tings to increase your odds, then it adds up to a better decision, which you should absolutely still protect with risk management, because the future is not knowable.

>>51045993
momentum trading is for plebs

Also, I find the shape of a falling stock much easier to notice than rising ones. A falling stock is a big drop down, then a little rise up, then it's time to buy a put--not that I can afford to play options though.

>>51036064
>professional futures trader
ill spoon feed you retards once

technical analysis is any analysis of a price chart, some of the greatest traders of all time use it
indicators are derived from the price action on the chart, usually with lag
you need to realize that testing indicators in a vacuum is useless, these indicators are tools to add to the confluence of factors which are the reasons to take a trade

>>51040770
>ICT
proven scammer who faked all his profits

>>51047720
>im turning $100 into $500k in approx 150 trades including losers
no you dont you stupid scamming jeet

>>51049822
cope and sneed

>>51040770
>If You can win 50 times and lose 50 times, Then you'll make it.
That's exactly the point where you don't make any money. Either extreme, you could inverse your decision and guarantee an edge.

>>51049837
take your ganga bath, schizo

>>51050120
u reported my posts didnt u

>>51036064
bump

TA unfortunately is bullshit. I also went through a few years of trying every indicator- even weird russian ones, and backtesting all the settings. I have settled on direction neutral trading. not sure why there's not more discussion about thetagang on here in general. i wish crypto had options, because I don't particularly like stonks. I also think options on crypto would help reduce the insane volatility. i sell options about 180 days out and also buy put debit spreads to hedge my position. it is a solid strategy and makes about 20% per year, but it requires portfolio margin.

in short, everything is totally fucking random. good news comes out, something tanks anyway; etc. fuck picking direction.

last point. .50 delta means 50-50 chance. options are priced so efficiently. no one has any clue what's going to happen, it's all a gamble, that's why at-the-money is .50 delta even on solid companies. not interested in gambling, need a definite source of income. selling options (not the tastytrade way or 'the wheel' though).

just learn wyckoff and stop crunching numbers you stem retard

>>51036064
its all a probability game and margins are thin af, day trading is a meme you wont be making money because of fees

>>51036329
You are somewhat competent because you were able to program such a thing but if you think shitcoin or crypto has any kind of consistency regarding indicators you are deluded
Maybe the stock market is more predictable but crypto is just gambling and being so arrogant as to try to solve it with some indicators is hard cope.
Just use the high volatility to make small frequent profits by buying when it goes up and selling right after with the bot

>>51036064
I can confirm this
I did the same with every stock
It failed everytime

>>51045993
Hey Op, I trade professionally using TA/statistics something like what you're trying to do. Some advice for you: All your targets and stops need to be based on a volatility indicator and not arbitrary percentage. I use ATR. Ie: target is 4X atr away from entry and stop is 2X atr away from entry. This takes volatility into consideration and will work in all market conditions at all times. If you want to have hard limits like "never take a trade if it's riskier than x%" you can just program it that if your stop atr multiple would be higher than x% it doesn't take the trade. You mentioned that you tested indicators and they all didn't work as intended. That's totally fine, one of my strategies uses an indicator in the completely opposite way it's supposed to and it's very profitable. Different markets, timeframes and market caps makes a big difference. I don't trade lower than 30m in crypto because it's too random. In general: stocks revert to mean, forex trends, crypto trends. Momentum strategies will work much better in crypto and reversion to the mean strategies will work much better in stocks. A single indicator by itself is pretty useless, there's always more you need, a confluence of factors. I keep my strategies very simple but it still might be ATR + MACD + volume or something. Good luck fren

>is TA a lie?
The only people that make money off TA alone are the grifters selling courses on how to make money with TA.

TA is only part of it. TA will show you when is a good time to enter or exit your position. TA isn't some substitute for doing all the other things you should do as a trader: like actually doing research, coming up with trading ideas, managing risk, hedging and so on. Ignore all that stuff and use TA exclusively if you enjoy losing money.

>>51036064
i can find a profitable backtesting with literally any indicator you mentioned on 1d chart in a few minutes. not that it means anything for future results, but still, you're just bad

>>51055730
>i can find a profitable backtesting with literally any indicator you mentioned on 1d chart in a few minutes.
bot is running on SOL3L 4hour with MACD 9,49,11 right now.

>over 1500 candle backtest
Total Profit: 1702.77%
Winrate: 50%
PF: 3.85
Average Bars: 22
Drawdown: 26.35%

the 1D is going to be rough which is why i dropped to 4hour, there isn't even 1500 candles worth of 1D.. i think MACD might be too slow for the daily

>>51054729
this is why i'm trying to mix in grid here instead of straight signal. bot right now on the 4 hour is running a 60% wide grid, which so far has made 1% in grid profit over the past 12 hours and 6% in regular pump profit. however it's going to eat atleast a 30% loss on this timeframe before it get's out. i guess it's better than 100% loss with an always long strategy

>>51053573
> learn wyckoff
i don't even know what this is; looking it up now.

>>51048600
>you need to realize that testing indicators in a vacuum is useless, these indicators are tools to add to the confluence of factors which are the reasons to take a trade
>confluence of factors
like what?
>the whitepaper is filled with lies
>the devs are are scamming bulgarians or jeets or shady asians where "doxing" means nothing because they all look the same and have the same names
>the exchange the coin is on is liable to steal all of your funds
>VC "investors" of the token are liable to pump and dump at any time
>the token is a security propped up with nothing but good thoughts
i mean if you dig deep on any of the coins they are all bad but they still moon and make what should be clearly detectable 45 degree angles on the charts

>>51036064
>macd, aroon, mesa, simple roc, bbands, psar
Those are absolute midwit indicators. KISS and stick to the basics.

>>51036329
You can’t bruteforce the right parameters for your system, the parameters space is too large. It’s more likely that your hypothesis is wrong, I would focus more on the maths behind the indicators and they are trying to capture about the market

Also look up qualitative approaches for nonlinear dynamics, imo the Strogatz is a good book to know

>>51055892
>bot is running on SOL3L 4hour with MACD 9,49,11 right now.
I backtested those same settings on tradingview and got 90% drawdown and pretty much all of the profits were made in the last month.

bump

TA is absolutely as meme. What is not a meme is price momentum and trend following. These are serious signals, well documented in financial research and puts actual bread on your table.
The only thing you can do with TA is to sell courses on it to retards.

>>51040461

https://www.youtube.com/watch?v=Q1vjKGie3yg

>>51036064
>The only thing that sounds reasonable to me at this point is splitting the bag on 100 pairs
i've done this. it does not work. tried it for 10 months. had a wallstreet consultant involved, and an ai researcher. could never make it work irl or in backtesting

>>51036064
some people can do it, by pairing multiple technicals.
if it was easy, everyone would do it. it's not.

>>51037757
Retard c0tt0nc4ndyta is a known fraudster within the TA community, he is so far from a trader.